ACG Control Loop Help

[321Mark]

This is my first time designing an AGC amp for an RF application. I'm using a commercial AGC amplifier and detector (Teledyne Cougar). The AGC amp has a 0-5V control signal, 0V is full gain. The detector has a 0-5 volt output. 0 volts occurs at -40dBm and 5 volts at +10 dBm. I want the output to be at -5dBm which related to a control voltage of about .5 volts.

I fed the detector output into the (-) input of an op-amp configured as an error integrator with a resistor and capacitor in series in the feedback loop (single pole, single zero). On the (+) input of the op-amp integrator I connected a reference voltage to set the operating level.

I followed the integrator stage with an inverting amp, A=1 to re-invert the signal and connected the output of the inverter to the AGC amp. I guess I didn’t know what to expect, but the loop does control the RF signal, but not like I thought it would. I see “pulses” coming from the detector and these are integrated into triangular waves by the integrator. These triangle waves feed the AGC amp. Looking at the RF level when the loop is operating, the level looks constant.

I though the loop would reach equilibrium and the output of the op-amp would be a steady DC level, but I guess because of the integrator it is constantly operating around some point.

Should the loop behave this way? Any explanations of insights would be helpful.

 

[Skogsgurra]

Sounds like you forgot the resistor in series with the detector signal. What you have now is more or less infinite gain. (G=Rfb/Rin). Rin = zero makes G infinite. Use a 10 kohm resistor between detector and opamp inverting input. Then tune feed back resistor (capacitor shorted) for fast and well damped response. Finally, adjust capacitor (starting with a large value, like 1 uF) to get good rsponse.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[321Mark]

Sorry, I have a 47K input resistor. This sets the low frequency gain and then you see the slope of the integrator. Works as described previously and oscillates with the triangular waveform.

Mark

 

[321Mark]

Text should read:

I fed the detector output through a 47k resistor into the (-) input of an op-amp configured as an error integrator with a resistor and capacitor in series in the feedback loop (single pole, single zero). On the (+) input of the op-amp integrator I connected a reference voltage to set the operating level.

 

[Skogsgurra]

OK. Then proceed as said above. You may also need to filter the output from the detector. Do that by dividing your input resistor in two halves and add a capacitor to ground to get a T-filter. You will probably need around 50 - 100 ms time constant in the filter.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[321Mark]

Thanks for your help so far...

When you explained to..."Then tune feed back resistor (capacitor shorted) for fast and well damped response"

Should I do this by switching the RF ON and OFF in a step-like manner and monitor the control voltage until I get a good response?

Also, is the oscillatory nature of this circuit normal?

Thanks

 

[Skogsgurra]

Switching on/off/on is not a good idea. The controller will usually not operate in its linear range when you do that.

You can simply increase gain until your signal gets unsteady. Then reduce to about half. Goes for P as well as I. When you think you are OK, then make a ten to twenty percent step amplitude change at about half power level and observe the step response. Some overshoot usually is OK, but no ringing.

No. An AGC should not be unstable. But it has to reflect the nature of the incoming HF. Is it AM modulated? If so, you need the filter I mentioned before.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[sreid]

1) Do pulses come out of the detector with a fixed AGC level?

2) Any chance you've got the loop connected with positive feedback so you've created an oscillator?

 

[321Mark]

Yes, pulses come out of the loop with a fixed AGC level. I think I have created a fast oscillator. Its definitely negative feedback - the detector is on the (-) input and my reference level (DC) in on the (+) input. Even though the loop is oscillating it controls the RF output and I can adjust the level.

I get nice triangle waves out of the intergrator. Maybe I have too much DC gain, or the AC gain is not dropping below unity at the crossover frequency.

I do have some questions though. The error amp feedback is AC (resistor and cap), should I also have another feedback resistor from (-) in to output to set the DC gain?

 

[Skogsgurra]

No DC from the detector? Then DO filter the signal. As said above.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[321Mark]

OK, I made some progress. One problem - as I would bring the RF level up on the AGC amp a point would be reached where I thought the AGC circuit was oscillating. Turns out that something between the AGC amp and detector module was causing the problem. Operating open loop, (no AGC)I would see nasty stuff coming from the detector at high RF levels. Looking at a wider span on the spec analyzer, I saw the 2nd harmonic rising up as the RF level increased. This was somehow reacting with the detector and causing the oscillation. The AGC amp feeds a 3dB divider and one side of the divider feeds the detector module. Placing a bandpass filter between the power divider and the detector to filter out the 2nd harmonic solved a lot of noise/oscillation issues.

After this, things worked better. Now I was able the raise the RF level even higher, but eventually I saw oscillations again. The error integrator is followed by an inverter, gain =1. While the circuit was oscillating, I scoped the output of the integrator - looked clean. Then I scoped the output of the inverter and IT was oscillating??? op-amps are OP07s, I havent checked, but I think they are unity gain stable. I placed caps across the feedback R, but that didn't help. Then I placed a cap after the inverter input resistor to GND, and that did it. No oscillations and the circuit seems to be operating well.

On Monday, I'll be investigating why the inverter was oscillating.

Thanks for all the help so far.

Mark