homemade EFIS system and EMI [Archive]

Peter

January 25th 05, 09:30 PM

> As a pet/learning project I made a data aquisition unit using an 8051
> microprocessor and an EPIA M motherboard running a variant of Linux to
> process and display the information. It all works but the EMI from the
> EPIA M causes way too much noise to the aircraft radios. I have tried
> shielding the whole device in an aluminum case with very little
> improvement.
>
> Any ideas on how to smother the EMI or some other small motherboard
> which may not have as much an issue (as a test I took my portable
> aviation radio and within 6' of any my home computers the same occurs
> which gives me little hope)

Physically separate the offender from the victim, especially the point of
entry of the signals into the victim. Don't forget about coax cables, comm
antenna and things like that, not just the radio box. You could try
reorienting the motherboard 90 degrees in various axis.

jcpearce

January 25th 05, 11:36 PM

This is without the system connected to any of the sensors, just
turning it on and with an external battery so it is physically isolated
but within 6' or so it causes interference. Rotating the motherboard
does not help either. Thanks for the reply.

Blueskies

January 25th 05, 11:57 PM

"jcpearce" > wrote in message oups.com...
> This is without the system connected to any of the sensors, just
> turning it on and with an external battery so it is physically isolated
> but within 6' or so it causes interference. Rotating the motherboard
> does not help either. Thanks for the reply.
>

I have the same problems with my desktop machine and handheld radio...Wrap with aluminum foil and see what happens...

UltraJohn

January 26th 05, 02:23 AM

jcpearce wrote:

> This is without the system connected to any of the sensors, just
> turning it on and with an external battery so it is physically
isolatedseparateithin 6' or so it causes interference. Rotating the
motherboard
> does not help either. Thanks for the reply.

Possibly try remotely mounting it back in the fusilage and running
fiberoptic to it. You can get inexpensive hobby FO and FO to rs232 modules
designed for experimenting.
Other than that try a non-aluminum case and use feed thru caps on all leads
and ferrite beads also. try some additional filtering on the PS and/or try
using separate PS for processor.
Just some thoughts.
John

ps those hobby aluminum cases do lousy for rfi prevention!

kumaros

January 26th 05, 05:14 AM

Bob

January 26th 05, 01:12 PM

The "noise" could be coming through the connector and could be emitting
from the wires.

Try using a connector with a metal housing and backshell and wires with
shields, terminate the shields to the connector backshell/housing which
will ground to the chassis.

The other end of the wires, try to terminate their shields where they
connect to.

Aircraft environment is not the same as home environment. Different
applications different techniques.

jcpearce wrote:
> As a pet/learning project I made a data aquisition unit using an 8051
> microprocessor and an EPIA M motherboard running a variant of Linux
to
> process and display the information. It all works but the EMI from
the
> EPIA M causes way too much noise to the aircraft radios. I have tried
> shielding the whole device in an aluminum case with very little
> improvement.
>
> Any ideas on how to smother the EMI or some other small motherboard
> which may not have as much an issue (as a test I took my portable
> aviation radio and within 6' of any my home computers the same occurs
> which gives me little hope)
>
> Thanks

jcpearce

January 26th 05, 03:44 PM

Thanks for the suggestions, no I have not posted what I did anywhere
but I think I will as it took me a little while to learn rudimentary
assembly, come up with a variant of linux running off flash memory,
write the code to process the serial output and display etc..

I did not use a hobby aluminum case, I custom built one with no gaps
(motherboard has a temperature sensor so I could check this for
possible overheating) and this was grounded. Without hooking up the
data aquisition card and just powering up the EPIA M in the seperate
aluminum case with no connections of any kind to the airplane I get the
interference in the radio. There are no connections coming out of the
aluminum box except for power and this line has caps on it for
filtering.

So it is coming solely from the motherboard, some chip on the board is
oscillating in the 107~130 Mhz range, given this occurs against one
home computer running at 2Ghz, another at 2.4 Ghz and the EPIA and 500
Mhz I would guess it is some supporting chip, but even if I knew I am
not sure that would do me much good. Perhaps changing the aluminum case
to a different size/shape would catch the offending frequency but I am
in the dark here and would be shooting in the dark.

There only seems two generic routes,
A) find a small computer which does not emit these frequencies (but I
do not know what is emmiting them so I would not know which computer
choice would alleviate this)
B) Some vastly better shielding approach for the motherboard.

Ideas?

Thanks

Bob wrote:
> The "noise" could be coming through the connector and could be
emitting
> from the wires.
>
> Try using a connector with a metal housing and backshell and wires
with
> shields, terminate the shields to the connector backshell/housing
which
> will ground to the chassis.
>
> The other end of the wires, try to terminate their shields where they
> connect to.
>
> Aircraft environment is not the same as home environment. Different
> applications different techniques.
>
> jcpearce wrote:
> > As a pet/learning project I made a data aquisition unit using an
8051
> > microprocessor and an EPIA M motherboard running a variant of Linux
> to
> > process and display the information. It all works but the EMI from
> the
> > EPIA M causes way too much noise to the aircraft radios. I have
tried
> > shielding the whole device in an aluminum case with very little
> > improvement.
> >
> > Any ideas on how to smother the EMI or some other small motherboard
> > which may not have as much an issue (as a test I took my portable
> > aviation radio and within 6' of any my home computers the same
occurs
> > which gives me little hope)
> >
> > Thanks

ELIPPSE

January 26th 05, 06:26 PM

jcpearce wrote:
> As a pet/learning project I made a data aquisition unit using an 8051
> microprocessor and an EPIA M motherboard running a variant of Linux
to
> process and display the information. It all works but the EMI from
the
> EPIA M causes way too much noise to the aircraft radios. I have tried
> shielding the whole device in an aluminum case with very little
> improvement.
>
> Any ideas on how to smother the EMI or some other small motherboard
> which may not have as much an issue (as a test I took my portable
> aviation radio and within 6' of any my home computers the same occurs
> which gives me little hope)
>
> Thanks

Hi, JC!
As others have mentioned, it is important that the conductors carrying
signal and power in and out of the case be prevented from carrying the
RFI/EMI out of the box; ferrite beads and high-quality ceramic caps or
filter connectors can take care of that problem. Try not to use too
high a value of cap as the self-resonant frequency decreases with size.
Usually around 100pf is good at VHF. Your circuits have many frquency
dividers which reduce the initial oscillator frequency down to much
lower values. These square-waves are rich in harmonics. If you tune an
FM receiver or a VHF receiver across each band you can recognize the
various harmonics by seeing where the amplitude peaks. You can make a
list of these and see the periodicity of the responsible waveform. The
rf energy inside your box induces currents in the metallic structure of
the box. Any gaps, no matter how small, form a slot antenna which
radiates. Each fastener around the periphery of the box where the sides
and top and bottom attach when tightened cause a slight arch in the
surfaces which form the slots. Think of the pan and rocker-arm covers
opn a car engine! There are several ways to stop this. Use conductive
elastomer gaskets at each joint, make the box out of extremely thick
material with machined mating surfaces with joggles, or cover the box
with rf absorbent material with an overall metal cover such as aluminum
foil. Any rf energy leaking out of the box gaps will be absorbed
somewhat as they pass through the absorber, will reradite from the foil
back through the material, and then through many passes back and forth
be attenuated. One of the manufacturers of rf gaskets has an excellent
description of the radiation from a box in their product manual. Paul

jcpearce

January 26th 05, 06:41 PM

Thanks for the input

Aluminum can act as an RF shield but not as a magnetic field shield,
given that the interference is in the RF band I figured aluminum would
be OK (and to keep it light) but some iron based material would do both
and perhaps is a better choice.

Using the PC architecture has so many advantages (and obviously at the
moment a showstopper bug), I can easily store the history of a trip on
a removable USB card, on this USB card is a directory for music that
can be played and piped into the stereo intercom, high tech displays
with graphs, charts, etc.. is easy, it is inexpensive, and easy to
program with high level languages. Doing the assembly on the 8051 data
controller was interesting but to do the same functions as described
above would be very very hard (for me). I have gone so far down this
road I am loath to junk it (even if that is the right thing to do) and
tell myself if I bang my head against the wall enough times the answer
will show up. I know people use their laptops in cockpits without this
problem so there must be a way.

[email protected]

January 26th 05, 07:51 PM

RFI is definitely a problem, and fixing RFI problems can be a black
art.
You don't actually have to have _any_ chip running at the problem
frequency. Square waves have a lot of odd harmonics. So, something
switching regularly at 1/3 or 1/5 the problem frequency can cause
problems. That something can be a software routine in a chip running
much faster than the problem frequency.

You might want to reconsider your preferred form factor. PC
motherboards are amazingly inexpensive, but they're relatively large.
Have you considered the PDA form-factor? Smaller, less power-hungry,
built-in display, and most of the RF problems will be already handled.
Should still be able to handle the audio and display functions.

If you prefer Linux over PocketPC or Palm OS, I know the Zaurus PDA has
had Linux ported to it.

Glider pilots use PDA's as glide computers, so they're known to run
without problems in systems with aviation band radios. If you're
sending the data from the 8051 via the serial port, this is how the
glide computer PDA's interface with the GPS and/or vario systems.

Downside is, RS232 ports seem to be going away in favor of USB, and USB
OTG (where the PDA can act as a host instead of a peripheral) seems to
be very slow in coming in. OTOH, older PDAs with serial ports are
dirt cheap on EBay.

And, for 8051 chips, the Dallas 89C440 is a pretty cool little chip.
It can run internally at up to 4x the xtal frequency, 1 machine cycle/
clock instead of 12.
Flash, so you can reprogram it in-circuit. Built-in loader so all you
need to program it is a PC and a serial port. 32K bytes of program
space. (64K on the 89C450)
2K bytes of xdata memory on chip, and the usual 256 bytes of the 8052.
There are C compilers for the 8051 architecture, so you're really not
limited to assembly for development. About five bucks a pop, small
quantities.

Tim Ward

[email protected]

January 26th 05, 08:30 PM

I think your idea of putting a noisey mother board inside an aluminum
box for shielding is a reasonable one. The problem may be that the
aluminum box your're thinking of as a shield, looks like an antenna to
your VHF receiver. You need to make the potential of the shield box be
the same as the ground of the receiver. The lower he inductance this
connection is, the more alike the grounds will be and the receiver
won't be able to notice the potential of the case going up and down at
100MHz.

Some other things, make sure the ground of your MOBO is connected with
a thick stap to the case.

