In electronic design one of the somewhat 'black arts' is known as EMC testing. Talk to any electronic designer and they'll explain the various different types of methods they use to guarantee their devices pass EMC testing and they usually involve sacrificing animals, rabbit feet or lucky sprigs of heather!
EMC stands for ElectroMagnetic Compatibility and any electronic product needs to be designed not to emit electromagnetic interference. In the bad old days we used to have the vacuum cleaner interfere with the televisions and when you turned on your computer the radio stopped working! The various governments don¹t like electronic systems interfering with each other so they decided to create a set of measures to stop this happening.
This is what is well known in the EU as CE marking and in the US as FCC testing. The important points are:
- A device should not radiate interference above a certain level (radiate like a radio transmitter does)
- A device should not conduct interference (down the power cable like the old vacuum cleaners)
- A device should be immune to external interference (i.e. a TV shouldn't look got all funny when the WiFi kicks up!)
So we spend a large amount of time thinking about all the things that can make electronics circuits radiate noise and design power supplies to reduce the amount of noise that goes out the cables.
Slice is no exception to this rule and we have had to pass the device in a special chamber with a very special antenna which sniffs out all the noise we're emitting and make sure we are below the line. Of course what is more interesting to
us recently is that the Raspberry Pi Compute Module customers are also going to have to be able to do this, so we made sure the Compute Module IO board (which is a development device and therefore in some ways doesn¹t require such stringent testing) does indeed pass class B (suitable for use in the home), and also therefore that we are able to pass Slice!
James and I have been working hard on the Slice project and this is obviously another product that requires suitable testing to be done. We wanted to show that the Slice device could pass with or without the aluminium case and therefore
to show that a customer could create a device from the compute module and pass it for use in the home.
Our first testing was done with a very quick sniff to understand how close we were going to be:
This test shows a relatively broad band noise signal which comes between the class B line (the lower one) and the class A line (the upper one). Clearly not a class B pass! The issue here is once you¹ve run a test you have to go and find out where the noise is coming from, this can be done in a number of ways.
Physics says that the right antenna size around these frequencies is something in the range of c / 2f = 3x10^8 / 1.5x10^8 = 2m (this is a half wavelength). So just about the right length for one of the cables (power, hdmi, usb), our next test is to look at the various frequencies and spread, if the noise was from HDMI it would be a spike around the pixel frequency (148MHz) but our noise is much more general, again USB would be something like 480 or 960MHz. So it is relatively safe to assume the noise is coming from the power cable, so I began with measuring the noise from the cable itself
Picture above is our very very expensive oscilloscope with some very very expensive probes that are designed to measure this kind of noise... (Obviously had to get some gratuitous LED action going!)
Each one of those peaks is possibly a point we're interested in...
The pictures above shows me testing the noise on the power cable using an H probe (H is magnetic field). After a comment from James I tried changing the power supply from the one I got off of Ebay to the Pi power supply with a new connector soldered on.
Lo and behold the noise disappeared! We returned to the testing chamber to try again and see what it looks like now
The picture isn¹t a UFO looking like it¹s about to take off but the LEDs on Slice making lots of flashy lighties! The room is a very special box which is effectively a faraday cage (think big metal box to keep the outside noise out and inside noise in)
This image shows the results from testing Slice after changing the power supply, quite clearly we¹re down to three peaks all of which are below the line (the line has margin built in).
The left peak is around 27MHz which is a result of the PWM talking to the LEDs (and the LEDs talking between each other), the HDMI at 148MHz (with the marker on it, that is the pixel frequency) and the Ethernet at 250 Mhz.
All we need to do now is repeat with the new production Slice PCB
Any questions you've got about this 'fun' stuff I'll be happy to answer, it's all kind of a bit of a black art, but I hope you've got a good grasp of the types of issue people have when trying to make real products!