Based off the example on the H11F1M Photo FET Datasheet , I'm creating Automatic Gain Control with the PhotoFET and TL072 Opamp. This AGC circuit works successfully by itself, but when I try to have two AGC circuits in series, they effect one another and don't reach their full gain potential. I've tried adding a buffer in-between, as seen in the picture, which results in still some audio getting through, but AGC not working at all. I've tried many different resistor values for the AGC feedback. Wondering if there is a different way to decouple these two circuits and keep them from influencing one another. Or my values for the resistors around the OpAmps are wrong...

Also open to any suggestions on noise reduction. There are filters on the digital waveforms coming into the photoFET's, and SSR's(used for audio-switching) but its still pretty noisy on the output. I've already implemented separate digital and analogue grounds.

 

? The power rails for the Op Amps are not specified - making the +4.5V reference point questionable . . . as related to valid common mode operation range for a given supply . . . some about in https://forum.allaboutcircuits.com/...et-of-opamp-tl072-as-voltage-follower.164915/

 

You can get 70db of attenuation in one stage. Why do you need two stages?

It looks like the supply is 0V and 9V. Is the 4.5V well connected to 0V with capacitors? How do you make 4.5V?

C11, output capacitor, Many audio gear is 10k input impedance. Have you look at 0.1uF & 10k for low frequency roll off?

 

You can get 70db of attenuation in one stage. Why do you need two stages?

It looks like the supply is 0V and 9V. Is the 4.5V well connected to 0V with capacitors? How do you make 4.5V?

C11, output capacitor, Many audio gear is 10k input impedance. Have you look at 0.1uF & 10k for low frequency roll off?

Its for a Tremolo guitar pedal, where the first stage is being controlled by the Trem's faster waveform, and the second stage is controlling the overall volume of the pedal. So both needing distinct AGC circuits that are being controlled by 2 different sources.

Sry, yes, a 9V supply. Using Voltage divider with multiple capacitors for the 4.5V.

Usually 10K input but some use higher pull down resistors to fight pops on the output as well.. Could do some more low pass filtering...

 

? The power rails for the Op Amps are not specified - making the +4.5V reference point questionable . . . as related to valid common mode operation range for a given supply . . . some about in https://forum.allaboutcircuits.com/...et-of-opamp-tl072-as-voltage-follower.164915/

Thanks for sharing that. Yeah there was a 10K pull-down resistor on pg29 of the data sheet for Unity Gain amplifier.. will try that out.

 

? the magnitude of the R5 , R8
? the signal in between C2 , R10 , C3 , C10
? the outputs of the U1A , U2B "colliding"
? C6 , C12 set up a frequency dependence of the output gain - is that the intention (i'm not familiar with all fancy audio effects)

Thanks for sharing

Actually it's the "Recommended Operating Conditions: TL07xH // all except" ~ Vss+1.5V ... Vdd // Vss+4V ... Vdd.
In practice simulation the "all except" for total supply of 9V it's (Vss+300mV a macro model) Vss+2.1V for an equivalent transistor model.
the min output load resistance from the d/s shows out @ Fig. 6-45 --OR-- Fig. 6-36 (+25°C) ≥ (±15V / ±26mA ≈) 576Ω (( for an optimistic full-scale swing ))

 

I think I entered your circuit into SPICE correctly. I tried to keep the ref # the same. Resistor H11F1 and SSR are not really resistors. Look at the blue trace. It shows the frequency response of the input buffer. The low end is -3db at 20hz like you planed. The green trace shows the response of the entire circuit. See if you can see why. Gain=-30db at 100hz.


Changed color to see better. edited-------------

 

some visualizations • input induced • output induced -- such are more likely to become an issue with the unity gain amp (a non inverting configuration with the 1x gain)
Edit : the dif-amp case

 

The value of R5 is only 1k ohms and kills the low frequencies fed to it through C2 and C3.

C2 and C3 should not be used with a SSR because the DC voltage at each end of the SSR will be the same.

 

What is "5V_Waveform"? Are you really controlling both FETs with the same signal? Is it derived from "AudiUT"? If not, then how do you expect an AGC circuit to function?

You have pops in the output because C2 and C10 are unbiased when the SSRs are off, and both charge up or down through R10 when the SSR's come on.

Between U? and U? you have three hipass filters and two lowpass filters. Each of these introduces a phase shift. When the two signal paths combine at R10, they no longer will be in perfect phase throughout the audio bandwidth.

ak

 

when I try to have two AGC circuits in series, they effect one another and don't reach their full gain potential.