Solving EMI problems is like peeling an onion. Each layer makes you
wanna cry. You hit the low order effects first because they have the
largest effect and are easiest to fix. Eventually you get to a place
of diminishing returns (EMI gaskets, etc) were you can live with the
"birds" and squelch them out and live with the loss of sensitivity on
some channels of your receiver.

Regards

jcpearce wrote:
> As a pet/learning project I made a data aquisition unit using an 8051
> microprocessor and an EPIA M motherboard running a variant of Linux
to
> process and display the information. It all works but the EMI from
the
> EPIA M causes way too much noise to the aircraft radios. I have tried
> shielding the whole device in an aluminum case with very little
> improvement.
>
> Any ideas on how to smother the EMI or some other small motherboard
> which may not have as much an issue (as a test I took my portable
> aviation radio and within 6' of any my home computers the same occurs
> which gives me little hope)
>
> Thanks

Don Hammer

January 26th 05, 09:35 PM

Do a Google search on mumetal. Some years ago I had a shielding
problem on a Gulfstream and was able to kill it with a mumetal shield.

ELIPPSE

January 26th 05, 11:07 PM

jcpearce wrote:
> Thanks for the input
>
> Aluminum can act as an RF shield but not as a magnetic field shield,
> given that the interference is in the RF band I figured aluminum
would
> be OK (and to keep it light) but some iron based material would do
both
> and perhaps is a better choice.
>
> Using the PC architecture has so many advantages (and obviously at
the
> moment a showstopper bug), I can easily store the history of a trip
on
> a removable USB card, on this USB card is a directory for music that
> can be played and piped into the stereo intercom, high tech displays
> with graphs, charts, etc.. is easy, it is inexpensive, and easy to
> program with high level languages. Doing the assembly on the 8051
data
> controller was interesting but to do the same functions as described
> above would be very very hard (for me). I have gone so far down this
> road I am loath to junk it (even if that is the right thing to do)
and
> tell myself if I bang my head against the wall enough times the
answer
> will show up. I know people use their laptops in cockpits without
this
> problem so there must be a way.

Hi, JC!
Actually, any highly conductive box with minimal gaps will shield
against magnetic field intrusion. The way it works is that the incident
magnetic field generates a current in the metal surface, and this
current then in itself produces a magnetic field which opposes the
original one. That is why it is very important to use highly conductive
material and to have highly conductive, well-sealed joints. By the way,
RF doesn't reflect from a metal surface; the field generates a current
in the surface, which in turn radiates the energy, This is why the
counterpoise (ground-plane) to be effective must be of high
conductivity for good 1/4 wave antenna radiation. Aluminum and copper
make very good enclosures, both for RFI and EMI. Soft aluminum, such as
1100 0, is the best and has an IACS relative resistance of 1.69 making
it more conductive than some of the harder alloys such as 2017 T4 and
2024 T4 which are 3.33, double that of 1100, or 5056 H18 at 3.70. Don't
use brass; it's not as good as aluminum. I had a small circuit that had
to have extreme isolation from incident fields to work properly. I made
a box out of double-sided circuit board, and soldered all the inner
surfaces together as well as all of the outer surfaces except the lid,
which had conductive fingers all the way around on the inside, then
soldered on the outside. Worked great! You might try this if it is just
to be a one-off.
Here're some references you might find informative: Interference
Handbook, Nelson, Radio Publications Inc., 1981; Grounding and
Shielding Techniques in Instrumentation, Morrison, Wiley, 1967; Good
grounding and shielding practices, Electronic Design, 1977 Jan. 04,
p.110; Sniffer probe locates sources of EMI, EDN, 1998 June 04, p.155,
as well as the previously mentioned catalogs, which are from Chomerics
and Metex. Amuneal Mfg. has mumetal shields. Paul

AINut

January 27th 05, 02:06 PM

Look up info on "Tempest" PC's. Lots of good methods for RFI suppression.

jcpearce wrote:
> Thanks for the suggestions, no I have not posted what I did anywhere
> but I think I will as it took me a little while to learn rudimentary
> assembly, come up with a variant of linux running off flash memory,
> write the code to process the serial output and display etc..
>
> I did not use a hobby aluminum case, I custom built one with no gaps
> (motherboard has a temperature sensor so I could check this for
> possible overheating) and this was grounded. Without hooking up the
> data aquisition card and just powering up the EPIA M in the seperate
> aluminum case with no connections of any kind to the airplane I get the
> interference in the radio. There are no connections coming out of the
> aluminum box except for power and this line has caps on it for
> filtering.
>
> So it is coming solely from the motherboard, some chip on the board is
> oscillating in the 107~130 Mhz range, given this occurs against one
> home computer running at 2Ghz, another at 2.4 Ghz and the EPIA and 500
> Mhz I would guess it is some supporting chip, but even if I knew I am
> not sure that would do me much good. Perhaps changing the aluminum case
> to a different size/shape would catch the offending frequency but I am
> in the dark here and would be shooting in the dark.
>
> There only seems two generic routes,
> A) find a small computer which does not emit these frequencies (but I
> do not know what is emmiting them so I would not know which computer
> choice would alleviate this)
> B) Some vastly better shielding approach for the motherboard.
>
> Ideas?
>
> Thanks
>
>
>
> Bob wrote:
>
>>The "noise" could be coming through the connector and could be
>
> emitting
>
>>from the wires.
>>
>>Try using a connector with a metal housing and backshell and wires
>
> with
>
>>shields, terminate the shields to the connector backshell/housing
>
> which
>
>>will ground to the chassis.
>>
>>The other end of the wires, try to terminate their shields where they
>>connect to.
>>
>>Aircraft environment is not the same as home environment. Different
>>applications different techniques.
>>
>>jcpearce wrote:
>>
>>>As a pet/learning project I made a data aquisition unit using an
>
> 8051
>
>>>microprocessor and an EPIA M motherboard running a variant of Linux
>>
>>to
>>
>>>process and display the information. It all works but the EMI from
>>
>>the
>>
>>>EPIA M causes way too much noise to the aircraft radios. I have
>
> tried
>
>>>shielding the whole device in an aluminum case with very little
>>>improvement.
>>>
>>>Any ideas on how to smother the EMI or some other small motherboard
>>>which may not have as much an issue (as a test I took my portable
>>>aviation radio and within 6' of any my home computers the same
>
> occurs
>
>>>which gives me little hope)
>>>
>>>Thanks
>
>

Bob

January 27th 05, 02:13 PM

Some IMPRESSIVE experience in this group.

I have not heard "tempest" in years.

But again, in my experience, if you've got a metal box already. It's
most likely the connector and harness.

Carry those grounds (on the harness shields) through!

AINut wrote:
> Look up info on "Tempest" PC's. Lots of good methods for RFI
suppression.
>
>
> jcpearce wrote:
> > Thanks for the suggestions, no I have not posted what I did
anywhere
> > but I think I will as it took me a little while to learn
rudimentary
> > assembly, come up with a variant of linux running off flash memory,
> > write the code to process the serial output and display etc..
> >
> > I did not use a hobby aluminum case, I custom built one with no
gaps
> > (motherboard has a temperature sensor so I could check this for
> > possible overheating) and this was grounded. Without hooking up the
> > data aquisition card and just powering up the EPIA M in the
seperate
> > aluminum case with no connections of any kind to the airplane I get
the
> > interference in the radio. There are no connections coming out of
the
> > aluminum box except for power and this line has caps on it for
> > filtering.
> >
> > So it is coming solely from the motherboard, some chip on the board
is
> > oscillating in the 107~130 Mhz range, given this occurs against one
> > home computer running at 2Ghz, another at 2.4 Ghz and the EPIA and
500
> > Mhz I would guess it is some supporting chip, but even if I knew I
am
> > not sure that would do me much good. Perhaps changing the aluminum
case
> > to a different size/shape would catch the offending frequency but I
am
> > in the dark here and would be shooting in the dark.
> >
> > There only seems two generic routes,
> > A) find a small computer which does not emit these frequencies (but
I
> > do not know what is emmiting them so I would not know which
computer
> > choice would alleviate this)
> > B) Some vastly better shielding approach for the motherboard.
> >
> > Ideas?
> >
> > Thanks
> >
> >
> >
> > Bob wrote:
> >
> >>The "noise" could be coming through the connector and could be
> >
> > emitting
> >
> >>from the wires.
> >>
> >>Try using a connector with a metal housing and backshell and wires
> >
> > with
> >
> >>shields, terminate the shields to the connector backshell/housing
> >
> > which
> >
> >>will ground to the chassis.
> >>
> >>The other end of the wires, try to terminate their shields where
they
> >>connect to.
> >>
> >>Aircraft environment is not the same as home environment.
Different
> >>applications different techniques.
> >>
> >>jcpearce wrote:
> >>
> >>>As a pet/learning project I made a data aquisition unit using an
> >
> > 8051
> >
> >>>microprocessor and an EPIA M motherboard running a variant of
Linux
> >>
> >>to
> >>
> >>>process and display the information. It all works but the EMI from
> >>
> >>the
> >>
> >>>EPIA M causes way too much noise to the aircraft radios. I have
> >
> > tried
> >
> >>>shielding the whole device in an aluminum case with very little
> >>>improvement.
> >>>
> >>>Any ideas on how to smother the EMI or some other small
motherboard
> >>>which may not have as much an issue (as a test I took my portable
> >>>aviation radio and within 6' of any my home computers the same
> >
> > occurs
> >
> >>>which gives me little hope)
> >>>
> >>>Thanks
> >
> >

COLIN LAMB

January 27th 05, 02:58 PM

You are experiencing a problem that can be solved easily, using techniques
known by radio amateurs. In particular, The Radio Handbook, bu Bill Orr,
has a chart showing the attenuation of signal levels by various bypassing
methods used at the case of a shielded piece of equipment.

At 100 MHz, the attenuation of a standard bypass capacitor is not very high.
Additional filtering will reduce the level to where you cannot hear the
noise coming from the microprocessor. There are special capacitors, called
feedthrough capacitors, which do a great job of attenuating the signal
level, and that combined with a pi-network will result in no signal coming.

I would suggest that you borrow a high bandwidth scope (200 to 400 MHz) to
look at the levels coming out.

Where ultimate attenuation is required, double shielding is often used,
however I do not think you need that level of attenuation.

Colin N12HS

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.783 / Virus Database: 529 - Release Date: 10/25/04

ELIPPSE

January 28th 05, 02:23 AM

COLIN LAMB

January 28th 05, 03:41 AM

"Here's a couple more ideas on containing signals in the box. First, have
plenty of closely spaced fasteners along the joints; not just in the
corners. Second, the thin copper desoldering braid makes a fairly good RF
gasket. 'Best with your project!"

