THS4215: THS4215 high speed peak detect circuit

Hey Frank, 

Couple of things, 

1. Your hold cap is way too high, basically a short circuit at higher F, 

2. Normally you use a CMOS or JFET output stage to reduce droop, 

3. The OPA836 I used here may not be fast enough, what you really needs is super high slew rate to get the 1st stage tracking the input as it exceeds the held voltage from the prior cycle, 

here is a modified circuit using the new OPA817, needed to use +/-5V on that stage, this is 10MHz were the filtered held output should be 1.4V, I get 1.37V rippling a little, 

And this file, just something to start with, 

OPA836 Peak Detect with OPA817 output.TSC

Hi Frank,

I see three issues:

1. The 1N5817 is a 1A rectifier which is very unsuited for HF rectification. You need a special small signal Schottky rectifier with ultra low junction capacitance to be able to rectify 100MHz.

2. As already mentioned by Michael, the sampling capacitance is way too big. For 100MHz much lower sampling capacitances are usually chosen.

3. The circuit may suffer from instability caused by the capacitive load being directly connected to signal ground. Depending on the circuit and the used OPAmp the introduce of an isolation resistor in series to the sampling capacitor can help.

We already have discussed these topics in these threads (and many more):

https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/968283/ths4631-about-the-effect-of-r3-in-peak-detector

https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/940064/how-to-improve-the-accuracy-of-peak-detector-as-high-as-posssible

https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/997908/high-speed-peak-detector

https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/1088815/peak-detector-cs0951017

https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/1012676/peak-detector

Kai

Thanks for Kai and Michael's reply.

I modify my schematic diode with a SMP1331-040LF, which is an RF diode with very small capacitance. Is it necessary? Or is BAT17 enough?

And changed the sampling caps to 100p series and 22R as you suggested in your other post.

But seems still have some attenuation on it?  I think the slew rate and bandwidth of THS4215's is enough for this signal. Is LTC6268IS8#TRPBF a better choose? 

Another question for it, 

My signal is about 1~1.4V. I can only measure the high voltage (1.4V) of this signal with this peak detect cirucit. How to measure 1v here?

My idea is the stucture below, AC filter first so I can get the vp voltage for the signal 

High voltage  = peak detect voltage 

LOW voltage =  peak detect voltage  - 2vp ( AC filter)

Idea the output should be about 400mV or below coz the attenuation, but it's turn out 700mV.

Thanks,

Frank

Hello Frank,

Are you still in need of help with this thread?  I am reading over it now so I can become familiar with the discussion so far.

Best,

Alec

Hi Alec,

yes, I still need some guidance if possible.

Thanks,

Frank

Hi Frank,

have you already checked Michael's circuit? It may be the better choice.

Kai