How to design an AC coupled differential input comparator

[myatham]

Hello,
I need to design a differential comparator with maximum of 100V common mode range, with 200mV is hystersis, AC coupled, 10KHz Bandwidth. This is to detect pulse inputs. It would be great help if anyone can suggest design steps.

Thanks in Advance!

 

[KlausST]

Hi,

design it like it is DC. (two R as voltage divider at each input).Values depend on common mod input range.
Then add the positive feedback to generate hysteresis. (It depends on comparator output voltage and input impedance)

Then add a series capacitor at each input.

Each path will be: C1-R1-R2

(to be correct: with the "C" the hysteresis my differ a little. I don´t expect much, because R1 >> R2.)

Easy to simulate with ltspice or other simulation tool.

Klaus

 

[myatham]

My requirement is the circuit shall support both single ended signals (1Vpp min) and differential signal with 500mV peak to peak with common mode range 100V, Normal mode range of 0 to 200V.

If I add resistor divider to scale down to the comparator common mode range (assume 10V max), i need to scale down with a factor of 10 from 100V. But, when my input signal is 500mV, the actual output at the comparator will be 50mV. So, i need to compromize with hysteresis here.

One more point is for single ended inputs, how this circuit will work. Even the input voltage pulses between 5V and 4V, it will be considered as pulse train though it is oscillating in logic high state. How shall we design for this?

Please provide me the notes in detail if possible. Thanking You,

Regards,
myatham

 

[KlausST]

Hi,

the actual output at the comparator will be 50mV.

Sure: output 50mV?
Hopefully "input at the comparator".

***
About hysteresis:
100V with 200mV hysteresis.
Divided by 10 gives:
10V with 20mV hysteresis.

Klaus

 

[chuckey]

If you AC couple the inputs, providing you have sized the capacitor correctly and made provision for its charging current, then you can ignore any common mode DC voltages. If however the " 100V" is actually AC, then its a different problem, i.e. looking for pulses sitting on a common mode 50 HZ sine wave. Filtering might help.
If its the DC case then your pulses will appear across the resistor from op-amp input to earth, so there will be some amount of equal area sharing which could introduce distortion if the pulses are too wide for their PRF.
Frank

 

[myatham]

Typo, it is "input at the comparator will be 50mV". Thanks for your quick reply.

Regards,
myatham

 

[Warpspeed]

and differential signal with 500mV peak to peak with common mode range 100V, Normal mode range of 0 to 200V.

I would suggest a fully floating isolated input, coupled by an opto isolator which should easily carry the required 10 Khz bandwidth.

An input comparator with the required 50mV hysteresis, protected by some back to back diodes, should meet all your stated requirements pretty easily.