I would not use desoldering braid. Copper reacts with aluminum. The
electrolysis will form aluminum oxide, which is an excellent insulator, and
the rf tight enclosure will no longer be secure. It could also cause
rectification, and allow external signals to mix, resulting in many spurious
signals.

Aluminum likes stainless steel, since the electric qualities are similar.
Aluminum and copper plus moisture will form a battery.

Colin N12HS

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.783 / Virus Database: 529 - Release Date: 10/25/04

ELIPPSE

January 28th 05, 06:06 PM

COLIN LAMB wrote:
> "Here's a couple more ideas on containing signals in the box. First,
have
> plenty of closely spaced fasteners along the joints; not just in the
> corners. Second, the thin copper desoldering braid makes a fairly
good RF
> gasket. 'Best with your project!"
>
> I would not use desoldering braid. Copper reacts with aluminum. The
> electrolysis will form aluminum oxide, which is an excellent
insulator, and
> the rf tight enclosure will no longer be secure. It could also cause
> rectification, and allow external signals to mix, resulting in many
spurious
> signals.
>
> Aluminum likes stainless steel, since the electric qualities are
similar.
> Aluminum and copper plus moisture will form a battery.
>
> Colin N12HS
> Hi, Colin!
Excellent consideration! Thanks for the heads-up! Paul
>
> ---
> Outgoing mail is certified Virus Free.
> Checked by AVG anti-virus system (http://www.grisoft.com).
> Version: 6.0.783 / Virus Database: 529 - Release Date: 10/25/04

Dean Wilkinson

January 28th 05, 11:47 PM

RFI is best squelched at the souce, i.e. on the PC board. The biggest
source of RFI is the clock that drives the processor, etc. If it isn't
routed on the board very well or properly terminated, it will radiate like a
pig. Also, effective grounding of the PC board to the case is important.

Given that you have a noisy PC board, the only way to effectively shield it
is the fully enclose it in a conductive case, preferrable made out of steel
since that will also block the magnetic portion of the radiation. Any gaps
or slots in the case that are as large as the wavelengths of the radiated
signals will allow the signals to escape. In an airplane, aluminum cases
are preferred due to their light weight and lack of magnetic effects on the
compass.

Also, wires that run out of the enclosure will often carry the RFI signals
and re-radiate them by acting as antennas unless they are also well shielded
and properly terminated to the case ground. Ferrite cores can also be
effective in blocking such signals from traveling out through the wires.

Dean

These are all band-aids for the original problem of having an overly noisy
PC board.
"jcpearce" > wrote in message
ups.com...
> As a pet/learning project I made a data aquisition unit using an 8051
> microprocessor and an EPIA M motherboard running a variant of Linux to
> process and display the information. It all works but the EMI from the
> EPIA M causes way too much noise to the aircraft radios. I have tried
> shielding the whole device in an aluminum case with very little
> improvement.
>
> Any ideas on how to smother the EMI or some other small motherboard
> which may not have as much an issue (as a test I took my portable
> aviation radio and within 6' of any my home computers the same occurs
> which gives me little hope)
>
> Thanks
>

jcpearce

January 30th 05, 04:55 AM

Thanks Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc & Tim, I
really appreciate the advice.

Your comments rekindle my drive, I had hit a low point after designing
the data capture board, OS, circuits etc.. and then... being killed
with RF noise.

An odd, but beneficial I suppose, confluence of circumstances
involving a broken window pane and a mischievous dog and a storm lead
to water getting on the DC to DC power supply (11~14 DC -> 12V, 5V,
3.3V +-) for the EPIA M motherboard. Not surprisingly it decided not
to continue functioning. I then pulled out a traditional power supply
to use, and.. amazingly 80% of the noise went away! I would hate to
curse but the blasted PW-60 #!@! (
http://store.ituner.com/ituner/pw60caatxpos.html ) power supply was the
primary culprit. It is a switching power supply with evidently little
RF and EMI protection. I am trying to find another DC->DC power supply
without these issues, they all appear to be switching power supplies, I
would rather deal with energy/heat loss than RF noise.

I don't want to go through the space and elaborate hastle of going from
the 11-14V bus to 120V AC to a bulky PC power supply (space is
critical) to 12,5,3.3V so I need to find a suitable non noisy power
supply and then deal with the issues you knowledgeable fellows
enumerated in the posts.

Thanks again to Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc &
Tim.
Your comments rekindle my drive, I had hit a low point after designing
the data capture board, OS, circuits etc.. and then... being killed
with RF noise.

An odd, but beneficial I suppose, confluence of circucumstances
involving a broken window pane and a mischevious dog and a storm lead
to water getting on the DC to DC power supply (11~14 DC -> 12V, 5V,
3.3V +-) for the EPIA M motherboard. Not surprisingly it decided not
to continue functioning. I then pulled out a traditional power supply
to use, and.. amazingly 90% of the noise went away! I would hate to
curse but the blasted PW-60 #!@! (
http://store.ituner.com/ituner/pw60caatxpos.html ) power supply was the
primary culprit. It is a switching power supply with evidently all
the wrong harmonics. I am trying to find another DC->DC power supply
without these issues, they all appear to be switching power supplies, I
would rather deal with energy/heat loss than RF noise.

I don't want to go through the space and elaborate hastle of going from
the 11-14V bus to 120V AC to a bulky PC power supply (space is
critical) to 12,5,3.3V so I need to find a suitable non noisy power
supply and then deal with the issues you knowledgable fellows
enumerated in the posts.

Thanks again to Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc &
Tim.

AINut

January 30th 05, 05:13 AM

You're welcome! Keep us informed, please, of your trials and successes.

jcpearce wrote:
> Thanks Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc & Tim, I
> really appreciate the advice.
>
> Your comments rekindle my drive, I had hit a low point after designing
> the data capture board, OS, circuits etc.. and then... being killed
> with RF noise.
>
> An odd, but beneficial I suppose, confluence of circumstances
> involving a broken window pane and a mischievous dog and a storm lead
> to water getting on the DC to DC power supply (11~14 DC -> 12V, 5V,
> 3.3V +-) for the EPIA M motherboard. Not surprisingly it decided not
> to continue functioning. I then pulled out a traditional power supply
> to use, and.. amazingly 80% of the noise went away! I would hate to
> curse but the blasted PW-60 #!@! (
> http://store.ituner.com/ituner/pw60caatxpos.html ) power supply was the
> primary culprit. It is a switching power supply with evidently little
> RF and EMI protection. I am trying to find another DC->DC power supply
> without these issues, they all appear to be switching power supplies, I
> would rather deal with energy/heat loss than RF noise.
>
> I don't want to go through the space and elaborate hastle of going from
> the 11-14V bus to 120V AC to a bulky PC power supply (space is
> critical) to 12,5,3.3V so I need to find a suitable non noisy power
> supply and then deal with the issues you knowledgeable fellows
> enumerated in the posts.
>
> Thanks again to Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc &
> Tim.
> Your comments rekindle my drive, I had hit a low point after designing
> the data capture board, OS, circuits etc.. and then... being killed
> with RF noise.
>
> An odd, but beneficial I suppose, confluence of circucumstances
> involving a broken window pane and a mischevious dog and a storm lead
> to water getting on the DC to DC power supply (11~14 DC -> 12V, 5V,
> 3.3V +-) for the EPIA M motherboard. Not surprisingly it decided not
> to continue functioning. I then pulled out a traditional power supply
> to use, and.. amazingly 90% of the noise went away! I would hate to
> curse but the blasted PW-60 #!@! (
> http://store.ituner.com/ituner/pw60caatxpos.html ) power supply was the
> primary culprit. It is a switching power supply with evidently all
> the wrong harmonics. I am trying to find another DC->DC power supply
> without these issues, they all appear to be switching power supplies, I
> would rather deal with energy/heat loss than RF noise.
>
> I don't want to go through the space and elaborate hastle of going from
> the 11-14V bus to 120V AC to a bulky PC power supply (space is
> critical) to 12,5,3.3V so I need to find a suitable non noisy power
> supply and then deal with the issues you knowledgable fellows
> enumerated in the posts.
>
> Thanks again to Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc &
> Tim.
>

UltraJohn

January 30th 05, 04:41 PM

> curse but the blasted PW-60 #!@! (
> http://store.ituner.com/ituner/pw60caatxpos.html ) power supply was the
> primary culprit. It is a switching power supply with evidently all
> the wrong harmonics. I am trying to find another DC->DC power supply
> without these issues, they all appear to be switching power supplies, I
> would rather deal with energy/heat loss than RF noise.
>

The 110v power supplies used in desktop computers are also switching
supplies. So the problem is the cheapness of the 12v unit you selected.
I'd try another make unit and see if that works if not try putting the power
supply in its own shielded case and shielding and ferrite beading the
input/output leads from it.
Good luck and let us know what you workout!
John

Dean Wilkinson

January 31st 05, 06:23 PM

"jcpearce" > wrote in message
ups.com...
> Thanks Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc & Tim, I
> really appreciate the advice.
>
> Your comments rekindle my drive, I had hit a low point after designing
> the data capture board, OS, circuits etc.. and then... being killed
> with RF noise.
>
> An odd, but beneficial I suppose, confluence of circumstances
> involving a broken window pane and a mischievous dog and a storm lead
> to water getting on the DC to DC power supply (11~14 DC -> 12V, 5V,
> 3.3V +-) for the EPIA M motherboard. Not surprisingly it decided not
> to continue functioning. I then pulled out a traditional power supply
> to use, and.. amazingly 80% of the noise went away! I would hate to
> curse but the blasted PW-60 #!@! (
> http://store.ituner.com/ituner/pw60caatxpos.html ) power supply was the
> primary culprit. It is a switching power supply with evidently little
> RF and EMI protection. I am trying to find another DC->DC power supply
> without these issues, they all appear to be switching power supplies, I
> would rather deal with energy/heat loss than RF noise.
>
> I don't want to go through the space and elaborate hastle of going from
> the 11-14V bus to 120V AC to a bulky PC power supply (space is
> critical) to 12,5,3.3V so I need to find a suitable non noisy power
> supply and then deal with the issues you knowledgeable fellows
> enumerated in the posts.
>
> Thanks again to Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc &
> Tim.
> Your comments rekindle my drive, I had hit a low point after designing
> the data capture board, OS, circuits etc.. and then... being killed
> with RF noise.
>
> An odd, but beneficial I suppose, confluence of circucumstances
> involving a broken window pane and a mischevious dog and a storm lead
> to water getting on the DC to DC power supply (11~14 DC -> 12V, 5V,
> 3.3V +-) for the EPIA M motherboard. Not surprisingly it decided not
> to continue functioning. I then pulled out a traditional power supply
> to use, and.. amazingly 90% of the noise went away! I would hate to
> curse but the blasted PW-60 #!@! (
> http://store.ituner.com/ituner/pw60caatxpos.html ) power supply was the
> primary culprit. It is a switching power supply with evidently all
> the wrong harmonics. I am trying to find another DC->DC power supply
> without these issues, they all appear to be switching power supplies, I
> would rather deal with energy/heat loss than RF noise.
>
> I don't want to go through the space and elaborate hastle of going from
> the 11-14V bus to 120V AC to a bulky PC power supply (space is
> critical) to 12,5,3.3V so I need to find a suitable non noisy power
> supply and then deal with the issues you knowledgable fellows
> enumerated in the posts.
>
> Thanks again to Dean, Elippse, Colin Lamb, Bob, AINut Don, Kayroc &
> Tim.
>

Ahhh.... yes. I should have thought to mention the importance of a quiet
power supply. A PI filter between the supply and the PC is a good idea to
help ensure that switching noise from your supply doesn't get into your PC
board.

