About the preamplifier in the comparator design.

[carpa]

latched comparators by ken martin

In my opinion, the main contribution of the preamplifier is its prevention from the kickback effect generated by the following latch during its transition from track mode to latch mode [ref: David Johns & Ken Martin]. Although the preamplifier CAN attenuate the input offset of the latch, it has offset itself, which needs also extra circuits to reduce it. Therefore I think acturely the preamplifier contributes nothing in the aspect of offset-cancellation.
Am I right?

 

[zmliu]

I think the effect of the preamp on reducing input offset is dependent on its gain.However, since GBW is fixed, preamp will actually not play this role in high-speed applications.Other ways for cancelling offset should be used.

 

[gevy]

Let's consider the comparator consisting of the preamplifier and latch.
The latch functions in a large signal mode with output logic levels and usually it should provide the maximal speed. Hence sizes of transistors of latch should be minimal and consequently the latch will have the large input offset.
The amplifier functions in a small signal mode at the comparison time. Its gain gets out small from 5 up to 20. Therefore amplifier can have the large sizes of transistors and large enough GBW and small input offset. Besides the amplifier reduces kickback signal, it is very important.

 

[yaxazaa]

Besidez, offeset, speed, noise is also one factor. Preamplifier acts as amplifier and buffer as well.

 

[dreamheater]

If we use two stages of amplifiers with autozero caps in between, then theoretically the offset from the 1st preamp will be totally cancelled out, and only the offset from the 2nd preamp presents at the output. Therefore the 1st preamp can live with small input devices, while the 2nd preamp needs not to have too much offset. A strategy is to use minimal channel length in the 1st preamp to boost the gain there, but not in the 2nd preamp to avoid otherwise high offset.