I'd like to use atmega328p to read force with strain gauge. I'm using one 1k strain gauge and 3x 1k ohms dummy resistors.

My circuit looks like that: enter image description here enter image description here

I'm using DAC to remove offset from bridge.

As a result I'm getting so much noise on atmega pin.

It looks like that: 1) No averaging (It's like only 0's and 1023's): enter image description here

2) 30 readings running average: enter image description here

I've tried changing R4 to 2,4k, 3k and 10k ohms. 2,4k and 3k helped just a little bit, but while using 10k output of ad8237 goes to 0 in a few seconds. I've removed C11 - nothing better happened.

AVCC is 3.3V with 10uH inducto in series. It's used to power strain geuge bridge, AD8237, DAC, atmega328p analog VCC.

Atmega328p AVCC and AREF are decoupled with 0.1uF capacitors.

Aver unues atmega pin is set as LOW output.

Is there any way I could remove that noise?


Everything solved by removing C23, thank You @EinarA

Now different problem is noise when touching sensor body. On the photo: lower amplitude - sensor lying on its own, higher amplitude - im touching sensor body with one finger. That's not big difference but it is. Sensor body is made od 1.4301 steel. Sensor body is not grounded. Is it a good idea to connect cable shield to body or just ground body? Cable shield is connected to ground on just pcb side. enter image description here

  • \$\begingroup\$ Do you have an oscilloscope? Kind of looks like something is oscillating. If you don't have a decent scope handy, try a 100nF across the 300K just as a check. \$\endgroup\$ Mar 30 '19 at 23:24
  • \$\begingroup\$ How long are the leads to the strain gauge? Are they shielded in any way? \$\endgroup\$
    – TimWescott
    Mar 31 '19 at 0:19
  • \$\begingroup\$ You did not include an RF input filter nor show your shielded twisted pairs to interface both sensor and DAC has the shared ground induced noise. Also CM noise possible aliasing noise with RF and lack of aliasing filter can result in more noise than your signal. Isolate each source to locate the noise and correct as I suggested. Read about EMC from Henry Ott. \$\endgroup\$ Mar 31 '19 at 0:24
  • \$\begingroup\$ Make R30 and R40 be 1,000 ohms. Make C21 and C22 be 1uF. \$\endgroup\$ Mar 31 '19 at 3:21
  • \$\begingroup\$ use a ground plane under c11, c22 and u6 \$\endgroup\$ Mar 31 '19 at 3:27

Question 1: Remove C23. It is making the amp oscillate.

Question 2: The signal you are getting from the now working Amp looks like power line noise. This is a problem with just using one strain gauge. If I could walk into your lab I could tell right away if there is an easy fix or if it is going to be hard. Since I can't it will be a lot of shots in the dark. The first thing to try is grounding the ' body' to the amp power supply via a separate wire, not the shield. If the 'body' is not connected to anything major this could be enough, but if it is connected to other things that are connected to the power line this could be worse, You should look up information on how to fix this and get some more ideas as I am not really able to lead you through each possiblity one by one.

  • \$\begingroup\$ Sensor is not connected to anything right now, but will be in the future - to the motorcycle frame (so nice ground point). I'll test that next weekend. \$\endgroup\$
    – Anrzej
    Apr 1 '19 at 17:23
  • \$\begingroup\$ Anrzej, I have combined my first comment into this answer. Please indicate that you made use of it. \$\endgroup\$
    – EinarA
    Apr 5 '19 at 3:14
  • \$\begingroup\$ I have installed sensor on the motorcycle and problem seems to be gone. Now body is connected to ground. As I said before I've removed C23 already. \$\endgroup\$
    – Anrzej
    Apr 10 '19 at 6:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.