I'm used to designing my own power supplies for systems like this, so I
didn't stop to think that you might be using an off-the-shelf supply.

Dean

Evan Carew

February 1st 05, 02:15 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

JC,

Having tackled this issue myself for other apps, I can point you to an
excellent resource, "The Art Of Electronics" by Horowitz & Hill. In
there, they outline some techniques witch you may consider paranoid, but
I have found to work. For instance, building an enclosure within an
enclosure. The outer enclosure is little more than a way to terminate
your outside wire connections (including power), allowing you to bring
them into a controlled environment where you can then apply such gross
tricks as caps & ferrite filters on your power lines and data lines
(don't forget fuses!). After you have performed whatever magic you can
on the in/outbound leads, you then bring the signal & power into the
inner enclosure.

Also remember that for homemade enclosures of bent & riveted aluminum,
small gaps where one flap is riveted to another can have gaps in then
which WILL leak RF. To fix this, your local electronics parts supplier
(or Digikey) will sell conductive tape. Welding also works. When
attaching the top of the enclosure, you could employ either one of those
fine mesh springy wire gaskets, or ... more tape.

Good luck,
Evan Carew

P.S.
IF you want me to forward some of the specific steps outlined in TAOE,
let me know.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFB/48apxCQXwV2bJARAl5fAJ9SEYjU1Cv42YAG12kRkQVl2034PwC fV4Wv
SmcILTAIIIE+cHexj2fbBE0=
=xelG
-----END PGP SIGNATURE-----

Ernest Christley

February 5th 05, 02:31 AM

> jcpearce wrote:
>> So it is coming solely from the motherboard, some chip on the board is
>> oscillating in the 107~130 Mhz range, given this occurs against one

That frequency is awful close to the bus frequency of modern systems
(133Mhz). If you're using DIMM memory, do you have a card in every
slot, or do you have some of those little radiators hanging free?

jcpearce

February 7th 05, 03:01 AM

Thanks Evan, I have eyed that book at the local bookstore and given
that you mention there is a section (and paranoid no less, a good
thing) on this I'll purchase a copy. Books like this can save a lot of
hair being pulled out.

The enclosure I made is riveted and there are very small gaps in
addition to where the mandrel pulls out leaving a small hole. Would
encompassing the whole thing in aluminum foil (as a test, not for
deployment) accurately simulate these sophisticated seals? Before going
through that effort I would want to know what, if any, gain I may
derive.

Ernest the MB only has one DDR slot, where a single 512 Mb card is
plugged in.

I am encouraged that if the power supply issue can be solved via
shielding I will be set. The people at ituner admitted the PW-60 DC-DC
power supply was a noise daemon but there new super improved model the
M1-ATX ( http://www.mini-box.com/s.nl/sc.8/category.13/it.A/id.356/.f )
model would solve this and they are sending me a unit, they are willing
to work with me which is quite good customer service.

If it is still too noisy I'll get paranoid with cases within cases,
welding, prayer, etc... If still too noisy (please no) I'll do my own
power supply,

Thanks for the replies.

Evan Carew

February 9th 05, 03:52 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

JC,

[Comments follow your text]

jcpearce wrote:
> Thanks Evan, I have eyed that book at the local bookstore and given
> that you mention there is a section (and paranoid no less, a good
> thing) on this I'll purchase a copy. Books like this can save a lot of
> hair being pulled out.
>
> The enclosure I made is riveted and there are very small gaps in
> addition to where the mandrel pulls out leaving a small hole. Would
> encompassing the whole thing in aluminum foil (as a test, not for
> deployment) accurately simulate these sophisticated seals? Before going
> through that effort I would want to know what, if any, gain I may
> derive.
>

Well, when last I did this, I remember the RF "bonding" tape was nickel
plated coper foil with a collodial silver stickey backing. As I
remember, it worked perfectly, if somewhat expensively. Since this was a
little more than 10 years ago, I imagine that there have been some
cheaper alternatives developed since then.

Your problem when using tin foil will be getting it at the precise
potential of the rivited box underneath it. To accomplish this, it needs
to be "bonded", ie electrically grounded to the main box at regular
intervals. What these intervals is, has either to be determined
experimentally, or solved via the paranoid method (nickel plated coper
tape with $ilver......).

Ultimately, I think the better solution would be to get some cheap
double or single sided FR11 PC board and solder it together with coper
foil (cheap at your local electrical supply). To seal the top, you are
going to have to order some king of conductive compression tape (cheap
from digikey.

If you are interested, email me off group & I'll see if I can find you
some hardware ideas from Digikey.

Evan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCCjHZpxCQXwV2bJARAhTAAJkBU0SgGlHr5CBIgFPOac FAc0nTAwCeOEWI
rqXJENBMFEO4l079skK5oHs=
=ZpGE
-----END PGP SIGNATURE-----

Morgans

February 9th 05, 09:59 PM

"Evan Carew" > wrote

Well, when last I did this, I remember the RF "bonding" tape was nickel
> plated coper foil with a collodial silver stickey backing. As I
> remember, it worked perfectly, if somewhat expensively.

Sounds similar to something you can get in the heating and cooling sections
of the big box hardware stores. It is thin aluminum tape with sticky on it.
Seems like reversing it once in a while, or at the ends would get the al to
al contact, to bypass the non conductive sticky.

I have not tried it, but advice is worth what you pay for it! <g>
--
Jim in NC

Evan Carew

February 10th 05, 04:33 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Morgans wrote:
> "Evan Carew" > wrote
>
> Well, when last I did this, I remember the RF "bonding" tape was nickel
>
>>plated coper foil with a collodial silver stickey backing. As I
>>remember, it worked perfectly, if somewhat expensively.
>
>
> Sounds similar to something you can get in the heating and cooling sections
> of the big box hardware stores. It is thin aluminum tape with sticky on it.
> Seems like reversing it once in a while, or at the ends would get the al to
> al contact, to bypass the non conductive sticky.
I have a feeling this would suffer from the same problem that riviting
suffers from, namely small gaps.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCC4zhpxCQXwV2bJARAkhVAJsGG5pgDcA3y4O8d+wgB4 JoQRTJyQCguxam
8wu9R6NIWX7xfaxKfTh5mqo=
=molB
-----END PGP SIGNATURE-----

Gary Thomas

February 10th 05, 06:57 PM

Evan Carew wrote:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Morgans wrote:
> > "Evan Carew" > wrote
> >
> > Well, when last I did this, I remember the RF "bonding" tape was nickel
> >
> >>plated coper foil with a collodial silver stickey backing. As I
> >>remember, it worked perfectly, if somewhat expensively.
> >
> >
> > Sounds similar to something you can get in the heating and cooling sections
> > of the big box hardware stores. It is thin aluminum tape with sticky on it.
> > Seems like reversing it once in a while, or at the ends would get the al to
> > al contact, to bypass the non conductive sticky.
> I have a feeling this would suffer from the same problem that riviting
> suffers from, namely small gaps.
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.0.6 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFCC4zhpxCQXwV2bJARAkhVAJsGG5pgDcA3y4O8d+wgB4 JoQRTJyQCguxam
> 8wu9R6NIWX7xfaxKfTh5mqo=
> =molB
> -----END PGP SIGNATURE-----

McMaster Carr sells copper tape 1/4" wide to 3" wide with conductive
adhesive. The adhesive surface resistivity is .oo65 ohms/inch. The
copper foil is .0014" thick. Adhesive thickness is .0035". A six yard
roll varies from ~$6 to $42 depending on width. 1" width sells for ~$15
for six yards part# 76555A714.

Gary Thomas

jcpearce

February 13th 05, 06:27 AM

Thank you everyone for the input, now I am looking for some
prioritization. I received the new power supply from www.mini-box.com
and it is substantially better than the previous version
(http://www.mini-box.com/s.nl/sc.8/category.13/it.A/id.356/.f ) however
I still get some noise in the radios. The new version is too big to
fit in the previous enclosure so it is sitting slightly apart from the
MB on its own.

I have started to jot down what the system wants to do, pictures and
status at
http://69.229.202.54:8080 Under the 'Image' folder one can see the
plane, the EPIA in the nose and the prototype ADC board.

So now I have a case that has to heavily part for the ATX power lines
to come into it & a non cased DC power supply. I have attached ferrite
beads to the power lines coming in/out with little improvement.

It takes time for many of the steps so I am looking for what you folks
think would be the 'biggest bang for the buck', or the right priority
to proceed with. (IE, build a brand new case for everything, build a
seperate case for the PSU, don't build a case right away but do X,Y,Z,
etc..)

Thanks once again for the brains helping noodle this over.

Evan Carew

February 13th 05, 07:58 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

JC,

My comments/questions interspersed with your message:

jcpearce wrote:
> Thank you everyone for the input, now I am looking for some
> prioritization. I received the new power supply from www.mini-box.com
> and it is substantially better than the previous version
> (http://www.mini-box.com/s.nl/sc.8/category.13/it.A/id.356/.f ) however
> I still get some noise in the radios. The new version is too big to
> fit in the previous enclosure so it is sitting slightly apart from the
> MB on its own.
>
> I have started to jot down what the system wants to do, pictures and
> status at
> http://69.229.202.54:8080 Under the 'Image' folder one can see the
> plane, the EPIA in the nose and the prototype ADC board.
>
> So now I have a case that has to heavily part for the ATX power lines
> to come into it & a non cased DC power supply. I have attached ferrite
> beads to the power lines coming in/out with little improvement.
Stupid question here JC, but it looks like you are using a PC style
switching power supply [high noise] to power your standard ATX style PC
board. Since most of the important voltages going into an ATX board are
either 5V or 12V, why aren't you simply using regulators off the plane's
battery to supply the power? It would seem to me to be both cheaper as
well as inherently less noisy (not to mention lighter). You wouldn't
even have to do any special parts sourcing on these regulators as they
are redily available at you local radio shack (karmic regulators anyone?).

