It took me few weeks to find out TM1637 produces noise

[Georgy.Moshkin]

I am working on a device that has a 4-digit 7-segment LED screen with a TM1637 driver. From the beginning of the development, I filled the screen with four dashes '- - - -'. I was aware of possible switching noises, so there is LC filtering. Plus, I was sure that I would spot any noises early because I had used the AGC amplifier, and switching noise from four active segments should have been visible if filtering was not enough.

I was wrong. After spending several weeks debugging the algorithm, checking power rails with the scope, recording and replaying digital signal back to the microcontroller from the PC, I accidentally left the '6' digit on the screen yesterday and noticed a huge noise, with only two more segments active! Maybe the voltage is dropping, or the noise content is very different; I will figure this out.

Attaching the spectrogram. Note how the switching noise is reduced when only four segments are active.

[mikerj]

This is a multiplexed display so the more segments you have switched on, the greater the total current change when the driver switches between segments (since LED current drops to zero in the switching period).  You need to ensure the TM1636 has plenty of supply decoupling capacitance, that the LED segment traces are routed well away from any sensitive parts of the circuit and the TM1636 is connected to a low impedance ground.

[Nominal Animal]

Small 0.91" (128×32) and 0.96" (128×64) OLED display modules also have similarly "spiky" current draw.  I don't have a proper oscilloscope to quantify it, though.   I do have both TM1637's and various models of these OLED display modules, and have observed this before.  Fortunately, I was aware of this beforehand.

I've been looking at the various filtering approaches (including LC and CLC filter on the VCC), and am leaning towards using a separate linear regulator for these.  (This is not a problem at all for me, since the modules and my MCUs work at 3.3V, but I always have 5V available.)

Problem is, which small regulator to use?  At least some of the ubiquitous '1117 variants suffer from failing short, i.e. providing too high voltage on the output when they konk out, and I don't like that.  Many require specific types of capacitors and capacitance values on input and output, or they will oscillate.  I'd rather use a low-noise low-dropout one, so that it would work well over the entire USB voltage range (4.5V to 5.5V, so somewhat less than 1.2V drop), with just 1µF or so capacitance on the input to the regulator (and plenty on the output for these noisy devices), as USB limits the total USB VCC capacitance to 10µF.  And I'd like to standardize on one, so I can buy 100 of them from Mouser in SOT23/SOT25/SOT26/SOT89/TO-92 package, with suitable capacitors in similar amounts.

Plus, using several instead of a single regulator, assuming there are no power sequencing issues, means better distribution of heat generated, making it easier to keep passively cooled gadgets cool in plastic enclosures (without heatsinks).

Decisions, decisions...

[Hiemal]

There's a good few parts from diodes.inc in the "1117" type that are explicitly stated to be okay with ceramics.

https://www.diodes.com/datasheet/download/AZ1117C.pdf

https://www.diodes.com/datasheet/download/AZ1117I.pdf

https://www.diodes.com/datasheet/download/ZLDO1117Q.pdf

Any of these should be alright.

[temperance]

Can you clarify what is in the spectrogram? What is the meaning of "7.5km/h"?

[Georgy.Moshkin]

Can you clarify what is in the spectrogram? What is the meaning of "7.5km/h"?

I measure speeds using radar sensor. 7.5km/h is a speed threshold. Spectrogram contains Doppler frequencies up to around 50km/h as a result of FFT. As soon as first speed was measured and displayed on the LED screen, algorithm started to glitch. For example, changing from "- - - -" (4 segments, one per digit) to  "6" (6 segments in first digit) resulted in a huge noise. I didn't know that noise was present. I tried to debug it by recording the signal, but couldn't find the problem, because during the recording screen was not updated. During one of firmware updates I haven't cleared the LED screen and realized that signal became very noisy.