Logarithmic Compressor and Decompressor circuits


 

Please see recent upload…

 

I started from basics. to create a simple log compressor for a limited range of input.

That seems to work well in an "ideal" simulation, using Univ OpAmp2.

Then I got inspired to test an equally simple DEcompressor by putting the circuit in the feedback path of another UniOpAmp2, which also worked (surprisingly) well.

 

Replacing the amps in the Compressor path with LT1012's appears to work just as desired, as well. But replacing the DEcompressor amps fails miserably.

 

What can I learn today?

 

Dave


 

Not to use costly LT1012s? Bur seriously, what happens if you put another real opamp as U3 and U4. I can't check, because I can't see how to change from the Uopamp2 to a real one. Do I have to delete the symbol from the schematic and put in a new one? I've never used Uopamp2.

On 2024-11-10 21:38, Bell, Dave via groups.io wrote:

Please see recent upload…

 

I started from basics. to create a simple log compressor for a limited range of input.

That seems to work well in an "ideal" simulation, using Univ OpAmp2.

Then I got inspired to test an equally simple DEcompressor by putting the circuit in the feedback path of another UniOpAmp2, which also worked (surprisingly) well.

 

Replacing the amps in the Compressor path with LT1012's appears to work just as desired, as well. But replacing the DEcompressor amps fails miserably.

 

What can I learn today?

 

Dave

-- 
OOO - Own Opinions Only
Best Wishes
John Woodgate
Keep trying

Virus-free.www.avg.com


 

Yes, delete and replace.

I haven’t tried other LT opamps yet, either

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of John Woodgate
Sent: Sunday, November 10, 2024 2:52 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

Not to use costly LT1012s? Bur seriously, what happens if you put another real opamp as U3 and U4. I can't check, because I can't see how to change from the Uopamp2 to a real one. Do I have to delete the symbol from the schematic and put in a new one? I've never used Uopamp2.

On 2024-11-10 21:38, Bell, Dave via groups.io wrote:

Please see recent upload…

 

I started from basics. to create a simple log compressor for a limited range of input.

That seems to work well in an "ideal" simulation, using Univ OpAmp2.

Then I got inspired to test an equally simple DEcompressor by putting the circuit in the feedback path of another UniOpAmp2, which also worked (surprisingly) well.

 

Replacing the amps in the Compressor path with LT1012's appears to work just as desired, as well. But replacing the DEcompressor amps fails miserably.

 

What can I learn today?

 

Dave

-- 
OOO - Own Opinions Only
Best Wishes
John Woodgate
Keep trying

 

Virus-free.www.avg.com


 

I just tried LT1022’s for U3/4, and they work fine.

Weird?

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of Bell, Dave via groups.io
Sent: Sunday, November 10, 2024 4:07 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

Yes, delete and replace.

I haven’t tried other LT opamps yet, either

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of John Woodgate
Sent: Sunday, November 10, 2024 2:52 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

Not to use costly LT1012s? Bur seriously, what happens if you put another real opamp as U3 and U4. I can't check, because I can't see how to change from the Uopamp2 to a real one. Do I have to delete the symbol from the schematic and put in a new one? I've never used Uopamp2.

On 2024-11-10 21:38, Bell, Dave via groups.io wrote:

Please see recent upload…

 

I started from basics. to create a simple log compressor for a limited range of input.

That seems to work well in an "ideal" simulation, using Univ OpAmp2.

Then I got inspired to test an equally simple DEcompressor by putting the circuit in the feedback path of another UniOpAmp2, which also worked (surprisingly) well.

 

Replacing the amps in the Compressor path with LT1012's appears to work just as desired, as well. But replacing the DEcompressor amps fails miserably.

 

What can I learn today?

 

Dave

-- 
OOO - Own Opinions Only
Best Wishes
John Woodgate
Keep trying

 

Virus-free.www.avg.com


 

Never mind; LT1022’s don’t work

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of Bell, Dave via groups.io
Sent: Sunday, November 10, 2024 4:10 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

I just tried LT1022’s for U3/4, and they work fine.

Weird?

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of Bell, Dave via groups.io
Sent: Sunday, November 10, 2024 4:07 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

Yes, delete and replace.

I haven’t tried other LT opamps yet, either

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of John Woodgate
Sent: Sunday, November 10, 2024 2:52 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

Not to use costly LT1012s? Bur seriously, what happens if you put another real opamp as U3 and U4. I can't check, because I can't see how to change from the Uopamp2 to a real one. Do I have to delete the symbol from the schematic and put in a new one? I've never used Uopamp2.

On 2024-11-10 21:38, Bell, Dave via groups.io wrote:

Please see recent upload…

 

I started from basics. to create a simple log compressor for a limited range of input.

That seems to work well in an "ideal" simulation, using Univ OpAmp2.

Then I got inspired to test an equally simple DEcompressor by putting the circuit in the feedback path of another UniOpAmp2, which also worked (surprisingly) well.

 

Replacing the amps in the Compressor path with LT1012's appears to work just as desired, as well. But replacing the DEcompressor amps fails miserably.

 

What can I learn today?

 

Dave

-- 
OOO - Own Opinions Only
Best Wishes
John Woodgate
Keep trying

 

Virus-free.www.avg.com


 

Well, LT1037’s work in U3/4. But NOT as U1/2 (oscillate wildky)

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of Bell, Dave via groups.io
Sent: Sunday, November 10, 2024 4:14 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

Never mind; LT1022’s don’t work

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of Bell, Dave via groups.io
Sent: Sunday, November 10, 2024 4:10 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

I just tried LT1022’s for U3/4, and they work fine.

