Actions
Forum Thread Details
  • Replies 15 replies
  • Subscribers 123 subscribers
  • Views 1639 views
  • Users 0 members are here
Related

The magic of Tin cans!

battlecoder

I've been learning more about electromagnetic interference for a while now, and in the process improving my tools and my understanding on how to measure, counter and prevent EMI, noise, crosstalk, etc. And this weekend I had the opportunity to put some of that knowledge in practice (although admittedly the most basic things).

I had this small switching module that takes 2.8-5.5V as input, and outputs ±12V, which is very useful for op-amps and other circuits, and it was supposed to be used connected to another module; a low-noise, 10x gain differential amplifier.

There were two problems with the board. The first one was that the voltage output itself was a bit noisy. This was to be expected since it is a switching supply with not a lot of filtering at the output stage. I fixed this with 79XX and 78XX regulators and a couple of capacitors.

But the other problem is that the module itself emits a crazy amount of EMI, which was very undesirable, given that this module was going to sit very close to a low-noise amplifier board.

To solve that problem I tried putting the board inside a small tin box, and here's the result:

I already knew that "tin cans" (or generally speaking; metal shielding) was an effective way of blocking EMI, but I wasn't expecting THIS much of a difference with a repurposed "hand cream sample" tin can Stuck out tongue

I've used "buck-boost" converters and other switching supplies modules in the past. To the point that they have become an essential "tool" to power electronics in my projects, especially when there's a very specific voltage requirement that can't be easily met with a battery pack or power adapter, or when using a linear regulator would dissipate too much heat, or won't be able to take advantage of the full voltage range of the batteries.

But I wasn't aware of how much EMI they can emit. I'm sure as hell going to be placing them all inside metal enclosures if they are going to be sitting next to noise-sensitive electronics or sensors, from now on.

As I always wonder about other people's experiences, I'd love to ask the community if you have ever needed to put a circuit inside a metal-shielded box in order to fix an issue caused by EMI (or to prevent it from ever happening) ?

Parents

  • I've used actual tin cans for temporary screening in blogs here on e14. Here's one where I was trying mainly to keep out mains noise, but also a low-level nuisance signal (a sawtooth between 40 and 50kHz) that comes off the antistatic mat. Connecting the tin to the circuit 0V and pushing the board to the far end away from the aperture worked nicely.

    /members-area/personalblogs/b/blog/posts/noise-transistor-preamp-3

    image


    In circuit design terms, I mostly worked in areas where the challenge was doing what you're doing here - keeping the radiation in, rather than shielding sensitive circuitry. No cans on boards, though. Just sealing metal enclosures and absorbing radiation on cables with ferrites, when the regs first came in, and then gradually improving things on the design side with each iteration (at that point I wasn't directly responsible for the emc side of things, I was doing design work, but I obviously had to work closely with the engineer doing the emc work).

    Presumably we're looking at the ringing at the switching node of your converter. Is the output unloaded (it looks like it might be cycle skipping)?

  • in reply to jc2048

    That's a nice project! I love how you did indeed use an actual tin can :D

    And you are correct, the output is unloaded in the video. I did run some tests with the circuit loaded, though with pretty similar results (the intended load is very small, like ~30mA, so it doesn't make a *huge* difference)

Reply
  • in reply to jc2048

    That's a nice project! I love how you did indeed use an actual tin can :D

    And you are correct, the output is unloaded in the video. I did run some tests with the circuit loaded, though with pretty similar results (the intended load is very small, like ~30mA, so it doesn't make a *huge* difference)

Children
No Data