I’m seeing a (IMHO) fairly high standard deviation (around 35) in readings with a fixed whetstone bridge (4.7K resistors) directly attached to the NAU7802 module, the gain is low at just 2 and the sample rate fairly low (40/sec) and I’ve directly attached the module with a short qwic cable to a qwic hat.
I think the NAU7802 supports separating the excitation voltage (VDD) from the voltage used to drive the I2C bus. I’d like to just try and use a small 4.5 volt battery pack to provide as stable a voltage as possible to eliminate the noise and lower the standard deviation. However it’s not clear to me from the schematics if that is possible? Pads that need to be cut? There’s warning about not putting anything more than 3.3 volts on the I2C bus - so I don’t want to just ‘give it a try’
[My ultimate design has 4 sensors attached to the QWIC mux board and using the I2C drivers and around 8 feet of Cat6 cable between mux and sensors - but in that configuration the standard deviation is way worse, around 150+ so I’m really trying to track down the source of the noise right now]
I took a chance and added a small 2 x AAA battery pack (so as not to exceed the 3.3V) and changed the code to not use the internal LDO voltage regulator and it appears to have greatly reduced the std deviation I was seeing (down to around 2-7, about 0.1%)
However I also noticed that attaching the battery pack lit the LED on the NAU7802 module, which makes me thing that the red/E+ VDDA pins are connected to the 3.3 power supply rails in some way? The schematic doesn’t seem to indicate that, is the chip passing the power back through in some way? Either AVDD or REFP pins are actually internally connected to DVDD?
Thanks for posting the PCB layers - dunno why I didn’t think of looking there too!
Yeah - I probed it and there’s no continuity between the VDDA pins and the I2C 3.3V Pwr, so I’m reasonably reassured.
The NAU7802 data sheet does say that an externally supplied AVDD power must not exceed the DVDD power by more than 0.3 volts, so I think my 2 x AA cell power pack is safe, and it definitely makes a significant difference to the stability of the readings.
Given the PCB layout I can only explain the LED being lit by there being some sort of internal path within the NAU7802 - but I don’t think it matters as long as I keep within the specs.
I’ve not used the NAU7802 but similar devices can use external excitation by simply leaving their EXC terminals open and installing the batt/PS on the cell or gauge wire pair. In other words, the instrumentation won’t ‘see’ the battery, except by way of the signal and that will be some dozens of millivolts.
Of course, your measurements must now take into account the unknown Excitation voltage.
