Fast peak detector for 10ns pulses

[electrov]

Hi,

I would like to build fast peak detector for 10ns pulses, PPM, 250ns repetition (4MHz), amplitude from 10mV to 5V.
How circuit I can use for it?

[sdouble]

I simple comparatot if you don't need any accuracy in the time measurement or a constant fraction discriminator if you need to be accurate.

[bson]

You need to start by figuring out what you need (not want) in terms of accuracy and precision, and what aspects of the pulse are important to you (periodicity, amplitude, phase, relative amplitude, pulse width...).

[Wolfgang]

an LT1715 should do (4ns comparator).

[Mechatrommer]

How comparator can detect 'analog' peak?

[Wolfgang]

by a sample and hold circuit.

[David Hess]

I would like to build fast peak detector for 10ns pulses, PPM, 250ns repetition (4MHz), amplitude from 10mV to 5V.

PPM?  Pulse position modulation?

What exactly is it that you want to detect if the repetition rate is 4MHz?  Do you want every pulse amplitude?

Is there a strobe signal available?

If a strobe is available, then there are some sampling ADCs which can do this directly.

If no strobe is available, then 10 nanoseconds should be possible with an open loop diode half-bridge based peak detection circuit.  Figure 21 on page 15 of Linear Technology application note 61 gives some idea of how this works.  A faster buffer, faster transistors, and lower hold capacitance would be needed.

If you are looking for peak detection like modern DSO use, then they just sample the input at the full sample rate and keeps the highest values within a given duration.  CCD based DSOs did it some other way in the analog domain but I am not sure how.

[electrov]

I want to measure laser pulses. This is reason why PPM.
I haven't strobe signal and I would like to build it with analog circiut. I know that FPA+ADC will solve the problem but is expensive.
I know methods of digital solutions. I need simply and uncomplicated analog circuit. I try witch OPA615 and aplication note example but without good results. It understands amplitude. I search a example schematic.

[mikerj]

I want to measure laser pulses.

What parameter of the laser pulses do you want to measure?  A "peak detector" is literally used to hold the peak value of a varying signal.  PPM suggests you have a digitally modulated signal which you want to demodulate, so a peak detector would not be useful for this.

If you simply need to detect the presence of each pulse over a wide range of amplitudes you should be looking for a "data slicer".  This circuit compares the incoming signal with the DC average of the signal, so the decision threshold is automatically adjusted as the signal amplitude changes.

[SiliconWizard]

Yes it's not 100% clear.

It's already challenging to design such a fast peak detector with good accuracy.
But a peak detector will hold the peak value until it's reset. What exactly is going to reset it in your case? Are you interested in each pulse's individual peak value? Are you looking for a max or an average over a certain number of pulses?

[ejeffrey]

What type of laser?  Does the laser have a strobe that tells you when the pulses are coming?  If so, you can just use that to trigger a sample-and-hold or an ADC with a fast sampling aperture.  If not, you will need to generate the trigger from the pulse itself with something like a CFD.

[Wolfgang]

Hi, I got another idea:

take your input signal, run it thru a wideband buffer amp (MMIC) and distribute it to, say, 10 channels.
At the end of each channel sits a comparator with a preset comparison voltage in steps of 1/10th to the maximum.
When a comparator fires, the output pulse is put into the trigger of a 74AC or even faster flip-flop.
The flip-flop with the highest comparator value is your maximum, with 10 steps resolution.

[sdouble]

like using a hammer to open a nut.

Hi, I got another idea:

take your input signal, run it thru a wideband buffer amp (MMIC) and distribute it to, say, 10 channels.
At the end of each channel sits a comparator with a preset comparison voltage in steps of 1/10th to the maximum.
When a comparator fires, the output pulse is put into the trigger of a 74AC or even faster flip-flop.
The flip-flop with the highest comparator value is your maximum, with 10 steps resolution.

[Wolfgang]

... yes, but the gentleman wanted something analog. A 16bit-ADC running at 500MHz plus a fat FPGA does a better job, agreed.

[mark03]

Everyone is brainstorming ways to measure the height of the pulses, but the OP stated that this is for PPM.  The pulse height shouldn't matter, only a pulse / no-pulse discrimination in each timeslot.  OP needs to clarify.

For PPM demodulation, I posted a circuit several days ago which I am in the process of prototyping.  It fell off the first page without any comments or replies, so it must have been flawless    I have since changed the op amps to slightly faster ones and simulated in LTSpice.  It seems to work fine, no pricey ADCs or FPGAs needed.

[StillTrying]

Everyone is brainstorming ways to measure the height of the pulses, but the OP stated that this is for PPM.  The pulse height shouldn't matter, only a pulse / no-pulse discrimination in each timeslot.  OP needs to clarify.

Yep, it's all as clear as mud.

For PPM demodulation, I posted a circuit several days ago which I am in the process of prototyping.  It fell off the first page without any comments or replies, so it must have been flawless

I remember looking at it now. If it's a quiet thread and you have new information to add, wait until the post is on page 2 of the index and then add the new info as a new reply to get it another outing on the 1st page.