STM CircleOS forum / mems noise ?

ThomasScherrer · 2007-10-30 14:10:25

Hi I tried your project,
in any of your screens the values jitter really much,
when I place the circle flat on table I should read 0 , 0 , 1000 ??
but I have jitter of 40-50 counts constantly, any idea what is wrong ?

Francis · 2007-10-30 19:40:50

<<I should read 0 , 0 , 1000 ??>>

Yes, with a table perfectlly horizontal, with a device perfectly soldered...
I am not an expert in MEMS but it is hard to say what is the part of the ambient vibration, the part of the noise on the power supply, etc...

armaniake · 2007-11-02 20:12:11

I have seen that jitter also.

Looking at the schematic, it is likely that the DC/DC switcher is the culprit. PSU rails have to be very clean with high res. accelerometer. I have used before a MEMS from the same family powered by a DC/DC and it was horrible. When added an LC low pass filter at the MEMS location, the jitter felt down to a reasonable +/-3LSB (FS2g, 12bit).

At this stage, software filtering needs to be used to clean a bit the jitter (averaging). But I strongly recommend one implements an LC filter next time between the DC/DC and the MEMS, or use a linear to drop from 5V. Do not connect directly the PSU of the MEMS to the PSU of uC.

Practically, the jitter is not linked to how horizontal is the MEMS, and not likely linked to the soldering (it could be in theory). It is linked however to any vibration you impose on the Primer. But you should start to feel this vibration if you see a jitter of 40 LSB.

dalewheat · 2007-11-09 02:50:20

I am seeing the same jitter from my MEMS outputs.

I'm wondering how difficult it would be to surgically install an additional inductor in the circuit. Looking at the schematic, it appears that there are two decoupling capacitors dedicated to the MEMS device, C36 (10uF) and C37 (100nF). I've looked closely at the PCB and see two undesignated capacitors immediately below U6 (the MEMS accerometer).

If these are the capacitors that I suspect that they are, would it be possible to make a tiny cut in the trace that leads from VCC3 to the capacitors and then to the IC, and then splice a surface-mount inductor (10uH, for example) into the circuit? If the board layout is not conducive to that modification, how about removing the capacitors altogether and installing a very small, flying PCB in its place containing the capacitors and the inductor?

I'm not suggesting everyone try this - I do this kind of work for a living and have the tools that make it straightforward. What I want to know is if the PCB is laid out in such a way that this sort of change is possible. I don't really feel like stripping all the parts off to find out

A peek at the production artwork would be helpful here.

Thanks,

Dale Wheat

dalewheat · 2007-11-11 12:52:45

OK, I found the PCB artwork on the documentation page - it was there all along!

The modification I suggested is possible but tricky. I'm not 100% sure that I want to try it at this time.

Any chance that future revisions of the STM32 Primer artwork will incorporate this LC filtering?

Thanks,

Dale Wheat

Francis · 2007-11-11 16:24:52

It is not plan to redesign the STM32 Primer in the near future. Perhaps when the next derivative will come... Anyway, it's important to collect the ideas of improvements/extensions.
Regarding the jitter from the MEMS, I was wondering whether a part of the problem is not caused by the high rate of the SPI communication. It would be easy to slow down the frequency of the read actions... (by increasing the timer period). At this time, we have optimized the rate of the read access, but it's could be that it is a bad idea... To be checked...

# 1 2007-10-30 14:10:25 mems noise ?

mems noise ?

# 2 2007-10-30 19:40:50 mems noise ?

Re: mems noise ?

# 3 2007-11-02 20:12:11 mems noise ?

Re: mems noise ?

# 4 2007-11-09 02:50:20 mems noise ?

Re: mems noise ?

# 5 2007-11-11 12:52:45 mems noise ?

Re: mems noise ?

# 6 2007-11-11 16:24:52 mems noise ?

Re: mems noise ?

Board footer