MCC118 Daqhat Coefficient Scaling Bug

[ehv]

Hi,

I'm working on getting my MC118 to output scaled data using the mccdaq daqhats Python library, but it's not working as I would expect.

If I call calibration_coefficient_write() with a scale of 1 and an offset of 0, I expect OptionFlags.DEFAULT and OptionFlags.NOSCALEDATA to output raw ADC codes when calling either a_in_read() or a_in_scan_start().

ai_hat.calibration_coefficient_write(0, 1.0, 0.0)
print(ai_hat.a_in_read(0, OptionFlags.DEFAULT))
print(ai_hat.a_in_read(0, OptionFlags.NOSCALEDATA))

This outputs:

-0.029296875
2042.0

The documentation says:
https://mccdaq.github.io/daqhats/python.html#daqhats.mcc118.calibration_coefficient_write

The coefficients are applied in the library as:

calibrated_ADC_code = (raw_ADC_code * slope) + offset

This looks like a bug to me, unless I'm doing something wrong.

Here is a test program that demonstrates the bug:

#!/usr/bin/python3

"""
Demonstration of a bug with scaling in a_in_read() and a_in_scan_start()
OptionFlags.NOSCALEDATA does not do what I expect.
ai_hat.calibration_coefficient_write() affects the result.
https://mccdaq.github.io/daqhats/python.html#daqhats.mcc118.calibration_coefficient_write
"""

from daqhats import mcc118, OptionFlags

address_118 = 0
ai_hat = mcc118(address_118)

def simple_read():
	print(ai_hat.a_in_read(0, OptionFlags.DEFAULT))
	print(ai_hat.a_in_read(0, OptionFlags.NOSCALEDATA))

def scan_finite(nsamples):
	ai_hat.a_in_scan_start(0b1111, nsamples, 10000, OptionFlags.DEFAULT)
	print(ai_hat.a_in_scan_read(nsamples, 1.0))
	ai_hat.a_in_scan_cleanup()
	
	ai_hat.a_in_scan_start(0b1111, nsamples, 10000, OptionFlags.NOSCALEDATA)
	print(ai_hat.a_in_scan_read(nsamples, 1.0))
	ai_hat.a_in_scan_cleanup()

def set_coefficients():
	print(ai_hat.calibration_coefficient_read(0))
	ai_hat.calibration_coefficient_write(0, 1.0, 0.0)
	ai_hat.calibration_coefficient_write(1, 1.0, 0.0)
	ai_hat.calibration_coefficient_write(2, 0.0, 0.0)
	ai_hat.calibration_coefficient_write(3, 0.0, 10000.0)
	print(ai_hat.calibration_coefficient_read(0))

simple_read()
scan_finite(1)
set_coefficients()
simple_read()
scan_finite(1)

Here is my output:

$ ./bugtest.py
-0.000786520708334848
2046.795118641332
MCC118ScanRead(running=False, hardware_overrun=False, buffer_overrun=False, triggered=True, timeout=False, data=[-0.010979898809909017, -0.0017999143186528954, 0.030290890733175146, 0.48317093277360357])
MCC118ScanRead(running=False, hardware_overrun=False, buffer_overrun=False, triggered=True, timeout=False, data=[2044.70751480613, 2048.674508288587, 2056.298605373789, 2147.9981528073927])
MCC118CalInfo(slope=1.0438019176008568, offset=-82.56079326441613)
MCC118CalInfo(slope=1.0, offset=0.0)
-0.0390625
2041.0
MCC118ScanRead(running=False, hardware_overrun=False, buffer_overrun=False, triggered=True, timeout=False, data=[-0.0439453125, -0.0390625, -10.0, 38.828125])
MCC118ScanRead(running=False, hardware_overrun=False, buffer_overrun=False, triggered=True, timeout=False, data=[2038.0, 2040.0, 0.0, 10000.0])

Only when slope=1 and offset=0 is when NOSCALEDATA outputs the raw ADC code. That's not right.

[ehv]

Ok, thank goodness for open source.
https://github.com/mccdaq/daqhats/blob/e6d96c8fb621c83696536a037f1c4fa373c46068/lib/mcc118.c#L1376

It's clear to me now that OptionFlags.NOCALIBRATEDATA and OptionFlags.NOSCALEDATA interact. The bitwise OR gives the raw ADC codes, as I wanted.

I suppose that I expected a single scaling to happen, not two. It's a bit weird to separate the calibration from the scaling and not provide a way to modify both.
If I had it my way, it would be a single scaling step that combines calibration and scaling.

I understand how to work around this, now. Calibrate the raw ADC codes against the calibration reference that I have, then use OptionFlags.NOSCALEDATA with my scaling values.

Sorry for the noise.