Experimental aircraft EFIS system using EPIA M

Some other things to remember:

EMI is black magic. It's based on sound scientific principals but there's so many variables that when you actually start attempting to fix things it's black magic. So don't feel bad if you can't make it work, or if the solution makes no sense.

If the EMI is happening at a specific frequency you can go looking for a circuit that works right at that frequency. I once worked for a month or so revamping software to work on different crystal frequency because with the old frequency the customers favorite tactical comms frequency was upset, even though the equipment squeaked by the EMI standards. For example if you have significant radiation at 120MHz look for a 10 or

The ideal EMI solution is to keep it from getting off the board in the first place. The second-best is to get all the beads in place _inside_ the case, then if that isn't enough make the case EMI proof to keep everything inside -- but this is much easier said than done. If you do address the case pay attention to Joerg's comments about finishes and use _lots_ of screws to attach the lid to the case.

motherboard

the

A good chunk of the interference is conducted out via the power supply wires, ethernet, video wires, etc.

To some extent you can put chokes on these leads but seeing as how the signal bandwidths over the Ethernet and video are in the tens of MHz you can't filter out the actual signal!

If you can, for example, use optical interfaces to carry Ethernet and other necessary signals from your box then you can come out way ahead. And using not-el-cheapo power supplies helps a lot with conducted interference over the power leads.

Are you using 400Hz 120VAC power, or 12VDC, or something else to power the EPIA? If it's 12VDC via one of those little 12VATX-like converter boards, keep in mind that those converter boards use almost completely unfiltered SEPIC and step-down converters and put out incredible hash even without a computer hooked to them.

Tim.

Quick thought: 100/120/133MHz front-side bus?

I am not familiar with your equipment (I'm a hang-gliding guy myself) but perhaps you could try "wrapping" the suspected emitter in grounded metal window screening (faraday cage) and running the test again. This is just to pinpoint/determine if the EMI can be shielded without building custom enclosures.

That said, any chance a notch filter could be installed in the receiving equipment? Better to fix the source of the problem rather than band-aid everything else though of course.

Hello jcpearce,

I believe the Long EZ is composite which would make it a bit more tough. The case is already a good start. Ideally it should not be anodized aluminum because of a sub-optimal surface conductivity. Maybe you could find alodine or something more conductive. I prefer conductively galvanized steel. But right now I wouldn't worry about it and instead try these measures first:

More important that the enclosure are the wires going in and out. Since you usually don't know all of the exact data rates or impedances, capacitive filtering is not so easy. So buy yourself a large bag of 43 ferrite material. Toroids, all kind of sizes. Ideally you should also buy some clip-ons and some of the Radio Shacks still have these. Then while running the radios and the EPIA keep adding toroids, as close to your enclosure as you can and see what happens. They have to go over the whole cable, including (and most of all) the power cable. If you slid one over +12V alone the DC current might saturate the core and then it does no good.

Please mind that this may void the airworthiness of the Long EZ. Especially if you use clip-on toroids these can fall off in flight and land in places where they might cause a lot of grief. Also, ferrite material is brittle and vibration against something else or a shock can shatter that stuff.

Another product that can come to the rescue if you have D-Sub connectors on there: These are also available as "EMI fortified" where there is a hole-pattern ferrite integrated over the inside pin contacts. See if they are protected against vibration though, for the same reason as mentioned above.

Your situation is better then the EMI job I had to do on a new composite design. The engine magnetos and a pump controller were messing up NAV/COM. So I had to crawl around while the prop and other stuff was running.

Regards, Joerg

Apart what has been told already, there are clock generators that sweep and thus distribute the noise over bands instaed of emitting at selected frequencies. Some motherboards, such as my new Athlon64bit let you select the clock spread option.

I you don't need the computing power an EPIA M provides, something simpler, such as an 8051 or an AVR, would generate less interferences.

Rene

I have a couple of home-built instruments who's cases are double-sided PC boards soldered together. I covered them all over with clear 2" wide shipping tape. The copper has turned green underneath the tape over the years, but it's a consistent, even green -- the only fingerprints are on the back panels which are (a) removable and (b) not taped.

Clear paint would probably also work, assuming you're careful about not painting the parts that are supposed to connect.

