Constant Gain

Imagine a sine wave that varies between 1v p/p and 100v p/p. Now full wave rectified it without filtering. You end up with a waveform consisting of both 1/2 cycles in the positive polarity (not DC) that varies between 1v peek and 100v peek. This is the "input" signal. Regardless of the input voltage between 1v peek and 100v peek, I need to "output" a 5v peek waveform. I need some kind of AGC. A Zener clipper won't work because I need to preserve the sine wave. I'm thinking an Op-Amp with a JFET voltage to resistance circuit in the feedback loop. Since both cycles of the input are being rectified positive, the op amp's -V supply can be GND (no need for + & - power supply for op amp, only +). So in summary, I have a full wave rectified unfiltered sine wave that varies between 1v and 100v peek and I want to convert this to a constant 5v peek. Any help would be appreciated.

AGC; look for apps. at thatcorp.com
Design Notes for THAT ICs

DrWhoWho said:

Imagine a sine wave that varies between 1v p/p and 100v p/p. Now full wave rectified it without filtering. You end up with a waveform consisting of both 1/2 cycles in the positive polarity (not DC) that varies between 1v peek and 100v peek.

Click to expand...

It's DC!

This is the "input" signal. Regardless of the input voltage between 1v peek and 100v peek, I need to "output" a 5v peek waveform.

Click to expand...

The output can't possibly be caused by constant gain!

I need some kind of AGC. A Zener clipper won't work because I need to preserve the sine wave. I'm thinking an Op-Amp with a JFET voltage to resistance circuit in the feedback loop. Since both cycles of the input are being rectified positive, the op amp's -V supply can be GND (no need for + & - power supply for op amp, only +). So in summary, I have a full wave rectified unfiltered sine wave that varies between 1v and 100v peek and I want to convert this to a constant 5v peek. Any help would be appreciated.

Click to expand...

You might use an OPA547. It is designed for 50V a single supply of 8 to 60V.

Additionally you must use a voltage divider not to exceed the maximum input voltage.

There are other high voltage input OpAmps on the market but they cost real $$$!

Boncuk

It's DC!

Click to expand...

Surely you understand that I mean unregulated DC. It's still a sine wave with both 1/2 cycles having the same polarity.

The output can't possibly be caused by constant gain!

Click to expand...

If I divide the input by 10 with a voltage divider, I end up with .1v peek to 10v peek, so indeed it is possible to create without ever going to a high voltage op amp. I'm thinking on the lines of dividing the input by 10 which is input to an op amp that has a UJT as a voltage controlled resistor in the feedback loop.

DrWhoWho said:

Surely you understand that I mean unregulated DC. It's still a sine wave with both 1/2 cycles having the same polarity.



If I divide the input by 10 with a voltage divider, I end up with .1v peek to 10v peek, so indeed it is possible to create without ever going to a high voltage op amp. I'm thinking on the lines of dividing the input by 10 which is input to an op amp that has a UJT as a voltage controlled resistor in the feedback loop.

Click to expand...

hi Who
Whats the frequency range of the 'original theoretical sine wave'..?

Post your UJT circuit.

We are confused!
you have 1/2 sign waves all +.
I think you want the output to be the same voltage if the input is 1 volt or 10 volts.
If you could wait 2 or more time periods to get the output level to respond then this is easy. It the first pulse must be at the right level that is hard because you don't know its level until it is finished but the gain must be set correctly at its start.

ericgibbs said:

hi Who
Whats the frequency range of the 'original theoretical sine wave'..?

Post your UJT circuit.

Click to expand...

For simplicity, 60hz and I said UJT but meant FET.

I'll try to attach the waveform conversion I'm trying to achieve along with an AGC circuit using the JFET.

DrWhoWho said:

For simplicity, 60hz and I said UJT but meant FET.

I'll try to attach the waveform conversion I'm trying to achieve along with an AGC circuit using the JFET.

Click to expand...

hi,
OK, 60Hz.
The problem will be the response time of the 'agc' circuit. If the 0.1V to 10V changes rapidly the 5V will not be constant.

