Hello...

I'm posting the following 2 peak detectors to try to understand why the first one is acting so weird!
They are taken from AoE from Paul Horowitz & Winfield Hill 3rd Edition, page 254, pictures 4.58A and 4.58B.

If needed I can take a picture and post them here!

Book says that version A of the circuit deals with the diode Vdrop because of the feedback from the capacitor voltage comparing with most basic peak detector topology that uses only a diode in series with a capacitor and Vout is measured between diode and capacitor.

The problem is that this A version of the circuit has a very weird wave form at the output. Why is this happening?
A 'B' version is an improved version of A that can deal with short peaks because it doesn't have to slew from negative saturation.

 

The circuits in the attached .asc files do not match the circuits on the waveform image.

 

I'm sorry. Probably I messed up uploading the .asc files but what I want to refer is the image circuits.

 

You are using a 10MHz input. That is very high for common op-amps. What is the GBW of the amplifiers?

 

Why is this happening?

Perhaps because the only discharge path for the cap is diode reverse leakage or opamp input bias current.

 

The germanium diodes have a large leak (especially at elevated temperature). If you need a fast diode with an acceptable leak, then take the Schottky diode BAS70.

 

Your simulation is using the generic op-amp. Do you know what the model is for that op-amp? I don't.

I suggest choosing a real op-amp before you try a detailed analysis.

For example, some, high speed op-amps have clamp diodes between the input pins. The clamp diodes create a discharge path from the input to the peak-holding cap.

 

Start with a low frequency signal, like 1kHz.
Work up to higher frequencies once you get the lower frequencies to work.

Generic op amps may have an open circuit output voltage limited only by the dynamic limit of the simulator, which can give strange results.
Use a model for a real op amp, as noted.

 

Thank you all for replying.

I tried lower frequencies as suggested and the results were way better but I still don't understand the behaviour of left circuit. It should hold the peak but it looks like it does absolutely nothing compared to the input signal waveform.

I tested 1Khz, 10Khz and 100Khz. The last frequency already revealed serious distortion/non-linearity for the circuit of the left.
So I keep wondering about the book words! This left circuit should be more or less like the one on the right but with lower peak value detected due to Vdrop.

Ok, so to start fixing the circuit, will BAS70 OpAmp be suitable for higher frequencies such as 10Mhz or 20Mhz??




I've also included the correct .asc file!

Edited;
I think someone asked about the parameters that rules these OpAmp models I used.
The parameters I can see are as follows:
AoL = 100k
GBW = 10Meg

 

Adding to previous post, I tested a BAS70H diode model from LTSpice library for the same frequencies of 1Khz, 10Khz and 100Khz. Results for circuit on the left were not very promising.

I did a last test for an higher frequency of 10Mhz. The circuit on the left, after settling, the output drops to 0V.

 

Ok, now for a more real opamp scenario, which one should I choose to be able to handle higher speeds?

 

Use GBW = 500Meg.
You use an ideal operational amplifier, which has no limitations on the output voltage.Look at the output of the first (left) op-amp and you will understand what's the matter.Your amplifier brings to the breakdown of the rectifier diode and as a result you have not a peak detector, but a voltage follower.

 

Use GBW = 500Meg.
You use an ideal operational amplifier, which has no limitations on the output voltage.Look at the output of the first (left) op-amp and you will understand what's the matter.Your amplifier brings to the breakdown of the rectifier diode and as a result you have not a peak detector, but a voltage follower.

With your suggestion:

 

So did you see what happens at the output of the first op-amp? In a real amplifier, the minimum output voltage can not be lower than the voltage of the negative source. Take the parametric model of an operational amplifier with power pins! You do not understand What I wrote about. Your first circuit is not a peak detector, because you can not bring the diode to the breakdown. Recover an ideal diode model that does not have a breakdown voltage parameter. In reality, the first scheme is bad because the operational amplifier will become saturated! From saturation the amplifier can go out for a long time! The second scheme does not have this bad regime. By the way, Models of some operational amplifiers may not simulate the saturation (or overload) mode.

 

So did you see what happens at the output of the first op-amp? In a real amplifier, the minimum output voltage can not be lower than the voltage of the negative source. Take the parametric model of an operational amplifier with power pins! You do not understand What I wrote about. Your first circuit is not a peak detector, because you can not bring the diode to the breakdown. Recover an ideal diode model that does not have a breakdown voltage parameter. In reality, the first scheme is bad because the operational amplifier will become saturated! From saturation the amplifier can go out for a long time! The second scheme does not have this bad regime. By the way, Models of some operational amplifiers may not simulate the saturation (or overload) mode.

Ah ok. I missed the output of the first OpAmp. It goes way down to -120V or so.
But the AoE book presents the circuit on the left as a peak detector, or better, a peak tracker.


Edited;
And I've just noticed I'm missing one resistor on the circuit of the right!

 

the AoE book presents the circuit on the left as a peak detector, or better, a peak tracker.

That circuit will track the highest peak and stay there until the capacitor is discharged or the leakage currents causes the output to drift from the peak value.

What behavior do you expect form a "peak tracker"?

 

That circuit will track the highest peak and stay there until the capacitor is discharged or the leakage currents causes the output to drift from the peak value.

What behavior do you expect form a "peak tracker"?

Yes but on the circuit of the left, what makes the capacitor to discharge? It's just leakages?

I found another version for a peak detector

 

Yes but on the circuit of the left, what makes the capacitor to discharge? It's just leakages?

I found another version for a peak detector
View attachment 156620

This is no longer a leak. This is the current breakdown. Do not you know that in the case of a breakdown in a real diode, the current is much more leaky. Some radio amateurs use conventional silicon diodes (with a fairly small leakage) as zener diodes for a voltage of 100-300 volts. This was done in the USSR.

 

I found another version for a peak detector

Apart from the buffering action of U1, that circuit would give the same results as omitting D2 and D3 and using just D1 and C1. D2 and D3 serve no useful purpose that I can see.

 

Apart from the buffering action of U1, that circuit would give the same results as omitting D2 and D3 and using just D1 and C1. D2 and D3 serve no useful purpose that I can see.

D2 stops the amplifier saturating on negative inputs and eliminates the recovery time from that saturation.
D3, at least partially, offsets the voltage drop of D1.