Voting resources, early voting, and poll worker information - VOTE. ... Adafruit is open and shipping.
0

Avoiding electrically noisy connections?
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

Avoiding electrically noisy connections?

by pneumatic on Sun Jul 19, 2020 2:40 am

I'm trying to use some microphones to detect loud sounds, and the problem I've been having is that I can't seem to get an electrically quiet connection.

I'm using a Feather M0 WiFi as my microcontroller and a MAX4466 amp with an electret mic feeding into A2. The connections on the protoboard wing look like:
Image

The white wire is the output of the mic, but the wire is too fine to make a good connection with the socket, so I've tried to beef it up a little by adding solder:
Image

But I still wasn't getting a good connection, so I soldered it to a single pin I cut off a header:
Image
(I know.. I should throw some heat shrink on that, but I'm still not getting a good connection.)

Any ideas how I should approach this?

pneumatic
 
Posts: 176
Joined: Sun Jul 26, 2009 3:59 pm
Location: Newark, CA

Re: Avoiding electrically noisy connections?

by adafruit_support_mike on Sun Jul 19, 2020 5:51 am

What kind of noise are you getting? There's a whole menagerie of different noise sources, each with its own way of handling.

Noise from mechanical issues is almost entirely make-and-break electrical contact between two conductors, and is easy to trigger by moving the conductors relative to each other. If that's the case, the general solution is to ensure better contact between the conductors. Instead of using pin header, cut a JST-PH2 jumper in half, solder one half into your protoboard and the other half to your MAX4466, and use the connectors to make and break the connection between the two:

https://www.adafruit.com/product/1131

If the noise seems to be independent of the physical connections, the next most likely suspect is ground bounce. All real conductors have some resistance, and sending current through them produces voltage. When the current changes suddenly (digital circuits are notorious for that), you get sudden changes in the voltage. If you have current from multiple sources passing through the same physical conductor, voltage spikes from any source will be visible to every source connected to the conductor.

The solution there is to avoid shared wiring as much as physically possible. Ideally, the current paths for different circuits should only meet at the terminals of the power supply.. and then you put a big capacitor between the supply's terminals to lower its AC impedance.

The next suspect after that is capacitive coupling. Any two conductors near each other form a capacitor whose impedance is inversely proportional to frequency. A signal whose voltage changes quickly (another realm of notoriety for digital circuits) can jump through the air gap between a pair of wires surprisingly well.

For that, there's an old saying: physicists solve coupling problems with distance; engineers solve coupling problems with shielding. Start by moving the conductors as far apart as possible, and making the cross sections that remain parallel to each other as small as possible.. if two wires have to cross, do it at 90 degrees. Real projects have space constraints though, and sometimes those don't allow you enough room to get suitable distance between conductors. When spacing things out as far as possible isn't good enough, then you start to add shielding.

A shield is a new conductor with a low-impedance connection to some fixed voltage (usually GND) that sits between two other conductors. It forms two new parasitic capacitors, one with each of the original wires. The distance from the shield to either original conductor is smaller than the distance bewteen the original pair, so the new capacitors are better at coupling signals than the one formed by the original pair. The shield's low impedance connection to a fixed voltage makes it hard for coupled signals to change the shield's voltage, so any coupled signal ends up being a small burst of current across a low-value resistor. The amount of signal that can pass through the shield from one original conductor to the other gets chopped down dramatically.

Inductive coupling behaves in a similar way, but tends to be more of a problem with circuits that carry a lot of current. The solution there is to reduce the enclosed area of any current loop as much as possible (run feed and return lines right next to each other so you have equal currents flowing in opposite directions) and to use twisted pair cable (two interlocked helices have exactly the same average path). And when those aren't enough, you add shielding that creates two new parasitic inductances to a fixed voltage again.

If none of those things seem to help, the noise is probably mixed in with the signal of interest. Then you need to start tuning the circuit's frequency response. That's basically a matter of adding high-pass filters (capacitors) between the conductor that carries the signal and GND. When using an ADC, the goal is to block any signal faster than half the ADC's sampling rate. The ADC can't process those signals correctly anyway (it takes a minimum of two samples per cycle to identify a signal's frequency, and more than two samples per cycle to estimate its amplitude) so it's best to get rid of them.

adafruit_support_mike
 
Posts: 61407
Joined: Thu Feb 11, 2010 2:51 pm

Re: Avoiding electrically noisy connections?

by pneumatic on Sun Jul 19, 2020 11:23 am

I think it's a mechanical connection because what I'm seeing in sudden voltage drops on the mic analong input, followed by slight rebounds: It looks like this when I plot it:
Image
or a really big on with a big rebound:
Image
It's happening frequently enough (about 1500 times in the last 8 hours), that I'm having trouble isolating it to when I touch or move a certain part.

I even soldered the wire straight to the protoboard, and that didn't help either, so it must be somewhere else. Maybe on the microphone end.

pneumatic
 
Posts: 176
Joined: Sun Jul 26, 2009 3:59 pm
Location: Newark, CA

Please be positive and constructive with your questions and comments.