Very high precission COMPARATOR

[Tom_Sawyer]

Hi all,

what kind of architectures could you suggest implementing a high precission comparator using a BiCMOS process.

It is a non latched comparator.. where I must have a precission of 1.5% ... and I want to avoid the auto-zero-technique (question of design time) for the offset compensation..

When using bipolar input devices - what about the non equal base currents when one input is comming from resistive dividers?

Looking for help...

Best regards,

Tom

 

[reltol1]

Can you tell me some words I cannot understand clearly.

> a precission of 1.5%
Does this means that input overdrive voltage is 1.5% of the full-scal voltage?
If so, the answer is "depends". I mean, depands on how many overdrive voltage. If it's very small (i.e. several mV) you may need some correction technique. If not, just amplify it.

> I want to avoid the auto-zero-technique (question of design time)
Sorry, I dont understand (question of design time)??

> When using bipolar input devices - what about the non equal base currents when one input is comming from resistive dividers?
Effects, I guess. It will changes the input voltage. you may need consider this and check by simulation.

I hope this helps you.

 

[Humungus]

Please specify the minimum detectable input voltage in VOLTS;-)

A reasonable minimum input offset without auto-zero techniques is about 1mV. Is that enough for you?

 

[Tom_Sawyer]

Well, thanks for the reply so far.

I will give a more precise description:

I have to design an "under voltage" and an "over voltage" block for a reset block.
The reset block to detect the under voltage or over voltage must have a precision of 1.5%.
I will operate with a 5 V supply. The signal to be detected for over or under voltage is 2.6V. As reference I will get a trimmed 1% bandgap voltage of 2.39V.

For this reason I need a very precise comparator..

I thought using either:

o a simple comparator together with auto zero technique (high design effort also for the clock generation)
o or a comparator with a big input device to get around the offset voltage.

What do you think about this ?

Best regards,

Tom

 

[IanP]

I think you try to overcomplicate this task.
This can be achieved using basic comparators such as 311 with a good and stabile voltage refference source (ΔVref/ΔT<50ppm/°C) such as TL(KA)431 ..

 

[reltol1]

Still cannot catch you... If you have a block diagram, it would be helpfull.

 

[hojjat-kaveh]

http://www.sdrforum.org/pages/sdr05/4.1 System Implementation 1/4.1-03 Anderson et al.pdf