What response time will be acceptable.?

ericgibbs said:

hi,
OK, 60Hz.
The problem will be the response time of the 'agc' circuit. If the 0.1V to 10V changes rapidly the 5V will not be constant.

What response time will be acceptable.?

Click to expand...

I did not design this circuit, but the response time is 1.2ms which is good enough for my application. I really don't like this design because it uses an op amp, transistor, and JFET. I'm really looking for something more simple. Seems a bit much to keep an output at 5 v with an input signal varying from .1 to 10 v.

I would first look at fixing the problem at the source. Why would you need to keep the "sine" shape of the waveform but completely ignore its change in amplitude?

Is this for a synchronisation or measurement system? If you explain exactly what the input waveform is from, and what you need to measure or test then we can help you by suggesting the best way to go about it.

DrWhoWho said:

Surely you understand that I mean unregulated DC. It's still a sine wave with both 1/2 cycles having the same polarity.

Click to expand...

This is called pulsating DC, doesn't matter if its regulated or not, it is not a sinewave. Maybe you can utilize an OTA (Operational Transconductance Amplifier) to make a variable gain amplifier. I have some CA3094E OTA's that you can have for free if you like.

Jaguarjoe said:

This is called pulsating DC, doesn't matter if its regulated or not, it is not a sinewave. Maybe you can utilize an OTA (Operational Transconductance Amplifier) to make a variable gain amplifier. I have some CA3094E OTA's that you can have for free if you like.

Click to expand...

OK, yes it is a sine wave its just not alternating in current every half cycle. Sine is a mathematical function that describes the "shape" of the wave. That's like saying a square wave isn't a square wave, it pulsating DC. The "shape" of the wave is what determines if it is sine or square, not the polarity or direction of current. If the wave is rising and falling by a function of sine, than it is a sine wave. Pulsating DC can be viewed as any waveform such as a square wave, triangle wave, sawtooth wave, sine wave, etc. that is not alternating in current. They are "DC" because they the current is not alternating, but they are all "waves" be it sine, square, triangle, or sawtooth until you filter them. The only difference between the sine and other waves is that sine is not linear and the other are. If you have square wave that rises to +12 then falls to ground, then rises to -12 then falls to ground, then you end up with a 12v peek to peek "AC" square wave because the current is alternating.

So back on subject, I have figured how to achieve what I need. I simply created an OP with AGC via an LED/CDS cell opticoupler feedback circuit. Works like a charm. The nice thing about this method is the better linearity of the CDS cell and response time compared to the JFET. It also allows independent adjustment in the "sustain/release" in the feedback loop. Thanks for all the input, it was much appreciated.

DrWhoWho said:

The nice thing about this method is the better linearity of the CDS cell and response time compared to the JFET. It also allows independent adjustment in the "sustain/release" in the feedback loop.

Click to expand...

Better linearity than the JFET, I have to agree. Adjustable sustain/ release, yes also.
I'm surprised the 'attack' time for the LED/CDS was better than the JFET though. What combination did you use? NSL32, or similar?

rogs said:

Better linearity than the JFET, I have to agree. Adjustable sustain/ release, yes also.
I'm surprised the 'attack' time for the LED/CDS was better than the JFET though. What combination did you use? NSL32, or similar?

Click to expand...

Actually, I looked at many similar to the NSL32 but found that there weren't many available that gave me what I needed so I made my own. I used a high bright white 300 mcd LED and an LDR 07 type photoresistor or CDS cell. The photocell resistance is somewhere between 100 ohms and 10M Ohms and allows excellent sensitivity with that range when coupled with a high bright LED. Whether the attack time is better or not in the JFET, I don't know, but really it doesn't matter that much in my application. Its only important that I can control the times independently. I'm betting the other plus is the response time. A JFET as a voltage controlled resistor in an op amp feedback circuit is several ms because the gate is driven by the rectified signal and charge of a capacitor along with the JFET on/off time, where an LED's brightness to change in current is very fast. I don't know this for sure yet, but I will measure it.

Sounds like a clever solution for AGC.

What was it for?