Strange plots analysis of audio compressors

[bombofa]

Hi, I modelled two circuits of tube and VCA compressor.
They seem to work well, but the only strange thing is the start point in the plots , there is in both compressors a wave ( oscillation ) for the first half of the analysis, and then they seem to adjust.
There are some errors in the circuits or in the .tran analysis?

In the tube comp, if I change the attack resistor from 81k to 500k ( the first one after the 2 tube diodes rectifer) the "oscillation" is a little bit less.
The VCA model is THAT 2181 , I found it in on the web and also I downloaded the that .lib.

View attachment 67151

 

[ericgibbs]

hi,
Regarding the tube asc, the inductors are very high inductive value, why is this.??

I dont have the THAT 2181 sub/mod.

E.

EDIT:
This is what I see when I run your file, I am assuming N009 is the top of the output transformer secondary.??

It helps if you use labels for your plot test points

 

[bombofa]

Yes, N009 is the output transformer (secondary), and I have the same result whit 500k resistor, but if I change that one whit 81 k ( the slow attack) the result has the oscillation.
And the same oscillation I had in the VCA comp, like I had post in the previous post.
Maybe it is right, but I don't know!

I used high inductance because if I used low inductance I have a cut of 6 db in freq resp! Can you solve this problem? I would like to have a 9:1 trafo!

 

[ericgibbs]

hi.
I have used a 81K, and this is what I see, the oscillation is much less than yours.???

I will look over the circuit, see if I can suggest idea's

E.

 

[ericgibbs]

hi,
I have opened the AGC feedback loop and used a linear voltage to control the overall gain of the amplifier.

You can see from the plot that the problem is with the amplifier itself [non linear gain control] and the way that AGC voltage thats being generated.

The plots are labelled on the circuit
E.

EDIT:
Modified the plots, as you can see the amplitude at which your trying to control the output level coincides with the close to 0V on the 12AX7 grids where the AGC voltage no longer has linear control.

 

[bombofa]

OK and what it means? There is an error on the amplifier? Maybe the tube that I've used?

I don't understand if there is a compression or not?

 

[ericgibbs]

OK and what it means? There is an error on the amplifier? Maybe the tube that I've used?

I don't understand if there is a compression or not?

There is compression, look at the plot of the output versus the AGC voltage.

What device will be driving the compressor input when the circuit is used,??

EDIT;
Try reducing then input signal from 0.3V to 0.1V, this will help you understand the problem.

The relationship between the AGC and the input signal is not tracking correctly.

 

[ronsimpson]

The VCA model is THAT 2181 , I found it in on the web and also I downloaded the that .lib.

Where did you find THAT spice models?

 

[bombofa]

Hi, in this tread:

https://www.groupdiy.com/index.php?topic=44581.0

 

[ronsimpson]

Thanks for the info on spice and THAT. I now have the .lib and will make the .sym later.

I have interest in Audio compressor/limiter/expander/scratch and pop filters/hiss removal............
My first limiter used a LED and Light Dependent Resistor as a VCA. The response was too slow but it did work. Next I used MOSFET as VCA. Then used 4 quadrant multiplier. Mostly back in the 1975-80s.

 

[bombofa]

Hi, I found a new strange thing:

When I run the transient simuation I can see the gain reduction of the compressor ( tube one ), but wheen I try to run the ac simulation, for the frequency response, I can't see the gain reduction.
It seems that there isn't compression....?

 

[ronsimpson]

Yes that is the way spice works.
AC simulation does not do real time things. (sorry can't think of the word)
To make AC simulation work you need to put a DC voltage on a VCA and force the gain to some value.

From THATCORP.COM

One cannot accurately simulate dynamic gain control while performing an .AC analysis. This is because SPICE is performing a small signal analysis on the devices based on how they are biased after the DC bias portion of the simulation. Large signal behavior, such as the exponential relationship between VBE and IC are not modeled, and this exponential relationship is fundamental to the operation of THAT Corporation's log/anti-log VCAs. Performance with static gain can be simulated with the .AC analysis, and dynamic gain control can be simulated with TRAN (transient) analysis.

Also, the results of .AC analysis with THAT Corporation's RMS detectors can be misleading, since the result only shows the ripple of the RMS detector's log filter, and not the actual DC output. Transient analysis yields the correct DC output, but the user may want to experiment with the .IC statement to speed settling times for the log filter.

 

[audioguru]

My local TV station has a horrible compressor that is too active for the news people and outside events.
The compressor is too slow and lets very loud sounds pass for the first half a second before it wakes up then everything compresses too much.

I would like to hear what the news people really sound like without the horrible compressor.

 

[ronsimpson]

My local TV station has a horrible compressor that is too active for the news people and outside events.
The compressor is too slow and lets very loud sounds pass for the first half a second before it wakes up then everything compresses too much.

That's easy to fix. Just turn a knob or two.
1) Decrease the attack time.
2) Reduce the signal into the limiter.