That's because the output of U2B gets fed back to the input of U3A via C10, the SSRs and C3.

Edit:
I don't see the need for the first SSR?

 

I think I entered your circuit into SPICE correctly. I tried to keep the ref # the same. Resistor H11F1 and SSR are not really resistors. Look at the blue trace. It shows the frequency response of the input buffer. The low end is -3db at 20hz like you planed. The green trace shows the response of the entire circuit. See if you can see why. Gain=-30db at 100hz.

Thanks for drawing this out. I see what you're saying with changing the resistors to 10K to move the filter pole down

 

LT SPICE is free and will help you. There are other free SPICE programs.
With 10k resistor.

If you choose to use SPICE there is probably no part for the H11f and the opto-switch.
For the switch, there is a switch that you can use. You can set the on/off voltage. I think on>1V and off<1V. On resistance=50 ohms, OFF resistance= 10meg. That is pretty close.
For the H11F1, we could make a part but for a start I would just use a resistor, then when you get good at SPICE change to a resistor that is voltage programmable.
ronsimpson

 

If you'd like an LTspice-compatible H11F1M model, here's one I made.

 

Updated schematic. So I changed the R5 and R8 to change the RC filter.
Added the missing C13 before R10.
Added R15 to the buffer, and that increased the gain a lot. Having trouble finding a balance in the resistors and gain, but clearly need to use Spice a little more. Thanks for the help, and Spice models.
View attachment 263579

 

(image wasn't working)

 

What is "5V_Waveform"? Are you really controlling both FETs with the same signal? Is it derived from "AudiUT"? If not, then how do you expect an AGC circuit to function?

You have pops in the output because C2 and C10 are unbiased when the SSRs are off, and both charge up or down through R10 when the SSR's come on.

Between U? and U? you have three hipass filters and two lowpass filters. Each of these introduces a phase shift. When the two signal paths combine at R10, they no longer will be in perfect phase throughout the audio bandwidth.

ak

One of the signals is a PWM waveform, and the other is technically 5V controlled by a digipot.

Added C13 before R10... Should SSR's be biased? From what i've gathered they shouldn't have DC on them at all. Thought the caps are necessary, but @Audioguru again also said they shouldn't be there. if you could explain, or reference something for this.

 

? the magnitude of the R5 , R8
? the signal in between C2 , R10 , C3 , C10
? the outputs of the U1A , U2B "colliding"
? C6 , C12 set up a frequency dependence of the output gain - is that the intention (i'm not familiar with all fancy audio effects)

Actually it's the "Recommended Operating Conditions: TL07xH // all except" ~ Vss+1.5V ... Vdd // Vss+4V ... Vdd.
In practice simulation the "all except" for total supply of 9V it's (Vss+300mV a macro model) Vss+2.1V for an equivalent transistor model.
the min output load resistance from the d/s shows out @ Fig. 6-45 --OR-- Fig. 6-36 (+25°C) ≥ (±15V / ±26mA ≈) 576Ω (( for an optimistic full-scale swing ))

So maybe a 500ohm resistor for R15 would be better? Appreciate those visualization links, though i didn't really understand them...

 

those visualization

The visualizations show the limitations of TL07x , TL08x family Op Amps - e.g.- they do not have full scale output swing nor is their input rail-to rail.
since i don't know your AC amplitudes and DC offsets at various stages and

when I try to have two AGC circuits in series

hints that your signal may exceed the "linear" range at some point

usually you define your max signal level - say ±500mV - and attenuate it and provide the required amplification at the last stage
( Contrary - if some fancy sound effects require driving the Op Amp to it's max. output values - you must make sure your I/O input and output signal grounds are residing correspondingly at correct DC offsets )

if you just need a volume control :
dedicated chips (not covered in this list)
https://www.google.com/search?q=linear+control+vga+op+amp
https://www.radiolocman.com/shem/schematics.html?di=580299
https://www.edn.com/agc-circuit-uses-an-analog-multiplier/

 

C1&R2, ...C3, C4, C5, C13 Five times you have a filter that blocks low frequencies. R is 10k or 82k. The 82k = about 20hz. The two 10k is about 350hz. Math in my head.
One RC has 20hz and you have three then the filter is down 3x what you think.
Then there is two more filters at 350-hz.
There is no low end.