For a break out of ATX voltages, see the following link:
http://www.7volts.com/atx2.jpg
>
> It takes time for many of the steps so I am looking for what you folks
> think would be the 'biggest bang for the buck', or the right priority
> to proceed with. (IE, build a brand new case for everything, build a
> seperate case for the PSU, don't build a case right away but do X,Y,Z,
> etc..)
>
> Thanks once again for the brains helping noodle this over.
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCD7F5pxCQXwV2bJARAhMmAKCQkWmHv3zP3v1jPgaekZ pWtSSMaQCeLvo9
RfniUGRXnB7y9YA784UsIwg=
=zPpX
-----END PGP SIGNATURE-----

UltraJohn

February 13th 05, 08:43 PM

Evan Carew wrote:

> board. Since most of the important voltages going into an ATX board are
> either 5V or 12V, why aren't you simply using regulators off the plane's
> battery to supply the power? It would seem to me to be both cheaper as
> well as inherently less noisy (not to mention lighter). You wouldn't
> even have to do any special parts sourcing on these regulators as they
> are redily available at you local radio shack (karmic regulators anyone?).

You still need the minus voltages hence probably still an inverting power
supply.

I'd try mounting both in the same case for starters filtering the input 12v
supply and see where you need to go from there.
John

Evan Carew

February 13th 05, 09:21 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

UltraJohn wrote:
> Evan Carew wrote:
>
>
>>board. Since most of the important voltages going into an ATX board are
>>either 5V or 12V, why aren't you simply using regulators off the plane's
>>battery to supply the power? It would seem to me to be both cheaper as
>>well as inherently less noisy (not to mention lighter). You wouldn't
>>even have to do any special parts sourcing on these regulators as they
>>are redily available at you local radio shack (karmic regulators anyone?).
>
>
>
> You still need the minus voltages hence probably still an inverting power
> supply.
>
>
> I'd try mounting both in the same case for starters filtering the input 12v
> supply and see where you need to go from there.
> John
>
There are many different voltages coming from a power supply. Here is a
quick breakdown of what the different voltages do:

* -5V - Legacy ISA
* -12V - Legacy ISA, Serial Ports (including PS/2)
* +3.3V - Motherboard Logic, AGP, PCI
* +5V - Motherboard Logic, Drive Logic, PCI, ISA
* +12V - Fans, Drive Motors, PCI, ISA
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCD8S/pxCQXwV2bJARApvRAKCLfnctGxRpkQ6AkbF4jtG3i6G8mgCgvm l9
rbKSUS/yb2x5QMbfGlU0skA=
=Pjh/
-----END PGP SIGNATURE-----

jcpearce

February 14th 05, 05:29 AM

Thanks for the input.
Perhaps going with a linear power supply would be the way to go, I did
not as there are already three switching DC-DC power supplies on the
MB, the power efficiency is less than 50% for linear and I would have
to deal with the heat issues. I could just string a bunch of LM78xx and
LM79xx to get the required positive and negative voltages.

I've created a seperate smaller case for the PSU as creating a new
single case would have taken longer and I don't know about what level
of cooling issues I will have. I've noticed the PSU manufacturer (not
surprisingly) has reducd the MTBF when it is in hotter conditions. (but
hey perhaps I could use the heat to warm my cold toes in the Long Ez)
If the weather clears tomorrow I will go flying and measure how much of
a problem I have with it operating.

Thanks for the continuing input.

John

February 14th 05, 09:17 AM

jcpearce wrote:

> Thanks for the input.
> Perhaps going with a linear power supply would be the way to go, I did
> not as there are already three switching DC-DC power supplies on the
> MB, the power efficiency is less than 50% for linear and I would have
> to deal with the heat issues. I could just string a bunch of LM78xx and
> LM79xx to get the required positive and negative voltages.
>
> I've created a seperate smaller case for the PSU as creating a new
> single case would have taken longer and I don't know about what level
> of cooling issues I will have. I've noticed the PSU manufacturer (not
> surprisingly) has reducd the MTBF when it is in hotter conditions. (but
> hey perhaps I could use the heat to warm my cold toes in the Long Ez)
> If the weather clears tomorrow I will go flying and measure how much of
> a problem I have with it operating.
>
> Thanks for the continuing input.
The LM79XX's will do you no good unless you have a source of negative
voltage to feed them. This generally will mean a dc to dc converter which
generally means switching regulators. Try putting your current powersupply
in a case and attach it directly to the MB's case and use feedthrough
capacitors in the common case bulkheads to feed the dc voltage in. Put
fedtrough caps and ferite beads on the dc in to the powersuppply case.
Cheep and will get you a good start to eliminating the noise.
John

John

February 14th 05, 09:33 AM

John wrote:

> jcpearce wrote:
>
>> Thanks for the input.
>> Perhaps going with a linear power supply would be the way to go, I did
>> not as there are already three switching DC-DC power supplies on the
>> MB, the power efficiency is less than 50% for linear and I would have
>> to deal with the heat issues. I could just string a bunch of LM78xx and
>> LM79xx to get the required positive and negative voltages.
>>
>> I've created a seperate smaller case for the PSU as creating a new
>> single case would have taken longer and I don't know about what level
>> of cooling issues I will have. I've noticed the PSU manufacturer (not
>> surprisingly) has reducd the MTBF when it is in hotter conditions. (but
>> hey perhaps I could use the heat to warm my cold toes in the Long Ez)
>> If the weather clears tomorrow I will go flying and measure how much of
>> a problem I have with it operating.
>>
>> Thanks for the continuing input.
> The LM79XX's will do you no good unless you have a source of negative
> voltage to feed them. This generally will mean a dc to dc converter which
> generally means switching regulators. Try putting your current powersupply
> in a case and attach it directly to the MB's case and use feedthrough
> capacitors in the common case bulkheads to feed the dc voltage in. Put
> fedtrough caps and ferite beads on the dc in to the powersuppply case.
> Cheep and will get you a good start to eliminating the noise.
> John

About $1.00 a piece from a couple sources I checked. So about $7.00 for the
PS voltages plus the dc input. (although some are sold only in 10 lots) if
you have multiple projects to use them up in you can order 100 quantity for
about $.50 each.
John

John

February 14th 05, 02:19 PM

John wrote:

>>>> Thanks for the continuing input.
>> The LM79XX's will do you no good unless you have a source of negative
>> voltage to feed them. This generally will mean a dc to dc converter which
>> generally means switching regulators. Try putting your current
>> powersupply in a case and attach it directly to the MB's case and use
>> feedthrough capacitors in the common case bulkheads to feed the dc
>> voltage in. Put fedtrough caps and ferite beads on the dc in to the
>> powersuppply case. Cheep and will get you a good start to eliminating the
>> noise. John
>
> About $1.00 a piece from a couple sources I checked. So about $7.00 for
> the PS voltages plus the dc input. (although some are sold only in 10
> lots) if you have multiple projects to use them up in you can order 100
> quantity for about $.50 each.
> John

Just checked Newark Catalog They have an interesting item:
A feed through teminal block with built-in filter capacitors
part # 95F895X
where x = number of terminals from 2 to 6.

Just something interesting to look into.
They have a lot of ideas for cap/ferite/filter ideas.
John

[email protected]

February 15th 05, 03:58 AM

isn't there a step down treansformer taking 110vAC down to 12vAC then
rectified to 12vDC????

Steve
A&P and Computer Guru (and very interested in your Idea)

John wrote:
> John wrote:
>
> >>>> Thanks for the continuing input.
> >> The LM79XX's will do you no good unless you have a source of
negative
> >> voltage to feed them. This generally will mean a dc to dc
converter which
> >> generally means switching regulators. Try putting your current
> >> powersupply in a case and attach it directly to the MB's case and
use
> >> feedthrough capacitors in the common case bulkheads to feed the dc
> >> voltage in. Put fedtrough caps and ferite beads on the dc in to
the
> >> powersuppply case. Cheep and will get you a good start to
eliminating the
> >> noise. John
> >
> > About $1.00 a piece from a couple sources I checked. So about $7.00
for
> > the PS voltages plus the dc input. (although some are sold only in
10
> > lots) if you have multiple projects to use them up in you can order
100
> > quantity for about $.50 each.
> > John
>
>
> Just checked Newark Catalog They have an interesting item:
> A feed through teminal block with built-in filter capacitors
> part # 95F895X
> where x = number of terminals from 2 to 6.
>
> Just something interesting to look into.
> They have a lot of ideas for cap/ferite/filter ideas.
> John

John

February 15th 05, 06:52 AM

wrote:

> isn't there a step down treansformer taking 110vAC down to 12vAC then
> rectified to 12vDC????
>
>
> Steve
> A&P and Computer Guru (and very interested in your Idea)
>
All kinds unfortunately this is for aircraft use, he's starting with 12v DC
(transformers don't work with DC) and needs to get +-12v +5 and +3.3v all
DC.
John

John

February 15th 05, 01:52 PM

>> All kinds unfortunately this is for aircraft use, he's starting with 12v
>> DC (transformers don't work with DC) and needs to get +-12v +5 and +3.3v
>> all DC.
>> John
>>
> John,
>
> I think you'll find that the -12V is used only in "legacy" PC design for
> ISA busses, or serial ports (which he probably isn't using).
>
> Evan

Yes I did a little research on it and the only thing the negative is needed
for is the rs232 serial ports. for this you can get a $1.50 ic to convert
+12 to -12 V low current.
That said, what I'd do is get a 1.5 amp 5v TO-3 regulator and a 1 amp TO-220
regulator along with one of the TO-220 adjustable 1 amp regulators for the
3.3 v and run it all off the 12V (highly filtered) input. That said,
(again) you might have problems with th 12V regulator if the input drops
below 13.8V! So you might need to use a separate battery for it and diode
isolate it from the aircrafts electrical systm and have it charged when it
exceeds the 13.8 volt figure.
John

Evan Carew

February 15th 05, 02:43 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

John wrote:
> wrote:
>
>
>>isn't there a step down treansformer taking 110vAC down to 12vAC then
>>rectified to 12vDC????
>>
>>
>>Steve
>>A&P and Computer Guru (and very interested in your Idea)
>>
>
> All kinds unfortunately this is for aircraft use, he's starting with 12v DC
> (transformers don't work with DC) and needs to get +-12v +5 and +3.3v all
> DC.
> John
>
John,

I think you'll find that the -12V is used only in "legacy" PC design for
ISA busses, or serial ports (which he probably isn't using).

