Expert Design - Peak Detect & Hold <100ns pulse

You should try moving the negative input of IC1 to the anode of the diode. Then the amplifier will eliminate the diode drop.

Yes, if sense input is from capacitor then it sure will reduce the diode drop. However, it may introduce capacitor leakage. Perhaps I can think of another buffer and then feedback.

Another problem is fast charger switch which can also be turned OFF fast. In that design, amplifier itself charges the capacitor. Perhaps a transistor switch might have beena better idea.

Look at using a fast JFET. Also, given the required timing you have indicated I think you can make the charging a lot easier by using a much much smaller capacitor. You might also ditch the 10k parallel resistor in favor of an active reset with a transistor.

My favoured method for high speed switching is to use a Lewis Bilateral gate using matched diodes, CA3039 for instance, the switching time can be measured in the 10's of pico seconds, depending on layout.

This then dispenses with the need for (ultra) high speed amplifiers with negligible input bias currents. A moderately fast op amp input buffer would suffice with a MOSFET voltage follower to buffer the storage capacitor...

Using the pulse for shutdown to drive an emitter coupled pair switching the gate should be ample.

John.

I used a similar design to detect the peak of the 2T pulse in the insertion test signal of a TV signal to monitor the transmission quality back in the mid 70's...

It was able to peak detect the 2T pulse with less than 2nS and to hold the voltage for the field time of 20 mS.

I've forgotton what Marconi Instrument that was used in!

John.

Dear John

Thanks for information on those fast diodes.

I have some of those diode arrays in metal can package and will try. I thought they were good only for making a mixer. I have those were from NS. Sure these will make a faster switch and good sample and hold. Yes, they were used as switch in early days when MOS switches were rather slow in micro second time. Intersil was a big name for many years. I used their I-8009 pico amp amplifier that never failed for my experiments in 20 years. It is no more produced.

I also have some pico-amp diodes PAD1 and PAD5 from Vishay (Siliconix design) and perhaps those also may work well.

How about the peak detection? I need some fast amplifier to take decision on the peak detection. Can you give more idea on your TV peak detector circuit for a cup of tea?

I remember some peak detection used in nuclear data ADCs in their early ND-100 MCA. They used first an analog switch to amplifier, then fast transistor amplifier, with fast current mirror source to charge some 100pF||100pF capacitor and their sense amplifier was isolated by a diode to reduce the sense current.

I will avoid discharge resistor and will use an analog switch to discharge the capacitor as suggested by Rcapper. Actually in that design, I had only hold and read functions and hence, discharge was made through resistor. Perhaps a transistor switch will be a much better idea.

I am also planning a double S/H circuit in place of one to make the decay rather small. First cap can be small and second cap can be big. First part peak detect and second only S/H. I will finally use some 12-bit ADC to get the peak data in about 1us. Or may be I can use the S/H of the ADC for the second part.

The reason for using the monolithic diodes in the RCA CA3039 package was mainly to have the diodes not only high speed but more importantly matched...

This gives very little offset between diode forward voltage drops and even more importantly the diode charge when switching on and off is balanced in the Lewis bridge so that the is practically no charge injection into (or out of) the storage capactitor, this also gives the incredible switching times possible, also, the leakage currents of these low leakage diodes is also balanced, so reducing the effective leakage by an order of magnitude...

Using a single diode or a FET etc... you will be introducing errors from the charge injected into or out of the storage capacitor, it can be nulled, but only over a limited temperature and operating range...

John.

Dear John

I am not sure if RCA is still there. Siliconix used to make a four FET bridge switch that worked dike four diodes. I am not sure if now this may be there with Vishay. Many great designs were lost in time transit.

Dear john

This IC CA3039 is no longer manufactured. I have few numbers, but I may never get this again. Hence, tell me some equivalent that is manufactured now. Any problem using schottky diode. They may sometime get damaged with too much current or over voltage. They are fast for sure. I need some plan here if this is to be tried out. CA3039 looks a history now.

Linear Technology has used this CA3039 in their application. See the diode bridge switch below. I also have some diode array from HP in cyclic chain used for frequency mixer. I will love to have these for analog switch. Those made for ESD protection may also work.

Hiiiiiiiiiii

I hav seen ur thread now itself and hope u got ur answer

or else iam also working on same peak detectors

i can give my design....

Go ahead and let us all know your design and its performance.

I have used a simple one. u want nearly 300nsec rise time right

Iam not getting how to upload my circuit

click on Green symbol button in this Page and it will ask for a file. use JPG or GIF file to load.

Iam using the circuit about

but i also want to use the automatic switch that can hold the peak for a specified time automatically. Can u help giving such type of circuits today and those should be as simple s possible like this....

In place of discharge resistance, use a transistor or FET switch.

sir can u send me a circuit diagram so that i can understand nicely plz if u don;t mind...

replace the resistance in parallel to capacitor and place a JFET switch to short the capacitor to discharge whenever you wish to do reset it.

Sir i was waiting for the circuit from u....

Sir if u don't mind can u give me ur mobile number so that i can call u if u don't mind and can u also say what time u may be online daily so that i can discuss some topics with u and call u plz...

I will be waiting for ur reply.....

I suggested part change in your circuit. Replace resistance with capacitor discharge switch controlled by some timer. Are you not able to understand? Short circuit the peak detector capacitor by a switch and remove the resistance you have placed.

I hav done sir but i didn't get the required output...

can u send me some circuit plz.....

what is that you get from your circuit? Make a sketch in paint brush.

I have planed NMosfet across capacitor and during 1st pulse of reset iam getting true output abut from next pulses it's taking much rising time to charge capacitor. I will send u my circuit....

rise time is affected by the current drive capacity of the OP Amp and diode. Replace OP AMP with higher drive current version or use a transistor at the output of the op amp to force more current into capacitor to charge faster. you can also reduce the capacitor value to say two 100pF Mylar film type in parallel.

Good day Sir!

Please send me or upload the final design which you have finialized after all this discussions as i also want to detect and hold peak value of the pulse of 100ns.

Best regards.

Dear,

I am working on very fast biomedical electronics and I should use some fast peak detectors, I sow that you are working on it, Could you find a good circuit for your problem, If yes can you help me either, if it is possible for you plese inform me by Zendegi_zibast_63@yahoo.com because I am not member of this forum.

Thanks in advance,

Behnood Barmaye

Hello all, I'm also searching for a medium fast peak detector for 10mV - 1V range pulse peaks. If you know some quality stuff matching my requirements please send me ASAP. indunilabaya@yahoo.com

I may be of help now.

write to me. I am working on this design now.