COMP - Performance questions

[snoopy20]

Reference states a +/- 3% tolerance, which is rather high.

1. Do any of the reference voltages exhibit a better tolerance?

2. Is it possible to roughly model the tolerance, so for example if the tolerance goes high when temperature increases and vice versa I can make a software based adjustment by using the chip temperature reading (which should be of less tolerance)?

There is no performance data on using the dividers to set the hysteresis points.

3. How are the dividers working, resistive?, and do they add additional tolerance?

The performance mode only states to affect the response time.

4. Does increasing the performance mode also tighten up any tolerance? PIC16 internal comp's do.

I think that's it. :)

[Kenneth]

Hi,

Sorry, but we don't have more electrical data than what is in the datasheet, usually the comparator is not intended for high precision. So I suggest to test it if it can work for you, and if you need higher accuracy I suggest simply use the adc to read the input at given intervals.

Kenneth

[snoopy20]

The ADC is occupied for other things. It's a bit shocking to say the COMP is less accurate then the ADC. It doesn't say that in the sales pitch!

But it is what it is. Now, the references have a tolerance of +/- 3%. It doesn't say if they're all at this tolerance, but it would be great to know how these references are produced, i.e are they bandgaps or zeners?, is the 1.8v a resistor divide from the 2.4v making the 2.4v more accurate?

If they're bandgaps I can guestimate temperature drift using something like the LM4040 datasheet, similar for zeners. I presume temperature is a portion of the tolerance and I can counteract it to some degree by selecting different COMP hysteric divider values.