Simulate RF interference coupling prior to testing

There are so many ways for RF to get into your circuit. You have to gp through each path one by one

mook Jonhon puked:

=================

Only real tests on a prototype have any value.

..... Phil

I doubt that Spice will tell you much.

In my experience, emi sensitivity is much worse in narrow bands where boxes, boards, and cables resonate. That often happens in to 100-300 MHz range.

Well, it can certainly be done; probably at low impedance the magnetic coupling would dominate, and at high impedance capacitive will dominate, but...

your components have resonances that will have to be modeled; how complex a SPICE model do you expect to support?

What is the GHz impedance of an electrolytic capacitor? I don't know, but it's the kind of thing that has to go into your model. Series resonant frequency of inductors is something usually specified, but of other components, you're on your own.

Me, I'd make a paddle out of something conductive or magnetic, and wave it around until I found a position that attenuated the crosstalk, and put a shield there.

================================================

Beware of this trolling, nut case, retard and congenital LIAR.

His idiot, fake handle ought be fair warning.

.... Phil

RF interference is /very/ dependent on the /details/ of the two systems, including their relative orientations and the external surroundings.

A spice simulation could, /at best/, give you information about one point in that design and implementation space.

Where circuits must be immune to RF interference, they techniques are based around presuming worst-case behaviour. The prediction and assessment of worst-case behaviour is a highly specialised skill, and even then it will be supplemented by careful testing.

Summary: I don't think you will be able to achieve your goals on your own.

EMI reduction uses the following strategies;

Your hypothetical model is not good enough. Needs more details.

Can you model a slot in a shield as a wire ? Can you design microstrip or lumped filters then add wave mechanics to get lumped elements on schematic to simulate a simple wire or structure.

Do you know the ESR, inductance or capacitance of wire?**

Do you know Mutual Coupling is just a ratio from C to C or L to L? This is like crosstalk from C to Z or L to Z.

Can you do it on Spice? Not likely. Although I have done it on Falstad's Sim., You may use tools like COMSOL to simulate the physics, or build and test it with a Network Analyzer or Spectrum Analyzer & scope is more likely.

- You can use the model solutions I listed above **** to know what works best.. and estimate values with **

Read the EMC book by Henry Ott as I did over 40 years ago.

It will give you more than my quick summary of my EMI experience.

One problem of interest was in early 80's when Burroughs had computers with large 14" disk drives on the top floors of the biggest Financial Investors. They were so "big" they had a bylaw that no other building in the city could be bigger. They discovered the data would get random data errors and discovered it was from the 1 second rotating RADAR at the international airport 10 miles away which used kW microwave pulses that got into the high impedance Read Data magnetic signals. (Fixed with braid shielding)

You can model all you like and spend $k on it, but the only way to really prove it is to build a prototype, in it's intended enclosure, put it in screened room and measure it. Too many variables to affordably model it, though no doubt some will try...

Chris

Do the opposite: connect a short stub antenna to a signal generator, plop it here and there, and sweep the frequency to find sensitivities. Fix them.

A few ferrite beads made a huge difference here:

They broke up some serious high-Q resonances.

That's a thermocouple input. The competitor's version of this controller could be shut down from clear across the room, with a modest signal generator. That wasn't pupular in an NMR system.

IEEE exposure limits:

Ummm... it's intermodulation productions. Please calibrate your spelling chequer.

You haven't seen strong RF fields until you've had to deal with marine radios. Depending on the size of the vessel and type of service, there's usually only one mast for all the radio antennas, where everything affects everything else. Here's a fairly simple sailboat mast and VHF radio simulation using 4NEC2: (Click on "Download full text PDF" button). Scroll down to the bottom of the PDF and look at Figure 9. There's no sane way to model exposure to some random piece of equipment at some random location in that mess of an antenna pattern.

As for the "rusty bolt" problem, it's quite real. I used to maintain several mountain top commercial 2way radio sites in the Smog Angeles area. Tightening the tower bolts to destroy the iron oxide diode wasn't very effective. Beating on the tower legs with a large hammer seemed more effective.

What really causes intermod on congested radio sites is having some of the transmit power being picked up by a different transmit antenna, going backwards into the RF output power amplifier, mixing with whatever frequency is being transmitted, getting amplified by the same RF output power amplifier, and radiating these intermodulation products on someone's receive frequency. There are ways to prevent this, but the most common is to install a ferrite circulator or isolator in the transmit path, so that nothing goes backwards into the power amplifier. Mountain top sites are crammed with such devices on every transmitter. It is possible to calculate most of the intermodulation frequency combinations: but there always seems to be some frequency that leaks through. Trying to do that with LTSpice doesn't seems like an exercise in futility.

Note: I'm not going to try to decode the original question until I see some numbers from the OP (besides the 1 ohm).

Well, no; I've done that, but rather than fixing 'em, used them to make the test jig (so there wouldn't have to be a dismount of the transducers to get at the few hundreds of contacts of their drive/sense circuits).

The real issue wasn't external RF, but internal (ground loops) and was traceable by hitting the DC/DC converters with freeze mist. Only one of the dozens of converters affected the pickup/ noise (ferrite bead on its power/return cable fixed that).

This assumes one is working on a single installation, not a field-portable or multiple units in varying interference conditions.

Dear Mook,

A capacitor can be added to each wire in your simulation. All caps get the same RF input noise voltage AC signal on one plate. Then common mode rejection can be observed.

Inductive models are not any better than capacitors. The input simulated noise waveform is important. And capacitor values can be wire length dependent.