Evan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCEgqopxCQXwV2bJARAvUpAJ9QJP/GjpzrhRBgI73a8UrVyWBD+QCgsgGj
ONCHJdB56/BtoKbGC18OFS4=
=Lq2H
-----END PGP SIGNATURE-----

Evan Carew

February 15th 05, 08:52 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

John,

I wonder if we are trying too hard on the +12V issue. Couldn't you
simply let it float free with something like a MOV to keep it from
getting too high? If the voltage goes too low, a simple current limiting
sacrificial resistor (AKA fuse) would protect the equipment. You could
even use one of those self healing plastic fuses.

Looking at the design again, there won't be any hard drives or fans to
suck juice from this voltage, so the only thing it will be powering is
the PCI bus.

John wrote:
>
>
>
> Yes I did a little research on it and the only thing the negative is needed
> for is the rs232 serial ports. for this you can get a $1.50 ic to convert
> +12 to -12 V low current.
> That said, what I'd do is get a 1.5 amp 5v TO-3 regulator and a 1 amp TO-220
> regulator along with one of the TO-220 adjustable 1 amp regulators for the
> 3.3 v and run it all off the 12V (highly filtered) input. That said,
> (again) you might have problems with th 12V regulator if the input drops
> below 13.8V! So you might need to use a separate battery for it and diode
> isolate it from the aircrafts electrical systm and have it charged when it
> exceeds the 13.8 volt figure.
> John
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCEmEhpxCQXwV2bJARAvYPAKC5/3ys03DwDrklsBMYLKhzbeeNtgCdFDb/
7wukjwVF2NQz4kVSOqNayms=
=ohFv
-----END PGP SIGNATURE-----

UltraJohn

February 16th 05, 01:50 AM

Evan Carew wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> John,
>
> I wonder if we are trying too hard on the +12V issue. Couldn't you
> simply let it float free with something like a MOV to keep it from
> getting too high? If the voltage goes too low, a simple current limiting
> sacrificial resistor (AKA fuse) would protect the equipment. You could
> even use one of those self healing plastic fuses.
>
> Looking at the design again, there won't be any hard drives or fans to
> suck juice from this voltage, so the only thing it will be powering is
> the PCI bus.
>

Yes it probably wouldn't matter, an MOV wouldn't do much good but you could
still feed a 12V regulator I guess and when the voltage started to go above
that it would regulate and when below it would just slump. I don't think it
would hurt the regulator to take the input too low.
That being said you could build a low differential voltage regulator for the
12V line it's possible to do it with less than 1 volt drop you just can't
do it with the simpler 3 terminal regulators.
The main point is you get rid of the switching regulators and their inherent
noise. It also makes it easier to filter the PC's noise out of the supply.
John
PS I checked out his site and found it quite interesting. I use to program
with hex and then onto assembler when I started out with computers in 1976.
I had a SWTPC 6800 with a whopping ~800KHZ clock! You had to build the whole
thing from a kit soldering on the buss connectors (SS50 bus) ic's
powersupplies etc. It did not have video you had to use an external
terminal although I later build a video board for it and modified the boot
rom's to use it. I started with a Mega memory of 4Kilobytes and eventually
got up to 48Kilobytes. Those were the days!

John

February 16th 05, 01:37 PM

Blueskies wrote:

>
> "Evan Carew" > wrote in message
> ...
>> John,
>>
>> I think you'll find that the -12V is used only in "legacy" PC design for
>> ISA busses, or serial ports (which he probably isn't using).
>>
>> Evan
>
>
> What is the +12 vdc for?
serial ports, fans and disk drives mainly

John

February 16th 05, 02:20 PM

Frank van der Hulst wrote:

> John wrote:
>>>>I think you'll find that the -12V is used only in "legacy" PC design for
>>>>ISA busses, or serial ports (which he probably isn't using).
>>>
>>>What is the +12 vdc for?
>>
>> serial ports, fans and disk drives mainly
>
> And nowadays CPUs.... late last year I upgraded a PC motherboard from
> 800MHz to to 2.4GHz by replacing the motherboard. The new motherboard
> had an extra connector to provide 12V to the CPU 9Celeron, IIRC), so I
> had to replace the power supply as well. I'm guessing that the CPU had
> onchip regulators to provide the other voltages that it required.
>
> Frank
Thats true. my athlon 64 bit has it too. it was originally for the P4 chip.
Although the project that was the focus of this thread delt with a mini-atx
case that does not have this. For that matter he was not using any disk
drives instead he's using an USB thumbdrive. so he probably could get away
with leaving out the 12v. I've never tried it so I don't know. Maybe this
weekend I'll pull out one of my older atx boards and see if I can get it to
boot without the other supplies.
John
PS Don't forget to emulate the pwr good signal!

Blueskies

February 16th 05, 05:51 PM

"Evan Carew" > wrote in message ...
> John,
>
> I think you'll find that the -12V is used only in "legacy" PC design for ISA busses, or serial ports (which he
> probably isn't using).
>
> Evan

What is the +12 vdc for?

Frank van der Hulst

February 16th 05, 06:57 PM

John wrote:
>>>I think you'll find that the -12V is used only in "legacy" PC design for
>>>ISA busses, or serial ports (which he probably isn't using).
>>
>>What is the +12 vdc for?
>
> serial ports, fans and disk drives mainly

And nowadays CPUs.... late last year I upgraded a PC motherboard from
800MHz to to 2.4GHz by replacing the motherboard. The new motherboard
had an extra connector to provide 12V to the CPU 9Celeron, IIRC), so I
had to replace the power supply as well. I'm guessing that the CPU had
onchip regulators to provide the other voltages that it required.

Frank

Evan Carew

February 17th 05, 07:21 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

As a general rule, if you are thinking of solidifying such a design, you
generally go out & purchase one of those hardened PC systems such as
PC104 or like. These systems come with features such as extended
temperature ranges, G tested, & slower clock rates so you don't have to
use fans. Most of the systems I have seen in this class also have slots
for CF cards & can get by on ~15 watts of power, or just about what your
average night light is rated for.

Evan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCFO6ipxCQXwV2bJARAoijAJ9uyFiGJsaghTBrb0hH+X cGVmSncwCfaoSK
0zoNS5oI56uVEBP39MDuFcA=
=KaXv
-----END PGP SIGNATURE-----

jcpearce

February 18th 05, 04:57 PM

Thanks everyone. A brief update.
With the new power supply the mini itx people sent me the problem
almost went away, I was able to do flight tests with it but .... very
rarely it would still cause enough interference to cut into the
Radio's, something clearly unacceptable as sooner or later it would
happen when the tower would be talking to me.

So I suppose I will pursue the linear regulator route. John you
mentioned there was an IC to take 12V and invert it to -12v, do you
know the part #?

Before I read the recent post I had all the TO-3's to make one and was
going to try and just omit the -12V and see what happens as I could not
find out what this was used for, but I see in the posts it is used for
serial communication which I do use. The ADC board I built sends over
the sensors data via serial and I log all the GPS NMEA data via serial.

I like the idea of linear regulators, low noise and a low part count.
There should not be a heat problem as the nose of the Long EZ has
circulating air (which is a bane at high altitudes).

Looking at the ATX power supply specifications they state there is a
sequencing of voltages and when everything is OK, a PW_OK line is
dropped low. I would rather not get into sequencing and keep as much as
possible to the K.I.S.S. mindset. Does anyone have an idea on whether
this is truly important? I was hoping to just apply power and be ready.

Thanks again to everyone and the recent replies from John, Evan &
UltraJohn

Evan Carew wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> As a general rule, if you are thinking of solidifying such a design,
you
> generally go out & purchase one of those hardened PC systems such as
> PC104 or like. These systems come with features such as extended
> temperature ranges, G tested, & slower clock rates so you don't have
to
> use fans. Most of the systems I have seen in this class also have
slots
> for CF cards & can get by on ~15 watts of power, or just about what
your
> average night light is rated for.
>
> Evan
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.0.6 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFCFO6ipxCQXwV2bJARAoijAJ9uyFiGJsaghTBrb0hH+X cGVmSncwCfaoSK
> 0zoNS5oI56uVEBP39MDuFcA=
> =KaXv
> -----END PGP SIGNATURE-----

UltraJohn

February 19th 05, 12:19 AM

jcpearce wrote:

> Thanks everyone. A brief update.
> With the new power supply the mini itx people sent me the problem
> almost went away, I was able to do flight tests with it but .... very
> rarely it would still cause enough interference to cut into the
> Radio's, something clearly unacceptable as sooner or later it would
> happen when the tower would be talking to me.
>
> So I suppose I will pursue the linear regulator route. John you
> mentioned there was an IC to take 12V and invert it to -12v, do you
> know the part #?
>

snip

> Looking at the ATX power supply specifications they state there is a
> sequencing of voltages and when everything is OK, a PW_OK line is
> dropped low. I would rather not get into sequencing and keep as much as
> possible to the K.I.S.S. mindset. Does anyone have an idea on whether
> this is truly important? I was hoping to just apply power and be ready.
>
>
> Thanks again to everyone and the recent replies from John, Evan &
> UltraJohn
>

http://www.jameco.com/Jameco/catalogs/c251/P38.pdf
This is one chip its a higher power version of the ICL7660/7662
TC962CPA
Converts +voltage to -voltage ie +12v to -12 volts or +5 volts to -5 volts.
About $2.50 a piece in single quantities.
They do use switching but with the low power should not be the problem that
the high power switchers in the +5 atx power supplies.
I think they're rated at 80ma 95+% efficiency it should be enough for rs232.
Also check weather your video uses - voltages it probably does. Put an
ampmeter in series with your - powersupplies to see what draw it needs.

As far as pw_ok is you could probably use a generic pnp transistor such as
an 2n2222 with a resistor from the + voltage to the collector, ground the
emitter and then a resistor from the + voltage to the base and a 10uf+
capacitor from the base to ground. connect the collector to the pw_ok. This
would keep the collector (pw_ok) high till the capacitor charged enough the
it would go to ground. You could also use a generic ic inverter to do the
same thing.

