Help finding ADC

[ronv]

What op amp are you using?
For the discharge fet use a real low gate charge. The 100 ohm could be in the drain and a little gate resistance might help. I think the gate/drain capacitance is putting a little bump on the end of the sample. Not sure it should matter unless it starts things over.

 

[ImperfectSeven]

I'm using a TLC27M7IP
The MOSFET is an SI2302ADS
it has a gate charge of 4nC typically and 10nC max, should I go even lower?

What do you mean by "starts things over"?

 

[ronv]

I can't find that ic, but here is a simulation with a coupe of changes:

I used OP743 fairly fast and low bias current.
Changed the diodes to BAT54 schottky
Made the cap a little smaller to charge faster.
Moved the first diode to the other side of the second diode. This gives a little overshoot, but maybe not to bad.

I think I see your second pulse now - after the reset?
I'll look some more later

 

[ImperfectSeven]

Ok, so first question is how do you get LTSpice to show the different signals on different graphs?

I too noticed that if I changed D1 ti a schottky diode that the extra pulse went away, but you replaced it and flipped it. Why?
(Picture showing circuit with schottky and no bump on the output of the first opamp)
View attachment 66070

Ya the second pulse occurs after the reset (falling edge of the disc signal.) in simulation this second pulse doesn't really affect thing, but when in a circuit this second pulse becomes larger than the first thus becoming the peak of the signal.

 

[ImperfectSeven]

In case I was not clear before here is another picture:
View attachment 66071

In the top circuit a Schottky diode has replaced the normal diode and as a result the the triangular pulse that occurs on the falling edge of the switch signal has been eliminated.
Is there a reason as to why this is the case?
Is there a way to eliminate the overshoot I am seeing between OUTPUT and the peak of INPUT?

 

[ronv]

Rev 2

OK, It takes so little into that high impedance circuit. I added a diode and resistor to keep the gate drain capacitance from charging up the little cap. Seems to take care of it. The diode in that direction keeps the 1st op amp from overshooting so bad until the second op amp can respond.
To add a plot pane right click in the waveform picture then add plot pane.
Hope the reset can be that long.
Not sure how well this will work in real life.

 

[ImperfectSeven]

If the reset time really is about 15μs then yes it is too slow. I need to keep it around 5μs, lower if possible.
Why won't this work in real life? <--- That is the question I really need answered because it already isn't working in the real life =(

 

[ronsimpson]

Post your schematic.

 

[bountyhunter]

Hi there, I am trying to build a peak and hold circuit that will capture the peaks of pulses that are about 5μs in duration or shorter.

That's actually a tough spec to meet. You don't specifiy the amplitude of the pulses?

I built a peak detector to extend the bandwidth of my meter to beyond 200kHz. It's designed to output negative voltages but you can flip the diodes and make it positive output.

The LF411 is a high speed buffer. The next two amps make up the peak detector.

 

[ronsimpson]

Some reading

 

[ImperfectSeven]

Here is my LTSpice simulation:
View attachment 66101

The problem I am having is on V(op1out) where there is a triangular pulse that occurs on each falling edge of V(switch). In simulation this isn't a problem, but when I actually build the circuit this bump is larger than my signal pulse and thus the peak is based off of it instead of my input pulse.

 

[ImperfectSeven]

That's actually a tough spec to meet. You don't specifiy the amplitude of the pulses?

Actually I have mentioned the amplitude of the pulses in the same paragraph you quoted me in =)
To prevent you from having to go back and look here it is again:

The amplitude also varies and is normally in the 1-50mV range but I am amplifying it using the PSoC so it is about 50mV - 900mV.

How much time does it take for your peak detector to reset? Like I said earlier I need a reset time of about 5μs or shorter, yours looks to be quite a bit larger than that.

Ronsimpson,
Thanks for the reading, I now know that there are a lot of ways to do peak detection however I don't know the differences between them or better yet which one I should use for this application. It seems most examples I find are regarding audio signals which would be great if I were doing audio signals. The circuit I am using was titled "High speed peak detector" so I'm hoping it will do a better job of fitting the bill, there are just some issues with it that I do not know how to correct =/

 

[ronsimpson]

Last year I make a circuit that needed a very fast low capacitance diode, much like your circuit.
I started out with the 1N4148 knowing it would not work. The I change to a BAT54 (?) and at the end I used a HSMS-286. This part and parts in the same family are used in test equipment made by Agilent. I needed less capacitance and less reverse recovery time. I really think you need better diodes. The data sheet for the HSMS-286 has the spice model numbers.

You have a 20k resistor form the output back to the input of the first amp. I think the RC low pass filter made by 20k and the capacitance of the amplifier is a problem. What happens if you change the resistor to 2k?

I believe the slew rate of the first amplifier is a problem. Spice does not model how a amplifier works when it output is at its power supply rail. The output of a amplifier gets "sticky" when it is at the supply voltage. I think spice does not show that but in the real world this is a big problem with your first amplifier.

I need to go work on my circuits now.

 

[ronv]

Rev 1.1.1

Here is one I think you will like. Couple of things:
The first op amp needs to be fast.
The diode needs to have low capacitance and fast recovery.
Got rid of the fet and replaced with a Tx to eliminate the blip after reset.
I'll post the asc. file so you can see the effects of different diodes etc.

 

[crutschow]

You might use a high speed analog peak detector such as this to capture the short ns pulses.