We are using ADVFC32SH / 883B Q 1044C chip to convert the signal from a temperature sensor into frequency.
By varying linearly the temperature (or the input voltage applied through a resistor connected on pin 2 = IN-) we noticed that the output frequency has a fairly linear variation. However, there are several zones where increasing the input voltage would not change the output frequency. Frequency would rather stay constant for a while and then it would start increasing again. Some of those zones where VFC32 does not respond, for a while, at an input voltage change are at the approximate frequencies of 60kHz and 76kHz (give or take). There are some other zones, too.
The overall linearity of VFC32 is still inside the specifications, i.e. +/- 0.05% @ Fmax < 100kHz. We actually expected that the nonlinearity to be more visible either at the extremities or in the middle of the voltage/frequency chart. But what we experienced is a linear response with 2-3 or more glitches between 20kHz-100kHz. It looks more like a discrete (and quite abrupt) correction is being applied from place to place to the v/f curve.
We changed several chips and all of them seem to have those "inflection points" in the same areas of frequencies.
In case you need more information, please see the attached image showing the frequency staying stable in time at 75584 Hz, although the input voltage is being increased constantly.
Therefore we are wondering if we should look for something wrong in our circuitry or this behaviour is normal for a vfc converter.