Others have covered the major factors to look at. Basically any wire coming out of the case is suspect. You could use feed thru caps to pass the power cables out of the case. I'm not sure how that would work on the video signals though, it might seriously degrade them. And that brings us to the point of my post, what are you using for a display? In my experience, displays and keyboards are notorious for emitting rf noise. Your EPIA itself may not be the real problem, I suspect it's probably not.

Hi Tim,

That is an excellent hint. Especially when considering that aircraft communications are still done in old-fashioned AM which is prone to interference.

I wonder whether the task at hand could be performed using a more quiet laptop. The Linux requirement might be a challenge here because all those nice 'plug and play' USB data acquisition pods are geared towards Win 2000/XP. Some even won't work under older NT variants. However, the OP mentioned an 8051 add-on and those usually come with RS232. That should help, except with some new laptops. The one I just bought is all USB, no other comm ports.

Regards, Joerg

Hello Anthony,

Oh yes. I remember when I did an EMI case in a screen room and part of the noise pattern just didn't make sense, didn't compute. Then it turned out it came from the flat panel display of yours truly, the analyzer.

Sometimes holding a hand in front can pinpoint that. But in a composite structure aircraft that might not mean much because touching anything conductive will likely alter the noise pattern.

Regards, Joerg

In article , Tim Wescott wrote: [...]

Almost :>

The ideal is not to make it on the board in the first place. Second is trapping it in only part of the PCB and keeping it from getting off the board only comes in third.

Source termination of things like clock signals can help a lot. On switcher supplies, softening up the switching will cost you in the losses but perhaps save you on the EMI filter.

Putting a shield around the offending portion of the PCB can help a lot. The smaller the volume you put inside the shield the more effective it tends to be so don't enclose unrelated stuff if you are doing this.

place _inside_

This works fairly well. Using inductors of higher values is usually less effective. You want the lossiness of the beads to suck up the RF power. If you don't, it will just find a new way to leak out. This is usually capacitively right past the inductors.

Or (1) make the box from brass and solder it (2) Use those RF fingers to connect all of the joint (3) buy some sticky backed copper tape and tape over the joints.

Hi Ken,

(1) Brass works but gets ugly over time. Fingerprints, oils, gunk all leave pretty permanent 'souvenirs'.

(2) I prefer not to use any copper-beryllium. I believe its oxide can be toxic.

(3) Copper tape works great but unless you spot solder it tends to come off during a hot summer. Then it sails into some electrical area and, bzzzzt - shazam, you have a whole other issue at hand.

Regards, Joerg

Something I have done to quiet down noisy PCs:

Obtain a large square (as large as will fit) of MosFoam. Digikey sells it, this is the stuff that you put chips in to protect them from ESD in storage.

Cut to fit, and place inside a large freezer bag, and glue the freezer bag with mosfoam in place in the case. The bag prevents any "crumbs" from getting into the other goodies, and keeps the square itself out of anything else.

RF inside a metal box, is like light in a box made of mirrors, it's gonig to bounce around till it dissipates, or finds a way out. The mosfoam is a nice material for it to dissipate in.

How about getting it gold plated?

Must resist temptation to -- oh to hell with it:

Why would you want to gold plate paint?

'Cause gold plating things is good!

But seriously, plating an aluminum case is a fair option. Nickel plating looks quite good and resists finger prints and scuff marks well.

Hello Tim,

Maybe because it'll give that bling bling appeal?

Regards, Joerg

Thanks

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,

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

Anything that's called "DC/DC" by an honest person is going to be switching -- if it's a linear supply it'll be called "linear".

You could always build a linear supply for the thing, if you think you can stand your 12V not being quite 12V sometimes. Plan on _big_ heat sinks, though.

I read in sci.electronics.design that jcpearce wrote (in ) about 'Experimental aircraft EFIS system using EPIA M', on Sat, 29 Jan 2005:

It's really difficult to convert DC voltages without switching! But in your case, you can tolerate the lower efficiency of a sine-wave supply. You need a sine-wave oscillator (not low distortion), an audio amplifier chip and a suitable transformer at its output, which you will need to design or have designed for you. I suggest you use a 400 Hz oscillator, not 50 or 60 Hz. This should not present a problem with core loss if you use a toroidal core, and a core for 400 Hz is much smaller and lighter than one for 50 or 60 Hz. Since the voltages are low, hand-winding is practicable.

You might be able to use just one rectifier to get 12 V and then regulate down to 5 V and 3.3 V, depending on what currents you need. But it isn't that difficult to provide more than one secondary winding to get different DC voltages with less heat loss.