0
\$\begingroup\$

I'm working on solving an issue in a metal detector that generates false detects when subject to electrical interference. The electrical interference usually stems from devices such as shortwave radio (walkie talkies) when being used in proximity, HF fluorescent lighting, Variable Frequency Drives etc.

I want to design/buy a system that can produce these frequencies with low energy. The idea being that I can repeatedly subject the metal detector to noise to understand the behaviour of the metal detector and in turn design a solution that mitigates the effects of the environmental noise.

  • Could I achieve this with a software defined radio?
  • If not, could I achieve what I want with a signal generator, a power amplifier and an antenna?

I understand that there may be concerns about intentionally generating radio frequencies but I'll ask that in an other question.

\$\endgroup\$
1
  • \$\begingroup\$ Probably the best solution is to buy an "intentional radiator" for the given frequency. That is, a device legally designed to send at that frequency - preferably something with continuous transmission. Maybe a walkie-talkie where you use tape to keep a button pressed down or such. \$\endgroup\$
    – Lundin
    Mar 3 at 9:45
3
\$\begingroup\$

Could I achieve this with a software defined radio?

You can produce any band-limited signal, given its bandwidth is within the capabilities of the SDR. So, yes.

You'll find that "shortwave radios used in proximity" is often incredibly high power density compared to what modern SDRs do, so not sure whether you can induce the same problems when doing that at a lower power. But that's a different question!

\$\endgroup\$
2
  • \$\begingroup\$ I was hoping that I could emulate the power issue simply by just getting closer to the metal detector. I also assume that emitting high power radio raves is a definite no-no. Thanks for the advice. \$\endgroup\$
    – ChrisD91
    Mar 3 at 9:44
  • \$\begingroup\$ hm, problem really is that you are trying to find a technical model what of the other devices' emissions are causing trouble – by emulating them, which already assumes a model. If you know your walkie-talkie has out-of-band emissions at some frequency, you don't have to simulate that, you add an appropriate filter. If you don't know where exactly it has emissions, how are you going to simulate a similar effect? \$\endgroup\$ Mar 3 at 10:04
1
\$\begingroup\$

I had the same problem with a Motorola walkie talkie that operated at around 500 MHz. I just keyed the walkie talkie now and then and tried "this and that" on and around the search head coils and front-end circuits until the problem was solved without detriment to the performance of the metal detector.

Usually you'll find that there is one weak spot in the design and once you hang capacitors to ground (for instance) on that vulnerable EMI point, you solve the problem for other frequencies as well.

could I achieve what I want with a signal generator, a power amplifier and an antenna?

Cut-out the "middle man" and use the thing that generates the problem i.e. use the walkie talkie to find a solution. Using the walkie talkie means you can see how robust your potential solution is at different distances just by moving nearer or further away. That's what I did anyway and it worked.

\$\endgroup\$
4
  • \$\begingroup\$ The only issue with your suggestion of "cutting out the middle-man" is that it is impossible to test every device that causes an issue. I can get some walkie talkies but I was hoping that I could simply use a software solution and go through a range of bands that may produce a problem. I'll take your advice on board though, thanks \$\endgroup\$
    – ChrisD91
    Mar 3 at 9:42
  • \$\begingroup\$ I've had problems with Motorola too - not EMC problems but worse, non-compliant radio interference. To the point where we had to enlist a test house to prove that it wasn't our equipment which was non-compliant, but the Motorola walkie-talkies. They aren't good at following the various narrowband requirements for sub-GHz short-range devices, the bare minimum being the EN 300 220 for European use. \$\endgroup\$
    – Lundin
    Mar 3 at 9:49
  • \$\begingroup\$ So step 1 might be to check if the walkie-talkie is at all legal. Hook it up to a spectrum analyser, check occupied bandwidth by carrier, then harmonics. (Max hold setting over time) \$\endgroup\$
    – Lundin
    Mar 3 at 9:51
  • \$\begingroup\$ @ChrisD91 there is usually one dominant weak spot and anything that can cause that weak spot to be exposed is as good as everything that can cause that weak spot to be exposed. Metal detectors are usually highly tuned and that weak spot is usually nothing to do with the coil tuning hence, if there is a weakness, it's usually wideband in nature so, anything to use to expose it is usually good enough to represent all the things than can expose it. \$\endgroup\$
    – Andy aka
    Mar 3 at 10:10
0
\$\begingroup\$

Actually for a budget option, you could totally buy two of the 433MHz "Ali Baba quality" radio modules that are popular with the hobbyists (Google "433MHz duplex radio module") - we always get questions here about how to get those working. Well - the aim here is the opposite, get it not working at all, which is something these cheap modules excel at!

You'll want a semi-duplex pair since those as a rule of thumb toggle their power amps in evil ways. These are excellent noise generators! Tune it up to max power, set it to OOK or some madness modulation like that, maybe also use an antenna with gain. Then watch it spew wide band noise all over ~400MHz plus harmonics. Most definitely the cheapest noise generator on the market and roughly the same frequencies as used by walkie-talkies.

Are they legal? Not at all, but that's Ali's problem since he's the one who put it on the market.

\$\endgroup\$
2
  • \$\begingroup\$ Something like this is exactly what I had in mind. Could I put the signal through an op amp with a wide enough bandwidth to amplify the power? \$\endgroup\$
    – ChrisD91
    Mar 3 at 10:48
  • \$\begingroup\$ @ChrisD91 Yeah sure, should work fine. I don't expect these modules to have much in the way of band pass filters. \$\endgroup\$
    – Lundin
    Mar 3 at 10:53
0
\$\begingroup\$

You basically want an EMI test lab, where they will put your device into a shielded room and just throw a sine wave with rising frequency at it until your device signals an error. For example when testing a computer system, you'd have it blink a LED, and the test runs until the LED stops blinking.

The frequency where the error occurred is likely to correspond to the geometry of the unintentional antenna, so there is a good chance that they can tell you "the problem is four centimeters long", and also try various mitigation techniques like shielding certain parts and testing if that solves the problem.

Basically, this is already available as a service from people who have a lot of experience zeroing in on the problem.

\$\endgroup\$
3
  • \$\begingroup\$ In the end, yes you'll want a test lab. But not while you are still trouble-shooting the issue. Also I'm not sure if I'd call the average test house experienced in zeroing in anything, they are very good at finding a problem, not the problem. Typically you get something like "Your DuT goes nuts at 1.2GHz 30V/m radiated susceptibility". Okay... thanks... but why. They can't help you there; they will just tell you to shower the thing in ferrite beads and shielding. \$\endgroup\$
    – Lundin
    Mar 3 at 12:08
  • \$\begingroup\$ The frequency information is the most important information here. With "1.2 GHz", I know I'm looking for something that is 12.5, 6.25 or so cm long. \$\endgroup\$ Mar 3 at 12:33
  • \$\begingroup\$ Only helpful if you are looking for openings in some metal enclosure or whatever, not in addressing the cause. What good is the value 12.5cm to me if the problem is for example an incorrectly specified SAW band pass filter? \$\endgroup\$
    – Lundin
    Mar 3 at 13:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.