Continue to Site

Welcome to

Welcome to our site! is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Peak detector with a twist.

Not open for further replies.


Mar 27, 2022
Reaction score
Trophy points
Activity points
This is my situation. I want to make a peak detector that always puts the peak value measured in the last 5 microseconds at the output. Conventional peak detectors have a reset pin. If triggered, a mosfet shorts the peak holding capacitor to ground and then gradually the voltage on it goes to the peak. I want the peak detectors response to be as below(red curve)


Any suggestions?

I'm not sure I understand the relationship between the red line and the peak in your image but I think all you need is to reset the measurement every 5uS using a timer. In your image the signal seems synchronous to the time interval, always falling to zero at 5uS intervals and peaking in mid interval, if that is always the case it would be more sensible to use a sample and hold circuit triggered in mid interval.



the fourth flat red line (from 15 to 20 us) is wrong. It does not represent the pekk of the time of 10..15us.

And the timing looks like "synced" tot the input. This usually is not the case, unless you specially design it.
But how do you define the "sync" condition?

Digital solution: ADC --> microcontroller --> DAC
Analog solution: two alternating peak detectors.



a pure analog peak detector would provide the peak value immediately, thus your red line should follow your blue signal and stays max at 2.5 µs, 7.5µs, 12.5 µs & 17.5 µs (including a reset between the intervals). Or does the red line shows the ADC read-in at 0 µs, 5 µs, 10 µs & 15 µs. Are you aiming to reset your peak detector e.g. right after the ADC read in e.g. by a timer relized within a MCU?

At which point are you aiming to evaluate the voltage? You are not able to provide the peak value from interval 5 µs to 10 µs over the time instance 10 µs to 15 µs, and "detecting" a way lower peak values in the time interval 10 µs to 15 µs. You would need two hold ciruitries as suggested @betwixt e.g. P-MOSFET or Analog switch & capacitor. To store the peak value of time instance e.g. 5 µs to 10 µs lets say MOSFET#1 is conductive during this time interval and charges C1, while MOSFET#2 is not conductive and holds the max value of time interval 0 µs to 10 µs. Short before time instance 10 µs you have to
  1. discharged C2
  2. turn off MOSFET#1
  3. turn on MOSFET#2

The waveform, except for the fault mentioned by KlausST can be generated by the combination of a peak detector and a S/H, both clocked by the 5 us tic. The behaviour for non-zero signal at clock tic should be clarified.

Not open for further replies.

Part and Inventory Search

Welcome to