Teensy 3.6 ADC

[Raptor]

The specifications list the Teensy 3.6 ADC as 13-bit which seems odd -- is it really?

I haven't really played with the ADC yet but want to read current values across a current sense resistor so knowing more about the way the analog inputs are handled in the Teensy 3.6 would be nice -- any pointers to information I need? I now the 3.6 is a 3.3V device so presumably the analog input is limited to 3.3V but, once again, the specifications seem to indicate there's no gain amp.

What's the max plus and minus voltage that each channel can handle and what's the effective voltage resolution at max resolution?


Brian

 

[defragster]

There are at least a few threads around about this - indeed the hardware does resolve to 13 bits of resolution - an odd and prime number.

However getting near that resolution - even 12 bits - requires proper setup and delivery of a clean measurable signal - and a quiet environment in the chip to discern it.

Indeed any value over 3.3V will read full scale - or damage the hardware.

 

[Raptor]

Thanks defragster. Yeah, the 2 least significant bits are usually suspect and keeping a clean signal path is useful.

In my test rig, a rig for testing Peltier modules, I plan to add a couple current sense resistors to monitor the current going to the Peltier module and the heater. The Peltier module will be turned on and left on at 100% power (no PWM) but the heater will be controlled via PWM so measuring the current could be a trick to get the timing right for when the PWM is on versus off. I have the PWM frequency set to 75Hz so a PWM value of 1 out of 255 would provide an on duration of 52uS so being able to measure the current during that period would require a sample duration of less than that AND starting the conversion at the very beginning of the PWM cycle -- not sure how to do that. My work around is to do a short test at the beginning of the test run with the PWM value at 255 and then calculating the power from that. I'd prefer to be able to monitor current continuously during the run but the work around should be sufficient.


Brian