Electromagnetic interference mapping | Hacker News Reader

Electromagnetic interference mapping

charleslabs.fr

142 pointspizza25 comments

This is the most awesome thing I have seen on HN for weeks.

It marries such distinct component disciplines together to come up with a simple, immediately graspable device.

Being able to see radio waves has long been a wish of mine - we walk through this invisible world suffused around us, and anything that steps towards revealing that world I am a fan of.

Agreed! And every component is highly complex, i.e. the design of the probe [1], or object tracking in videos. While the author is obviously standing on the shoulders of giants, he appears to be pretty tall himself.

[1] https://interferencetechnology.com/diy-near-field-probes-pre...

This is excellent! EMI testing is one of the things I have to do with my projects, but there hasn't been an affordable way to visualize them. I've played with an RTL-SDR for ADS-B traffic so this is right up my alley. I'll be sharing this with coworkers for sure!

It's worth noting that his bandwidth is only 2MHz on his SDR so this will require a lot of manual testing and tweaking for good EMI/EMC results. It might be useful for crosstalk but a directional probe would be better for that because you can account for trace alignment. A directional probe can be made in much the same way as his loop probe, but will be a mono-pole design and will probably require an amplifier.

Got any resources on EMI testing on the cheap? Would be a great thing to share in our makerspace, as many do semi-professional electronics.

If you have a decent oscilloscope, you can actually get pretty far with that and some DIY probes.

This is pretty neat! The next step: create an array of common-clocked RTL-SDRs connected to patch antennas, and mount them on a rail. Use beamforming techniques so you don't have to manually wand the device. Connect an antenna to a signal generator to roughly calibrate your array response. Slightly more expensive (the big cost would be a signal generator, but a lab should have one of these anyway), but a lot more potential for automation.

I actually built something similar to the above setup for work, years ago. RTL-SDR wasn't available back then, so it was a lot more expensive. It would be awesome to go back and do it over again, just to see how much it would cost today.

I'm doing EMC work at the moment. I've fantasized about putting a near-field probe on a 2-axis motion stage and building a 4-dimensional (X, Y, Frequency, Power) visualization with the data.

The place I work is pretty well outfitted, so I imagined it using a spectrum analyzer, a couple of Zaber motion stages we have sitting around, and a nice set of H- and E-field probes. Probably several thousand dollars worth of equipment.

Now, reading Charles' blog post, I'm blown away by how he's basically done the same thing in a very clever way using about $50 worth of parts.

I remember doing something like this manually with a C-64 and oscilloscope in my high school physics lab.

Can you elaborate? Is this a project a freshmen can tackle?

Get a loop of wire, hook it up to the oscilloscope, read the amplitude of the oscilloscope.

Move the loop and repeat.

I did this in the late 1980s in my high school physics lab. We had a bunch of tube-type Textronix oscilloscopes that we inherited from the Raytheon factory where they made Patriot missiles. When they broke we shuffled parts around between them to keep them working, the way my physics teacher did when he was in the air force.

The preamp stage of the oscilloscopes were good enough that we didn't need any extra electronics to do the work.

While it's a nice hack, the problem with the EM certification world (and why it's so expensive) is not about the devices themselves but about calibration and measurement stability.

Most EE will be able to guess the patterns this setup generates just by looking at the boards themselves.

This is useful when you fail your EMC testing. You have a baseline measurement and you're looking at relative change. For example you may have be over the limit by 6dB at 430 MHz. You would take a reading using this setup, and normalize it against your lab plot.

You then try a number of fixes, and take readings using this setup. It is easy to determine if you're close to 6dB drop that you need.

The problem is that even with a professional setup you should expect about +/-3dB tolerances.

Since there’s no expectation of that uncertainty being systematic, taking a relative change pushes your error bars +3dB.

That 6dB drop is under the error bars, so it’s pretty meaningless, and that’s with a professional setup.

In other words, there’s no expectation that the numbers you’ll get from these will translate into anything useful, at most you have a ‘hotspot’ detector.

Professional setup has much tighter tolerance than 3dB. I was using 6dB as an example. Usually you're trying to kill peaks that are 10-20dB over.

This kind of debugging is very routine. You do it manually with E/H field probes and a spectrum analyzer in between lab visits. This setup makes it more repeatable by automating it.

>Professional setup has much tighter tolerance than 3dB.

To comply with CISPR (as an example), a lab only has to validate <4.5dB (in the 30 MHz to 300 MHz range), and up to 5.2 dB (from 30 MHz to 1 GHz). See for instance [1].

By the way, if you are doing this kind of testing routinely, you need to check your PCB layout process. This fine grained tweeting should be a last resort option.

[1]http://ing.univaq.it/emc-chap-it/download/MUinEMC_Revision%2...

Yes please!

Finding the problem and fixing it is voodoo (including magical copper tape and metal charms for the wires..), any help is extremely welcome and can save so much money.

(Even better if it improves upon the usual EMV divining rod.)

Do you not find problems in simulation? I used to do modeling in Ansys and CST Microwave Studio, and found them to be fairly accurate.

Mhm.. I think you overestimate the level of this operation. Imagine a hackspace trying to get things aircraft-certified.

(It's a university, which just means we have more expensive things. There is no dedicated person for this, most of the work is done by students with limited time and no domain knowledge in anything hardware at the beginning. And we're still chronically underfunded, of course.)

(So yes, we do model, but usually the science side of it. Most of the instruments stay in the lab, go in cars or just outside worldwide (also ships, stratospheric balloons etc, nobody cares about EM radiation too much as long as you don't overdo it), so the hacky homegrown versions are totally fine. Building scientific instruments in a research group has some advantages because they can be highly custumized, but that means you need to have a person with it who knows which parts to kick for which problems, or where to add or remove some parts. The first iterations are eg expensive optic components, some basic micro-controller or embedded computer system, cardboard and tape.)

(For the more sophisticated things like satellite instrument design you do get dedicated more specialized staff on your team (probably working at ESA or NASA etc), so they will do all the electronics design and probably modelling, so you're free to do what is actually your job, optimizing the thing for science.)

I don't think this is aimed at people doing certification though, it seems like it's more for the prototyping stage.

It doesn't matter if the device isn't fully calibrated in that case.

Yes, I got that, but if you are closing a design loop around a noisy, bad aimed target, you'll get worse results.

I don't understand the need to build a video image out of the information.

Why not link the probe to a buzzer/led, so it tells you when the interference is high as you sweep the probe across the device?

It's fun, it's visual, it's a whole new dimension!

What's the big antenna for?

Url changed from https://www.rtl-sdr.com/using-an-rtl-sdr-and-opencv-to-creat..., which points to this.

Reply on news.ycombinator.com