Hope this helps.

John

ps via has a board similiar to what your running but it is designed from
scratch to run from just 12v it retails for $183.00.

Blueskies

February 19th 05, 01:22 AM

"jcpearce" > wrote in message ups.com...
> Thank you everyone for the input, now I am looking for some
> prioritization. I received the new power supply from www.mini-box.com
> and it is substantially better than the previous version
> (http://www.mini-box.com/s.nl/sc.8/category.13/it.A/id.356/.f ) however
> I still get some noise in the radios. The new version is too big to
> fit in the previous enclosure so it is sitting slightly apart from the
> MB on its own.
>
> I have started to jot down what the system wants to do, pictures and
> status at
> http://69.229.202.54:8080 Under the 'Image' folder one can see the
> plane, the EPIA in the nose and the prototype ADC board.
>
> So now I have a case that has to heavily part for the ATX power lines
> to come into it & a non cased DC power supply. I have attached ferrite
> beads to the power lines coming in/out with little improvement.
>
> It takes time for many of the steps so I am looking for what you folks
> think would be the 'biggest bang for the buck', or the right priority
> to proceed with. (IE, build a brand new case for everything, build a
> seperate case for the PSU, don't build a case right away but do X,Y,Z,
> etc..)
>
> Thanks once again for the brains helping noodle this over.
>

Check out this machine, only +5vdc required, linux capable...

http://www.diamondsystems.com/products/prometheus

I'm sure there are many more similar...

Evan Carew

February 19th 05, 02:37 AM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Blueskies wrote:
> "jcpearce" > wrote in message ups.com...
>
>>Thank you everyone for the input, now I am looking for some
>>prioritization. I received the new power supply from www.mini-box.com
>>and it is substantially better than the previous version
>>(http://www.mini-box.com/s.nl/sc.8/category.13/it.A/id.356/.f ) however
>>I still get some noise in the radios. The new version is too big to
>>fit in the previous enclosure so it is sitting slightly apart from the
>>MB on its own.
>>
>>I have started to jot down what the system wants to do, pictures and
>>status at
>>http://69.229.202.54:8080 Under the 'Image' folder one can see the
>>plane, the EPIA in the nose and the prototype ADC board.
>>
>>So now I have a case that has to heavily part for the ATX power lines
>>to come into it & a non cased DC power supply. I have attached ferrite
>>beads to the power lines coming in/out with little improvement.
>>
>>It takes time for many of the steps so I am looking for what you folks
>>think would be the 'biggest bang for the buck', or the right priority
>>to proceed with. (IE, build a brand new case for everything, build a
>>seperate case for the PSU, don't build a case right away but do X,Y,Z,
>>etc..)
>>
>>Thanks once again for the brains helping noodle this over.
>>
>
>
> Check out this machine, only +5vdc required, linux capable...
>
> http://www.diamondsystems.com/products/prometheus
>
> I'm sure there are many more similar...
>
>
All Right! Now that's what I call an embedded system. Actually I use
systems almost exactly the same as these in my line of work... Put Linux
on them too.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCFqZ1pxCQXwV2bJARAq0WAKCQoWnkN5Ncza2Q/smCjFLG3+Xs4wCgi6Gx
EoJRK6xJaoaAcln6lG69Hcg=
=8+p8
-----END PGP SIGNATURE-----

UltraJohn

February 19th 05, 03:14 AM

Blueskies wrote:

> Check out this machine, only +5vdc required, linux capable...
>
> http://www.diamondsystems.com/products/prometheus
>
> I'm sure there are many more similar...

Looks to be a nice system to play around with. It has the possibility of
handling both the data interface and computations. The main problem I see
with this system is no windowing system ie KDE and I think that was part of
his need.
John

UltraJohn

February 19th 05, 03:19 AM

> Hope this helps.
>
> John
>
> ps via has a board similiar to what your running but it is designed from
> scratch to run from just 12v it retails for $183.00.

Product info:

http://www.logicsupply.com/product_info.php/cPath/21/products_id/174

VIA EPIA TC10000 Mini-ITX Motherboard (on-board DC-DC converter) $183.00

The ultimate platform for low profile networked devices, integrating an
onboard 12V DC-DC converter and horizontal SODIMM for complete system
design flexibility.

Measuring 17 x 17cm, the ultra compact VIA EPIA TC Mini-ITX Mainboard is a
highly integrated and configurable x86 platform with rich connectivity and
multimedia options that enables the development of a wealth of small, slim
and low noise networked systems. The rich built-in feature set of the VIA
EPIA TC alleviates the need for system developers to source extra
components, reducing cost and time to market.

VIA EPIA TC Specifications
• VIA C3™ processor
• VIA CLE266 North Bridge, VT8235 South Bridge chipset
• 1 x 200-pin DDR266 SODIMM memory socket, up to 1GB
• Integrated VIA UniChrome™ 2D/3D graphics with integrated MPEG-2
accelerator, motion compensation and DuoView support
• 1 PCI Expansion Slot
• 2 x UltraDMA 133/100/66 IDE connectors, direct DOM support (VCC set to
5V/3.3V on the 20th pin)
• 12V, 5V, 60W Max onboard DC-DC Converter
• VIA Networking Tahoe VT6103 Fast Ethernet 10/100 PHY-ceiver onboard LAN
• VIA VT1612A 2-channel AC'97 codec onboard Audio
• Onboard I/O Connectors: 2 USB connectors for 4 additional USBv2.0 ports, 1
Front-panel audio connectors (mic-in and line-out), 1 Serial port header, 1
LPT port-connector, 1 CIR connector (Switchable for KB/MS), Wake-on LAN
connector, 2 Fan connectors CPU/Sys FAN, 1 + 12V power connector
• Back Panel I/O: 1 DC-in jack, 1 PS2 mouse port, 1 PS2 keyboard port, 1
RJ-45 LAN port, 1 Serial port, 2 USB 2.0 ports, 1 VGA port, 3 Audio jacks:
line-out, line-in and mic-in
• Mini-ITX (4 layer) form factor 17 cm x 17 cm (6.7” x 6.7”)

Additional Information:
• This motherboard has an integrated DC-DC power converter wich needs a 12V
power supply. ONLY use a 12V, 5A (60W) power supply with this board. If you
plan on using this motherboard in a car or boat, you will need to use a
power stabilizer.

jcpearce

February 19th 05, 06:12 AM

I built a test linear regulator power supply this evening and the TO-3
regulator bringing ~12v down to 3.3 Volts nearly burnt a hole in the
test board. I looked over the wiring which was OK so it looks like the
thermal output of bringing 12volts down to 3.3 for 3 amps is just way
too high. So so much for that idea (unless I am missing something)

I may just end up getting on the EPIA TC models as UltraJohn mentioned
and hope that their EMI/RFI solution is better. The plane battery
voltage is ~12volts idle and ~14 volts running, what is a way of doing
the "power stabilizing" necessary for the EPIA TC which requires
11.6-12.4 Volts? If I just put a 12V linear regulator on there the drop
off in voltage would prevent the system running off battery.

AINut

February 19th 05, 08:31 AM

How much does one of these development systems cost?

>
> Check out this machine, only +5vdc required, linux capable...
>
> http://www.diamondsystems.com/products/prometheus
>
> I'm sure there are many more similar...
>
>

Evan Carew

February 19th 05, 01:21 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

jcpearce wrote:
> I built a test linear regulator power supply this evening and the TO-3
> regulator bringing ~12v down to 3.3 Volts nearly burnt a hole in the
> test board. I looked over the wiring which was OK so it looks like the
> thermal output of bringing 12volts down to 3.3 for 3 amps is just way
> too high. So so much for that idea (unless I am missing something)
>
> I may just end up getting on the EPIA TC models as UltraJohn mentioned
> and hope that their EMI/RFI solution is better. The plane battery
> voltage is ~12volts idle and ~14 volts running, what is a way of doing
> the "power stabilizing" necessary for the EPIA TC which requires
> 11.6-12.4 Volts? If I just put a 12V linear regulator on there the drop
> off in voltage would prevent the system running off battery.
>
JC,

Look at the specs for the part. They often require X many square inches
of copper, or equavelent in a heat sink to disipate the thermal energy
they throw away.

Evan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCFz1ipxCQXwV2bJARAovDAKDKOkN1Gy5Q/cGya6uLaOOLOtjHKACgqokZ
560wu/RjK/LtNav/eepuZrs=
=scfz
-----END PGP SIGNATURE-----

jcpearce

February 19th 05, 01:22 PM

So far I have spent
$180 MB
$ 80 Power supply
$80 512Mb DDR ram
$90 on two USB flash cards
$30 data aquisition board materials
$30 odds and ends.
~$490
When I bought the EPIA the TC model was not out yet. This is what I
would have purchased these days saving $80 and a lot of headache. I
might end up buying this anyway and selling the old one on ebay.

Evan Carew

February 19th 05, 01:24 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

UJ,

Well, it doesn't come with a video card if that is what you meant,
however, that is a relatively simple matter to fix. PC104 video cards
are quite reasonably priced & even support 1/2 or 1/4 VGA standards.

Evan

UltraJohn wrote:
> Blueskies wrote:
>
>
>
>>Check out this machine, only +5vdc required, linux capable...
>>
>>http://www.diamondsystems.com/products/prometheus
>>
>>I'm sure there are many more similar...
>
>
> Looks to be a nice system to play around with. It has the possibility of
> handling both the data interface and computations. The main problem I see
> with this system is no windowing system ie KDE and I think that was part of
> his need.
> John
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCFz4opxCQXwV2bJARAnAmAJoDMEDGi2BwdSuFbfQo/kYo1o/tGwCfXD6v
P9jP1/H9v/OAWRmw3a++wi4=
=iPrK
-----END PGP SIGNATURE-----

Pete Schaefer

February 19th 05, 05:14 PM

It's a PC. He can install all the Linux junk he wants.

"UltraJohn" > wrote in message
k.net...
> Looks to be a nice system to play around with. It has the possibility of
> handling both the data interface and computations. The main problem I see
> with this system is no windowing system ie KDE and I think that was part
of
> his need.

Pete Schaefer

February 19th 05, 05:16 PM

He could also go with the Athena card instead. That does have VGA on it. And
it's a faster processor.

"Evan Carew" > wrote in message
...
> Well, it doesn't come with a video card if that is what you meant,
> however, that is a relatively simple matter to fix. PC104 video cards
> are quite reasonably priced & even support 1/2 or 1/4 VGA standards.

