[what happened] Seeing ~30s "step-like" periodic signal

[CamDavidsonPilon]

Like the stirring PID isn't stabilizing to its target value.

[VitorFrost]

Could you show the graphs you have? I was using TCS3200 to detect color/growth stage, but I got two big problems:

• very high signal noise.
• raspberry 2 and zero can't handle the sensor. This sensor sends a frequency modulation to signal the intensity. But as Rpi has no real-time signal reading, it demands a a lot to process this signals. I am testing wright now the sensor APDS-9960. It is a I2C color sensor with an IR sensor and bult-in IR-LED. The only drawback about this sensor is that only sends 255 IR values. At least the proximity (IR) sensor is factory calibrated, and you have 4 IR-LED strenghts. Maybe we can detect liquid agitation with this sensor!

My TCS3200 signals were like that (9 days prolonged exp, and also adding an interquartile mean on day 7):

P.S.: This was recorded directly to a CVS file and graphs were plotted using spreadsheet programs.

[CamDavidsonPilon]

I'll see if I can pull up the graph. It was very "step-like" though, not close to what you have.

Your data is looking good though! Frequency measurements will be hard with an Rpi. We use a TIA circuit to convert current to voltage (We are also only supporting a mono-chromatic signal).

P.S.: This was recorded directly to a CVS file and graphs were plotted using spreadsheet programs.

Nice - what software did you use exporting to csv and plotting?

[VitorFrost]

My python program logs directly to a CVS file (It was easier for me then using the MQTT) Then I imported it to LibreOffice or MS Excel