Weird?

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of Bell, Dave via groups.io
Sent: Sunday, November 10, 2024 4:07 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

Yes, delete and replace.

I haven’t tried other LT opamps yet, either

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of John Woodgate
Sent: Sunday, November 10, 2024 2:52 PM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

Not to use costly LT1012s? Bur seriously, what happens if you put another real opamp as U3 and U4. I can't check, because I can't see how to change from the Uopamp2 to a real one. Do I have to delete the symbol from the schematic and put in a new one? I've never used Uopamp2.

On 2024-11-10 21:38, Bell, Dave via groups.io wrote:

Please see recent upload…

 

I started from basics. to create a simple log compressor for a limited range of input.

That seems to work well in an "ideal" simulation, using Univ OpAmp2.

Then I got inspired to test an equally simple DEcompressor by putting the circuit in the feedback path of another UniOpAmp2, which also worked (surprisingly) well.

 

Replacing the amps in the Compressor path with LT1012's appears to work just as desired, as well. But replacing the DEcompressor amps fails miserably.

 

What can I learn today?

 

Dave

-- 
OOO - Own Opinions Only
Best Wishes
John Woodgate
Keep trying

 

Virus-free.www.avg.com


 

Dave,
 
What am I missing here?
 
You say that your uploaded decompressor fails miserably.  In what way does it fail?  Error messages?  (There were none.)  Widely divergent signals?  (There were none that I saw.)  Oscillations?  (I didn't see any.)
 
So I think you need to consider the settings you used, which apparently were not the same settings I used when I tried your simulation.
 
That being said - the whole idea of putting a circuit into the feedback loop of another op-amp is inherently risky.  There is a chance of AC instability, due to the fact that you have two amps in the feedback loop, contributing to degraded phase margin unless their primary poles are widely separated.  Have you considered that?  It may, in fact, be one reason why some op-amps oscillate in your simulations whereas others do not.  Also they might all be on the verge of instability.
 
Andy
 
 


 

OOPS.
 
I just noticed my LTspice was set to use the Alternate Solver.  I wonder how long I had left it like that.  :-(
 
No, changing back to the Normal Solver did not make any difference.  I do not see misbehavior.
 
Can you explain what you saw, which makes you say that it fails?
 
Andy
 
 


 

I will say that the U3/U4 combination looks bad.  It has two op-amps, both with the same GPB.  That is not good.
 
As a general rule of thumb, I think one of the amps should have much wider bandwidth than the other.  It's more complicated than that, but it is a place to start.
 
Andy
 
 


 
Змінено

Have you considered that:
  • U4 inverts,
  • U4's output leads to the inverting input of U3,
  • Therefore it is a positive feedback loop.
 
I'm only guessing here, but any DC offset of a 'real' op-amp might be amplified infinitely (well, OK, not infinitely) until one of them saturates (clips).
 
You might consider swapping the input pins of U3, but you might still need to deal with the AC stability problem.
 
Andy
 


 

I tried this many moons ago, not in simulation, but on the bench. When it miserably failed, I just turned on the brains and realised it couldn't work because there is a point where the sum of phase-shifts reaches 180° and the system is no more stable.

Le 10/11/2024 à 22:38, Bell, Dave via groups.io a écrit :

Please see recent upload…

 

I started from basics. to create a simple log compressor for a limited range of input.

That seems to work well in an "ideal" simulation, using Univ OpAmp2.

Then I got inspired to test an equally simple DEcompressor by putting the circuit in the feedback path of another UniOpAmp2, which also worked (surprisingly) well.

 

Replacing the amps in the Compressor path with LT1012's appears to work just as desired, as well. But replacing the DEcompressor amps fails miserably.

 

What can I learn today?

 

Dave


 

The LT1037 is not unity-gain stable - needs to be for the logamp.

--
Regards,
Tony

On 11/11/2024 01:20, Bell, Dave via groups.io wrote:

Well, LT1037’s work in U3/4. But NOT as U1/2 (oscillate wildky)


 

But first, you need to start with getting the polarities right!
 
Then worry about excess phase and unity-gain stability.  Those are secondary.
 
Andy
 


 

I admit, I hadn't looked at Dave's circuit yet. But I've been there before.

--
Regards,
Tony



On 11/11/2024 15:08, Andy I via groups.io wrote:

But first, you need to start with getting the polarities right!
 
Then worry about excess phase and unity-gain stability.  Those are secondary.


 

I know…

The decompression was a side thought; the simple log conversion was what I intended for use in an embedded data acquisition application. Low frequencies, constrained input ranges, but wider than a uC ADC can adequately fit over maybe 3 decades.

With compression, I can get good (enough) resolution at the low end and a single input range.

Could use a little temp compensation, but otherwise all I need.

 

Dave

 

From: LTspice@groups.io <LTspice@groups.io> On Behalf Of Andy I via groups.io
Sent: Monday, November 11, 2024 6:09 AM
To: LTspice@groups.io
Subject: EXTERNAL: Re: [LTspice] Logarithmic Compressor and Decompressor circuits

 

But first, you need to start with getting the polarities right!

 

Then worry about excess phase and unity-gain stability.  Those are secondary.

 

Andy