[NI 9361] Dynamic Averaging on Period Measurements not working as expected

[CurranE]

I am trying to demodulate a frequency modulated square wave by using a NI 9361 counter on a NI 9174 chassis. I am measuring the period length in ticks, but it is always quantized at 2k ticks. At lower frequencies, 2k ticks quantization gives a comparatively low quantization error, and its hard to notice, but at higher frequencies of my application, the quantization is 10x my signal amplitude.

With the 100MHz time base, this translates to what I would expect if I was discretely sampling a square wave at 50kHz. However, this absolute quantization is independent of signal carrier/mod freq/amplitude and also independent from the Dynamic Averaging Measurement Method parameters, namely, the user inputs for expected signal range ('minimum value', 'maximum value'), 'measurement time' and 'divisor'. It was my understanding that these parameters could be tweaked to affect how the dynamic averaging affects quantization error?

Is this an effect of the chassis perhaps? Or am I fundamentally misunderstanding how this counter works. Is it not theoretically possible to have quantization of one tick? This is how I've interpreted the datasheet.

[CurranE]

Solved it myself:

I was using a function generator to supply the modulated signals, which was operating at 50kHz. This was not a problem for the sine wave, which is "interpolatable", but not for the square wave, where there would be up to a 1/50000s error, leading to the constant 2000 tick quantization.