Pete Schaefer

February 19th 05, 05:17 PM

I bought one a couple of years ago for around $1k (+/- a couple hundred
bucks).

Arcom also sells some very reasonable development kits.

"AINut" > wrote in message
...
>
> How much does one of these development systems cost?

UltraJohn

February 19th 05, 07:11 PM

jcpearce wrote:

> I built a test linear regulator power supply this evening and the TO-3
> regulator bringing ~12v down to 3.3 Volts nearly burnt a hole in the
> test board. I looked over the wiring which was OK so it looks like the
> thermal output of bringing 12volts down to 3.3 for 3 amps is just way
> too high. So so much for that idea (unless I am missing something)
>
> I may just end up getting on the EPIA TC models as UltraJohn mentioned
> and hope that their EMI/RFI solution is better. The plane battery
> voltage is ~12volts idle and ~14 volts running, what is a way of doing
> the "power stabilizing" necessary for the EPIA TC which requires
> 11.6-12.4 Volts? If I just put a 12V linear regulator on there the drop
> off in voltage would prevent the system running off battery.

You could use an external wirewound resistor(around 20 ohm), mount it to
your metal chassis, this should drop it to around 5v at 3 amp draw and will
make your voltage regulator a lot happier!
No matter how you look at it dropping 12+ to 3.3 v you'll have to disipitate
some heat somewhere.
John
Unless you use a switching regulator <vbg>.

UltraJohn

February 19th 05, 07:16 PM

Pete Schaefer wrote:

> He could also go with the Athena card instead. That does have VGA on it.
> And it's a faster processor.
>
> "Evan Carew" > wrote in message
> ...
>> Well, it doesn't come with a video card if that is what you meant,
>> however, that is a relatively simple matter to fix. PC104 video cards
>> are quite reasonably priced & even support 1/2 or 1/4 VGA standards.

Looking closer at it the biggest shortcoming I see now is the lack of ram
the card only handles 32mb and Linux is very memory intensive and running
only a 486dx-100 mhz chip with that kind minimal memory it will be
relatively slow. Do you really want to watch a video screen at 1/2 or 1/4
vga resolution? I think the card would be great for his data acquisition
card and use the via card for the processor. at least he'd be using the
same operating system and compilers for both. Less to learn and get
confused with, I get very confused!
John

Pete Schaefer

February 19th 05, 07:56 PM

Look again at the specs. The Athena has 128mb on board, and uses a
400-600MHz VIA Eden.

"UltraJohn" > wrote in message
nk.net...
> Pete Schaefer wrote:
>
> > He could also go with the Athena card instead. That does have VGA on it.
> > And it's a faster processor.

> Looking closer at it the biggest shortcoming I see now is the lack of ram
> the card only handles 32mb and Linux is very memory intensive and running

UltraJohn

February 19th 05, 11:04 PM

Pete Schaefer wrote:

> Look again at the specs. The Athena has 128mb on board, and uses a
> 400-600MHz VIA Eden.
>
> "UltraJohn" > wrote in message
> nk.net...
>> Pete Schaefer wrote:
>>
>> > He could also go with the Athena card instead. That does have VGA on
>> > it. And it's a faster processor.
>
>> Looking closer at it the biggest shortcoming I see now is the lack of ram
>> the card only handles 32mb and Linux is very memory intensive and running

My reference was to the Prometheus board that was suggested, and it does
look like a viable source for the data acquisition.
http://www.diamondsystems.com/products/prometheus

I have not looked up the Athena. Who makes it? Even the 128mb is pretty
marginal for Linux except possibly in the straight data acquisition mode.
John

Pete Schaefer

February 20th 05, 02:13 AM

"UltraJohn" > wrote in message
ink.net...
> My reference was to the Prometheus board that was suggested, and it does
> look like a viable source for the data acquisition.

Actually, no, you referenced my post about the Athena. Anyway, that's just
a quibble.

> I have not looked up the Athena. Who makes it?

Diamond Systems. Same as the Prometheus. The Prometheus is actually pretty
old. They also make the Hercules, which won't fit in a standard PC/104 stack
chassis, but it has enough I/O on it that you probably wouldn't need to add
much of anything to it.

> Even the 128mb is pretty marginal for Linux except possibly in the
straight data acquisition mode.

Most every display I've ever put or seen in an airplane has way less (by an
order of magnitude) memory that that. If you're hitting the wall for an
airplane display with 128mb, then you really need to rethink about how
you're running it. Maybe a stripped-down kernel just driving OpenGL stuff
makes more sense. A full-up Linux installation doesn't make much sense for
use in an airplane.

Robert Bonomi

February 20th 05, 02:57 AM

In article t>,
UltraJohn > wrote:
>Pete Schaefer wrote:
>
>> He could also go with the Athena card instead. That does have VGA on it.
>> And it's a faster processor.
>>
>> "Evan Carew" > wrote in message
>> ...
>>> Well, it doesn't come with a video card if that is what you meant,
>>> however, that is a relatively simple matter to fix. PC104 video cards
>>> are quite reasonably priced & even support 1/2 or 1/4 VGA standards.
>
>Looking closer at it the biggest shortcoming I see now is the lack of ram
>the card only handles 32mb and Linux is very memory intensive and running

With some care in kernel tuning, you can get the memory footprint *way* down.

The _real_ memory pig is the X server.

With a text-only display, it is *amazing* how much you can do in 'small'
memory configurations. 8 megs is _plenty_ for running a dedicated 'firewall'
box, for example, and a _386DX/33_ will handle a full T-1, in that
application, with cycles to spare.

jcpearce

February 20th 05, 06:34 PM

I ended up ordering a Via TC motherboard as it will take up less space
than the CL with cumbersome DC converter and hopefully reduce RFI, and
if there is still RFI I have only one variable to contend with vs two.

I did not go with the PC104 format although I did look at this at
length. I already have considerable experience with the VIA now, a case
and airplane bay designed for it (17cm x 17cm) and a very lean kernel
tweaked for it. Also the VIA is a tad more mainstream with this
typically resulting in better doc and support for drivers. Also I can
boot off a write protected USB flash card (USB flash memory as IDE did
not work for me, too much data corruption in unexpected power off
scenarios). I don't know which PC104 have this ability. Anyone with
PC104 experience would probably argue with this, but I have had my fair
share in the past of getting odd behavior with a video, sound, ethernet
driver and having to slowly track it down. I know all the MB sensors on
the VIA are queryable in linux (ie CPU temp, all the voltages, etc..)

For me, memory is very important and 512Mb a mininum. Why so big when
my entire custom Linux distribution with X and Java fits on a 128Mb USB
flash card? As a Robert pointed out the kernel is a small memory
consumer, X is the real hog. And with potentialy complicated displays
this would grow even further.

Most importantly this distribution runs in entirely in memory,
there are no writes by the OS to the main USB flash card which is write
protected. As such there is no swap space, and hence any malloc calls
need to actually find real memory not virtual memory. Flash memory
degrades with each write, areas such as swap and log directories can
receive an enormous # of writes with often little effort. Some
benchmarks have shown corruption after only 10,000 writes. Having a
memory block go bad (which is just a matter of time given the nature of
Flash) can lead to odd/unpredictable behavior (the worse kind),
something I want to entirely avoid.

The secondary USB flash card, which is not write protected, holds the
log files and mp3 files. 'Non critical' files. I also wrote the data
logging program to only flush the OS writes every X seconds, not on
each log write. I could add an OS swap file to this partition but I
still could end up with corrupted swap space and the additional writes
would also shorten the life of the USB flash card. Additional memory is
cheap insurance (in additional $40) in avoiding these problems and has
an added benefit of performance.

If I was going to start from scratch I might go with the PC104 platform
but neither platform seems to be vastly superior to the other. And the
Devil you know...

UltraJohn

February 20th 05, 07:32 PM

jcpearce wrote:

> I ended up ordering a Via TC motherboard as it will take up less space
> than the CL with cumbersome DC converter and hopefully reduce RFI, and
> if there is still RFI I have only one variable to contend with vs two.

Nice write up of what your thinking is. As you said probably either system
could work but your more familiar with this avenue.
This has been a very positive thread with no would getting there"panties up
there crack" <g>.
I'd be interested in any updates your might have, I have already bookmarked
your web site and will check in from time to time.
I wanted to check on the cost of the Athena system unfortunately they
require you to log in and register for a quote, I wasn't that
interested<vbg>.
Good Luck
John

Pete Schaefer

February 20th 05, 09:26 PM

I think the Athena goes for around $600.

"UltraJohn" > wrote in message
ink.net...
> I wanted to check on the cost of the Athena system unfortunately they
> require you to log in and register for a quote, I wasn't that
> interested<vbg>.

Evan Carew

February 21st 05, 11:04 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

JC,

For what its worth, the previously described PC104 486 based system was
used by my company to controll a filter system & write to an X display
at the same time running Linux. That system only had 128MB of RAM & that
might have been too much. All applications ran from RAM.

The problem with X apps that run on Linux is that when they are
developed with a toolkit such as Motif or GNOME, you end up including
quite a lot of baggage along with the X app. One way to solve this is to
search for a "light" toolkit or to simply write your app in plain X
yourself. People often say of Such apps "Hey your X app really starts
fast!" That being said, you should not be surprised when X sucks 50% of
the performance of your CPU up on such a box.

On my production unit, the X Application was the only one running &
didn't need a desktop. Under the hood, my drivers supplied the needed
data to the X app. Hopefully, it goes without saying that the kernel was
a custom striped down version with my drivers & very little else.

Evan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFCGmkWpxCQXwV2bJARAgFMAKC9u8jXMDfDqNQ60wh+rx flGNn/XgCfQNeV
0QqkahRlAra3uhd6QPNS5xw=
=xcmq
-----END PGP SIGNATURE-----

UltraJohn

February 22nd 05, 02:07 AM

Evan Carew wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> JC,
>
> For what its worth, the previously described PC104 486 based system was
> used by my company to controll a filter system & write to an X display
> at the same time running Linux. That system only had 128MB of RAM & that
> might have been too much. All applications ran from RAM.

Good points Evan. A lot can be done if you set up a custom system. Up until
I switch to an AMD 64 bit system I ran slackware in which you can select
indiviual packages to install. You could get a very small system if you
wanted. Stick to on editor one window system and a custom kernel and you
can get a rather complete system pretty small.
Take a look at most 'Boxed' systems under /bin /sbin /usr/bin and /usr/sbin
and you will see a lot of programs that you will never use!
John