# Rail-To-Rail Comparator Design

I'm trying to detect small current pulses (~1.5mA). Currently I'm generating a ~20mV pulse on a sense resistor and feeding that to the input of an opAmp with some hysteresis. I'm having a bit of trouble where the output of the OpAmp will sometimes latch low and not return after the pulse passes and would like to try to use a comparator for the circuit. The problem I'm having with that is that my supply and my input are both at 6V.

Can someone recommend a good comparator that has full common mode input range with no more than 10mV of hysteresis? Or instead a better design overall?

I've found this part that seems to come close but the hysteresis might be too much.

*Based on some of the replies so far, I should mention I have a 3V3 rail available as well on the board if that helps.

EDIT: Redesigned circuit based on below suggestions- yet to evaluate

• If you're literally only looking at pulses. Why aren't you level shifting or using a capacitor to move the input signal away from Vdd? Apr 8, 2014 at 17:43
• I think that a better drawn schematic could help too. The capacitors' capacitances are missing, and you can't even tell the pulse current direction... Apr 8, 2014 at 17:48
• Fixed schematic- it's not only pulses, the pulses are on top of fairly constant current that can vary between 1 and 20mA Apr 8, 2014 at 17:57
• What speed are the pulses? How long do they last? Are they always producing a pulse of about 20mV? Apr 8, 2014 at 18:58
• The pulses are around 250us and roughly 50% duty cycle. The pulses are always 1.4mA so they should always be around 20mV Apr 8, 2014 at 19:36

Try this circuit with a dual op-amp. It uses the suggestion by @horta to move the voltage away from the rails.

U1A produces a stable voltage at +3V. The input is capacitively coupled to that voltage. Since your pulses are -22.5mV, the threshold is set at 3V - 11.25 mV approximately (set by the ratio of R9 to R8 approximately) +/- 2mV of hysteresis (set by the ratio of R7 to R8||R9 ~= R8)

simulate this circuit – Schematic created using CircuitLab

If you want to continue to use a R-R op-amp, you could adjust the circuit values in your schematic to give similar hysteresis and threshold, as in my diagram. If the input offset voltage of the op-amp is less than 1mV, similar values for the voltages could be used. In that case,

However, the failure of the circuit you have now to work properly indicates something else may be wrong- the bias current of the amplifier or the voltage offset is preventing the non-inverting input from rising high enough to reset the hysteresis. There also will be very little useful hysteresis for a 250usec pulse with that 47nF capacitor on there- the positive feedback will take a long time to build up (time constant 3.7msec).

• Makes sense, but how's it better than the original circuit? Apr 8, 2014 at 17:59
• "The problem I'm having with that is that my supply and my input are both at 6V." <- He's fixing this problem by making the input voltage to U1B operate around 3V rather than 6V. Apr 8, 2014 at 18:01
• It allows a non-RR op-amp. It isn't really operating right at 6V, the input pulses are 6V - 22.5mV and his threshold is set at 6V - 19.8mV with hysteresis of 0/-5mV. That's too much on the low side. The op amp might also have a few mV of offset. I'll add a comment based on spizzak's remark. Apr 8, 2014 at 18:06
• What kind of part should I be looking for for U1? Apr 8, 2014 at 18:53
• @spizzak It depends a lot on the pulse length. If it's audio-type frequencies, most op-amps with low Vos, rail-to-rail output and common mode range that includes 4V on a 6V supply will be good enough. Low Vos or you might run into problems with the small pulse. If your pulse is MHz you'll need something more specific. Apr 8, 2014 at 20:04

It would be wiser to amplify the signal far above the supply noise then use a Schmitt trigger or comparator with positive feedback as required.

One solution uses a current sense chip on either 3.3 V rail or 6V rail, whichever is cleaner.

http://www.maximintegrated.com/app-notes/index.mvp/id/1180 using a gain of 50 on a MAX4372.

• I like the current sense IC idea. You get a nice linear output based on current over the shunt resistor. And most of these work nicely in the 0-100mV range for detection. You can easily work with that IC's output for the OP's comparator with schmitt trigger output Apr 8, 2014 at 23:24