Amplification of the signal decay

Thread Starter

SAM-Mo

Joined Jan 19, 2019
38
I should amplify a low impedance signal with the peak of maximum 1.4V.
The tricky part is that the signal itself is not important but the tail of the signal is important which might show some decay (falling edge could reach to zero soon or later than normal) and these signal decay variations are very small (a few microvolts)

The thing is if I amplify the whole signal by a typical opamp, it easily sticks to the supply rails and clipped. because the signal itself is not small but its decay is very small.

The application is the battery powered and the single supply design is preferred.

is there any way to just amplify the variations?
 

ericgibbs

Joined Jan 29, 2010
10,201
hi SAM,
Welcome to AAC.
Do you have more details of the signal and the falling edge that you want to amplify.?
Frequency, Amplitude, decay time etc..
E
 

Thread Starter

SAM-Mo

Joined Jan 19, 2019
38
Sure

The maximum frequency is 1KHz. I think a picture helps more

The yellow rectangle is the tail which might show some very small decay change, but if I amplify that p[art or the whole signal, it will stick to the rail or if I amplify with the low gain, then the variation would not be sensible.
 

Attachments

danadak

Joined Mar 10, 2018
4,057
At uV levels your biggest enemy is noise. CDS (Correlated Double Sampling) is one approach to get you there.

http://www.cypress.com/documentation/application-notes/an66444-psoc-3-and-psoc-5lp-correlated-double-sampling-reduce-offset

http://www.cypress.com/file/119666/download


Another thought is digitizing and post processing. Do you have a spec for Latency,
eg. time you get the "tail" of the signal to the point in time you need post processing
finished. And what are you doing with signal once you get it ? What testing, evaluation
is being performed on the signal ?


Regards, Dana.
 

Thread Starter

SAM-Mo

Joined Jan 19, 2019
38
hi,
I suggest you clip the signal before amplifying the signal back edge.
E
Well. no. because still that yellow part has a few hundreds of mV amplitude (or at least much higher than microvolts range) and it will stick to the rails because I have to select a high gain for the microvolts variations.
if I select a low gain to stay below the rails, then the decay variations are a bit stronger but still very low.
 

Thread Starter

SAM-Mo

Joined Jan 19, 2019
38
sorry I don't have the circuit and oscilloscope in front of me to mark it up with the knobs and take a pictore.
the signal does not have a negative part. any shift is just an offset or oscilloscope is set to AC.

Time setting is 20uS. therefore each screen square means 20uS.
Volt division is on 500mV. each square means 500mV.
 

ericgibbs

Joined Jan 29, 2010
10,201
hi,
OK, I will wait until you scope the signal and post the image with the values marked on it.
The method I posted in post #11 could work, depending upon the signal parameters.
E
 

Thread Starter

SAM-Mo

Joined Jan 19, 2019
38
hi,
OK, I will wait until you scope the signal and post the image with the values marked on it.
The method I posted in post #11 could work, depending upon the signal parameters.
E
Actually your markups are almost correct. except that there is no negative part. as you see if I set the gain to 2000, with the supply of 12V, any amplitude above 6mV will be clipped to the rail (in case of a rail to rail opamp)
 

Thread Starter

SAM-Mo

Joined Jan 19, 2019
38
The oscilloscope can not save an image with knobs activated.

I tell you. The orange level you pointed as 0V is -700mV.

the 1.5V in your markup is 800mV.

The middle line is 0V.

The signal has some DC offset but I pass it through a capacitor and read its AC value. naturally, it does not have a negative part because there is no negative supply.
 

ericgibbs

Joined Jan 29, 2010
10,201
I tell you. The orange level you pointed as 0V is -700mV.
the 1.5V in your markup is 800mV.
The middle line is 0V.
hi,
You said in post #12 there is no negative part.???
So my original marked up image is close enough as a test signal.?
If yes, the circuit in post #9 is one option.
E
 

Thread Starter

SAM-Mo

Joined Jan 19, 2019
38
Hi,

Your circuit at the first stage is a buffer and the second has a gain of 10, how could it amplify the microvolt variations?
We discussed that we can use a low gain to avoid the saturation, but it is not enough to make the very small changes more strong.
 

ericgibbs

Joined Jan 29, 2010
10,201
hi,
The area of the decay curve you boxed in the image is approx 750mV, so I limited the gain to 10.
The gain could be increased to the level that the output limits to maximum swing of the OPA, say using a 12V supply.
E
 

Thread Starter

SAM-Mo

Joined Jan 19, 2019
38
hi,
The area of the decay curve you boxed in the image is approx 750mV, so I limited the gain to 10.
The gain could be increased to the level that the output limits to maximum swing of the OPA, say using a 12V supply.
E
So I mentioned from the begining that the tricky part of this problem is what you said now. Low gain is not useful and high gain sticks to the rails, therefore these methods can not be applied.
 

bogosort

Joined Sep 24, 2011
465
Sure

The maximum frequency is 1KHz. I think a picture helps more

The yellow rectangle is the tail which might show some very small decay change, but if I amplify that p[art or the whole signal, it will stick to the rail or if I amplify with the low gain, then the variation would not be sensible.
This sounds like an application for what the audio world calls a compressor: https://en.wikipedia.org/wiki/Dynamic_range_compression

In short, a compressor reduces the dynamic range of a signal by amplifying the signal only when the signal's amplitude falls below some threshold value. In other words, it does nothing to the loud parts, but the quiet parts get amplified. As your signal is in the kHz range and well-within standard audio voltages, an audio compressor would work well. There are tons of analog units to choose from, though they tend to be pricey. If you can digitize the signal, many free audio editing apps come with a compressor effect.
 
Top