Simplest AGC for op amp

[^saint^]

Hello
my current project uses a sensor that that produces a sine wave (DC...~5kHz) with varying amplitude (max Vp-p ~60mW). Noise from the sensor is about 3mV max. and output offset voltage is -0.2V to +0.2V
Current schematic:

What is the purpose of C10?

In the schematic the U1 has a fixed gain and U2 uses a pot to adjust the gain. What I would like to do is use a simple AGC (probably FET as VCR) and vary gain of U1 (x10-100) 20-40dB and keep U2 gain at 40dB.
The output signal from U2 is then converted to a square wave (simple comparator with histeresis, High Threshold: 3.2V, Low Threshold:2.8V, wave is centered at VCC/2) to count the frequancy of the wave so I only need to avoid distorting the frequency.

Been reading up on FET based VCR for AGC, I'm particularly interested in simple circuits like AN-32 FET Circuit Applications (2N3685 on page 6) or this one VARIABLE GAIN OP-AMP CIRCUIT. The problem is that most of them don't give any values or specify the FET (N-channel JFET, right?) plus FETs characteristics are inconsistent from device to device.
This topic JFET as variable resistor was quite helpful but didn't clarify all.

Hopefully someone can point in the right direction for choosing a particular FET. Most of the datasheets only specify Rds(on) and don't mention Rds(off)?

Or is there a simpler way to implement AGC for an op amp?

 

[Audioguru]

What is the purpose of C10?

It filters the +2.5V reference voltage for the opamps so they do not amplify power supply noise.

In the schematic the U1 has a fixed gain and U2 uses a pot to adjust the gain. What I would like to do is use a simple AGC (probably FET as VCR) and vary gain of U1 (x10-100) 20-40dB and keep U2 gain at 40dB.
The output signal from U2 is then converted to a square wave (simple comparator with histeresis, High Threshold: 3.2V, Low Threshold:2.8V, wave is centered at VCC/2) to count the frequancy of the wave so I only need to avoid distorting the frequency.

Why do you need AGC? Why do you need a comparator?
I do not know why your frequency measuring circuit cannot measure the frequency at the output of the second opamp. When the signal level is high then the output of the second opamp will be a squarewave. If the gain of the two opamps is high enough when the signal level s low then the output level of the second opamp should be high enough for your frequency measuring circuit to measure the frequency.

The 100pF capacitor parallel with the 1M negative feedback resistor cuts frequencies above 1.6kHz in both opamps so 1.6kHz will have a level of half the level of lower frequencies and higher frequencies will have their level reduced to half for each higher octave. 3.2kHz will have 1/4 and 5.4kHz will have 1/8th the level of much lower frequencies.

 

[^saint^]

Why do you need AGC? Why do you need a comparator?

Because actually my module (**broken link removed** radar detector) has two outputs, when sine wave is present one is leading 90 deg. I need to know which plan I also need to know the frequency (both waves are identical just shifted 90 deg). Here's where I'm at > link based on similar designs and this **broken link removed**

I do not know why your frequency measuring circuit cannot measure the frequency at the output of the second opamp. When the signal level is high then the output of the second opamp will be a squarewave. If the gain of the two opamps is high enough when the signal level s low then the output level of the second opamp should be high enough for your frequency measuring circuit to measure the frequency.

Amplified noise will trigger the comparator when no sine wave is actually present.

The 100pF capacitor parallel with the 1M negative feedback resistor cuts frequencies above 1.6kHz in both opamps so 1.6kHz will have a level of half the level of lower frequencies and higher frequencies will have their level reduced to half for each higher octave. 3.2kHz will have 1/4 and 5.4kHz will have 1/8th the level of much lower frequencies.

They will be changed for the final circuit, for now I'm well below the upper limit.

 

[FvM]

Applying a simplified quadratic model, the FET characteristic is fully defined by just two parameters Idss and Vgsoff. For a specific FET type, you also observe a typical relation between both parameters so that the parameter distribution is one-dimensional in a first order. The variation is still quite large, even for FETs with Idss respectively Vgsoff groups like BF245A/B/C.

For a single channel AGC amplifier this shouldn't be a problem because the control amplifier feedback compensates the type variation. If you e.g. intend a dual channel AGC circuit with gain tracking, the matter becomes difficult. Monolithic dual FETs are a solution, but they are rarely available and quite expensive these days.

For your radar detector that doesn't impose particular linearity requirements, a BJT as variable resistor could be sufficient. See below a simple circuit of a microphone amplifier AGC to illustrate the idea.

 

[Audioguru]

AGC does the opposite to what you want.
With a low signal level or no signal then the AGC gives full gain or even increased gain which amplifies noise. When a signal is present then the AGC reduces the gain which keeps the output level fairly constant even if the input level changes.

 

[FvM]

AGC does the opposite to what you want.
With a low signal level or no signal then the AGC gives full gain or even increased gain which amplifies noise.

At least the AGC should be supplemented by squelch-like function that blocks signal detection below a certain threshold level.

But as I understand the OP wants a kind of constant fraction trigger for variable doppler signal levels, this can be in fact achieved by an AGC circuit. I wonder however if a simple zero crossing detector preceeded by a fixed gain AC amplifier can achieve the intended function.