EMI prevention / protection?

On a sunny day (Fri, 30 Oct 2009 07:43:24 -0700 (PDT)) it happened "dave.harper" wrote in :

All of those possibly Use metal housing, Use feed through caps, perhaps some inductors, like in the form of ferrite beads, on all in an out lines, including supply.

10nF is close to a short at 144 MHz, but put some inductance in series and then it really attenuates.

Dave, Are you using an opamp for the Wein oscillator? Does it have a BJT front end? If so you might try switching to a FET input opamp. (Though it never killed the oscillations I made a Wein oscillator that would sometimes develop high frequency 'fuzz' as the oscillator crossed certain voltages. Changing to a FET got rid of the 'fuzz'.)

George.

Oh a little shielding (putting it in a aluminim box) is a good idea too.

Most likely the RF is trashing the front end of the opamp. As George suggests, try a jfet or mosfet opamp.

Layout matters a lot. Multilayer boards, with a solid ground plane and a tight layout, help a lot.

Try adding a resistor, 1K or so, in series with the oscillator output, close to the opamp, then maybe a cap to the ground plane. The output lead can be an antenna feeding RF back into the opamp. Power supply leads can conduct RF into the opamp as well; use ferrite beads and bypass caps.

Or use some harder circuit, like a digital square-wave generator and a passive lowpass filter.

You can't solder to aluminum foil. The best shield is a seamless conductive box with emi filters on all leads that penetrate the walls. You can make fun little boxes by cutting up copperclad FR4 and soldering the seams.

The Bud/Hammond-type diecast aluminum boxes are good if the PCB inside has a good ground plane, hard bolted to the box, and all i/o leads have some sort of filtering or bypassing.

This thing has a thermocouple front-end amp inside an aluminum shield. You can just see the unmasked copper ring on the pcb that the cover sits on.

ftp://jjlarkin.lmi.net/Ferrite.JPG

The ferrite on the input leads bought another 20 dB or so of emi rejection, specifically killing some narrowband resonant responses in the 100-300 MHz range.

John

Hmm I have no idea if that will make any difference. What are you using to control the feedback? Can you use a more robust oscillator (as John suggested) Some type of 'bang-bang' rather than sitting on the 'knife-edge' of oscillation with the Wein bridge. I needed the low harmonic distortion of the Wein bridge... Is that what you need?

George H.

I'm using the wein bridge to generate audio tones that are fed to the transmitter for digital radio communication (basically a homebrew ASK radio modem). I'd be open to using that, but I'm not aware of any radio modems that output square waves.

This oscillator also has 2 digital pots attached: 1 to trim the space tone, and 1 to trim the mark tone. There's a high speed switch between the two pots to rapidly switch between mark and space tones. I could use this same setup with a 555 and trim the resistor to vary the pulse duration, but I'm not sure what impact a square wave would have on transmission, reception, decoding, etc...?

Has anyone heard of a square wave being used as an audio tone for digital radio communication?

Thanks > Hmm I have no idea if that will make any difference. =A0What are you

Hi Dave, I know nothing about ASK radio modem. I think you can get a triangle wave out of a 555 (Or some other charge/ discharge osc.) You can then low pass filter the triangle to get rid of the sharp peaks.. There are also some trianlge wave to sine wave converter circuits that use diodes or transistors to 'smooth' the tiangle wave. If you don't care too much about harmonic content then either approach may work.

George H.

The amplitude-stabilizing part of a Wein bridge oscillator has a diode and capacitor. You've got a little crystal radio there, and it detects the transmitter and turns the oscillator down (WAY down) thinking it has detected the oscillator amplitude. To check this, put a voltmeter on that capacitor while you key the transmitter.

Put a bypass capacitor across the diode (to conduct RF) and shield that sensitive part of the circuit. A ferrite bead in series with the diode won't do any harm, either.

The amplitude-stabilizing part of the wein bridge I constructed uses 2 diodes (opposite directions) to limit amplitude rather than a diode and a cap. The amplitude limiting circuit I'm using is almost identical to the one shown here:

Would a bypass cap across or between the diodes help?

Thanks! Dave

Yup. A cap across anything that even remotely looks like a BE junction can also help.

Beads in front of opamp inputs or transistor bases also help. Must be as close to the device as possible. And don't forget radiation smack-dab into the chip itself.

More current -> better.

Altoid cans work quite nicely. If the OP is in Europe (can't read his posts because he uses Google) then Fisherman's Friends works well. But don't let one of those mints sit in your mouth and fall asleep :-)

[...]

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Sounds like a typical "audio rectification" problem, in which the input transistor PN junction acts as an AM detector (crystal detector). The other halfwave is removed and there is an biased audio waveform depending on the RF amplitude and the audio can be heard with headphones at the audio output.

However, if this a constant amplitude FM transmission, no audio will be heard, but the rectified DC signal will upset the stage bias, perhaps with several volts.

If this is a non-surface mount opamp, just insert small ferrite beads on the input pins, before soldering it to the board.

Paul

beads,

then it really attenuates.

...or resonates if chosen "wisely"...best add some R for real damnping.

(I was thinking of a FET/voltage-controlled-resistor stabilizer)

Yes, the addition of 200 pF across the diodes might do some good; RF reaching those diodes will modulate their ON threshold, which will lower their impedance, and that's what turns the oscillator amplitude down. The sensitivity of this kind of network is lower than the one I was thinking of, though.

Tin-plated soft steel (like tin cans) is an excellent shielding material; I think it's time to snip open some of the recycle-bin chili containers and apply shielding.

Only as a last resort, surely! The inputs and transistor bases are relatively HIGH impedance, and beads are low-Z by comparison; I'd save those beads for emitters where they have better matching... there has to be RF current into clamp diodes before an op amp input will benefit much from a bead.

Hey OM:

ASK is CW pure and simple. What you have is infinitely variable speed CW. And some times the speed varies so much you can't tell if it's a dit or a dah.

You can't beat CW, no way, no how, for the best source, of signal, in an EMI environment. Towit you can't beat ASK either.

Just keep on doing what you are doing.

As for square wave for audio, we are looking at an infinite number of harmonics in a square wave, and that may be a good thing, but is it what you want?

73 OM de n8zu

e

If you don't mind the size, I've seen circuits built on the top lids of metal one gallon paint cans.

George H.

,

I used a light bulb. Lots of circuits on the web, I started with one by Jim Williams.

George H.

How about a cap between the + and - inputs of the op amp.? A small one to short it at RF, but not effect the circuit where it operated.

I think you mean "AFSK". At least that's what I think you mean. ASK =3D Amplitude Shift Keying.

Something in the transmitter chain stops (or should stop!) the square wave from being square with all the odd harmonics going out to infinity.

This something may be in the audio stage (example: low pass audio filter), or the IF stage (example: bandpass filter), or in the RF stage (example: tuned circuits).

It is very doubtful that after going through the transmitter and receiver that it'll come out as a square wave on the other end. Choosing to transmit a square wave audio waveform is usually a poor choice because you know that it can't come out looking that pretty on the other end. This sort of design decision might be made for a very low-end radio control transmitter of the 60's or 70's out of ignorance, but today we know how to do far better with little extra effort.

Most designs make a conscious choice to be a friendly transmitter, and limit splatter and unnecessary bandwidth that would be in violation of the FCC rules, by running any square wave through a low pass audio filter AND additionally using a rational choice for the IF filtering too. Way better than nothing, is a simple RC low pass in the audio stage. Still to be nice the resulting audio level has to be carefully set to not cause splatter in subsequent stages.

Tim N3QE