peak detector design

[Joe_Wu]

Hi, I am building a peak detector and voltage comparator for my returning signal of the ultrasonic transducer.
Below are the circuit diagram of peak detector and voltage comparator:


And below are the screenshot of oscilloscope after peak detector and voltage comparator:


As you can see, it kind of works, it detect the peak at the very beginning, however, what I am interested is the returning signal after the first pulse (the signal that appears in the middle of the screen), and my peak detector is not able to detect it.
Any idea about how to do it?
Thank you.

 

[barry]

If you look closely, you'll realize that the output of your (crummy) peak detector is actually a fraction of the actual peak value, maybe 50%. When that second peak comes along the cap has already been charged to a value greater than 50% of the new peak.

You might want to consider using an active (opamp) peak detector, and perhaps changing the time constant of the circuit.

But are you saying you don't want to see that first peak at all?

 

[smijesh]

Active peak deter is an option
http://cds.linear.com/docs/en/datasheet/lt0111.pdf

You try to reduce the capacitor value in peak detector (100pf to 10pf) so you can detect the missing peak also

 

[Joe_Wu]

yes, all I care about is the signal in the middle that looks like triangle. The first peak is always there, the signal is the middle is the reflection signal that I want. I calculated the time constant based on:
capacitor-resistor combination will establish the time constant ��, which should be kept to a value above the period of the main component (carrier, 5MHz), but much lower than the modulating signal (in this case dictated by the frequency of repetition 7.8kHz).

T1 = 1/5000000 = 0.2µs

T2 = 1/7800 = 128µs

�� = R*C = 100p*220k = 22µs

∴ T1 < &#55349;&#57103; < T2

 

[FvM]

For the simple exponential peak detector decay you are apparently intending, it should be simple to estimate the optimal time constant. It can be sketched into the waveform with any mathematics. Obviously the time constant should be smaller than that you selected.

But it's more involved to make an ideal or at least good working peak detector. The upper waveform shows that your simple diode detector is far from being ideal.

 

[barry]

First, you need a better peak detector. Second, you're going to need some logic to ignore the first peak.

 

[d123]

Hi,

Maybe useful?

 

[BradtheRad]

Bats have a similar problem as they only need to hear the echo return of their call. To solve this, the nerves to the bat's ears turn off at the moment its mouth emits a call. Then its ears turn on again. Your most effective solution might need to do the same.