LA2A feedforward/feedback design?

[abechap024]

Hello,
Finishing putting together my la2a project, kinda interested in the theory behind modding it with a potentiometer to act more "feedforward". I imagine it probably wouldn't work, or sound good 100% feedforward just because of the self correcting nature of the opto cell, but interested if anyone wouldn't mind sharing their thoughts/experiments on this.
Thanks!

 

[ruffrecords]

| would guess that feed forward makes the resultant response a lot more dependent on the actual characteristics of the cell. With feedback, any variance in the sell characteristics tends to be smoothed out.

Cheers

Ian

 

[dandeurloo]

wouldn't that be just putting a pot on R11? 

 

[Kingston]

wouldn't that be just putting a pot on R11?

This is a global NFB loop of the line amp stage. Has absolutely nothing to do with the compression or the optical parts. No interaction between the optical stage and line amp whatsoever.

Perhaps abechap024 should also clarify what he means by "modding it with a potentiometer".

 

[abechap024]

Well Ive seen on some hot rodded La2a designs they have a pot that goes from feedback to feedforward. Dont know if this is a trade secret, but just curious how theoretically this would be implemented.

 

[Kingston]

Well Ive seen on some hot rodded La2a designs they have a pot that goes from feedback to feedforward. Dont know if this is a trade secret, but just curious how theoretically this would be implemented.

well there's R7. At the compress/bypass position it's "feedback" around the optical stage.

Increase the value of R7 and it sends less and less of the photo resistor affected "output" of the optical stage back to the sidechain (actually the light emitting foil).

In practise fiddling with this value makes surprisingly little difference. I wouldn't call this "feedforward". Just means you send less (or none if you like) of the photoresistive/compressed output back to the sidechain/lamp.

 

[ruffrecords]

Well Ive seen on some hot rodded La2a designs they have a pot that goes from feedback to feedforward. Dont know if this is a trade secret, but just curious how theoretically this would be implemented.

I think you need to be much more specific about this. There are no two points in the LA2A that you would obviously label feedforward and feedback. Can you draw us a simple schematic of precisely what you mean?

Cheers

Ian

 

[Kingston]

Well Ive seen on some hot rodded La2a designs they have a pot that goes from feedback to feedforward. Dont know if this is a trade secret, but just curious how theoretically this would be implemented.

I think you need to be much more specific about this. There are no two points in the LA2A that you would obviously label feedforward and feedback. Can you draw us a simple schematic of precisely what you mean?

Cheers

Ian

Maybe you've already seen this. The original schematic is much less clear about this.

http://www.twin-x.com/groupdiy/albums/userpics/LA2A-deconstructed.jpg

R7 is feedback of sorts - R7 bypassed especially - just not the electronic kind like the global NFB (R11 with its pre-filtering) of the line amp.

 

[dandeurloo]

Thanks again Kingston.  That makes sense.  I may have to try it and see if it sounds cool.  Seems like the kind of thing that would be just as happy on a on/off switch if it sounds good.  Although, it may be more of a headache then its worth like a number of the mods.  But heck at this point I have tested and tried all of them so I might as well at least give this one a listen.

 

[dmp]

So feedforward means that the signal triggering compression is taken before the compression is applied.
Feedback means the signal triggering compression is taken after the compression is applied.
Right?

So the signal in the original schematic is sent to the sidechain at the middle of the voltage divider at the opto cell, so any compression applied will lower the signal to the sidechain.
This sounds like a feedback design, yes?

To switch to feedforward, you would want the signal to the sidechain to come before the GR voltage divider, upstream of R6.
So try a pot in place of R6 (68k), where the wiper is connected to the sidechain.

 

[Kingston]

So try a pot in place of R6 (68k), where the wiper is connected to the sidechain.

This has the added effect that the signal going to the sidechain will be very loud (something like +10dB) at one end of the pot. It might have undesired effects on the sidechain, like the lamp being constantly on. I'm not absolutely sure but blasting the lamp (light emitting foil) too loud might shorten its life considerably.

It's easier to just replace R7 with like a 100k pot (non critical, could be 50k too, wire as "rheostat"), and omit the compress/limit entirely.

Notice that the compress/limit switch already does this same thing, just in a less drastic way.

 

[abechap024]

Cool. Yes just what DMP said. a tap take from before the opto-cell. I am wary of side effects of this, I don't really want to mess with the original design but might be fun to tinker around with.

Kingston that is an interesting approach, yea I guess I never thought about that, but a in limit mode it is more or less like feedforward.

Maybe taking a tap on the input side of R6 with a 68k series resistor to sidechain would get even more feedforward type results...

 

[dmp]

This has the added effect that the signal going to the sidechain will be very loud (something like +10dB) at one end of the pot. It might have undesired effects on the sidechain, like the lamp being constantly on. I'm not absolutely sure but blasting the lamp (light emitting foil) too loud might shorten its life considerably.

The threshold pot is after this so, no, it is not a problem, but yes, the threshold might need to be adjusted lower.

It's easier to just replace R7 with like a 100k pot (non critical, could be 50k too, wire as "rheostat"), and omit the compress/limit entirely.

It might be easier, but the signal to the sidechain is in the voltage divider, so it is still in-between methods.

 

[Kingston]

It might be easier, but the signal to the sidechain is in the voltage divider, so it is still in-between methods.

Actually the net result should be equal (if R7 is a 68k rheostat). The other has you adjusting "threshold" to compensate for increased sidechain input voltage (there's still feedback, similarly in between methods), the other has you increasing make up gain to compensate for lower volume resulted by the voltage divider.

 

[ruffrecords]

Maybe you've already seen this. The original schematic is much less clear about this.

http://www.twin-x.com/groupdiy/albums/userpics/LA2A-deconstructed.jpg

R7 is feedback of sorts - R7 bypassed especially - just not the electronic kind like the global NFB (R11 with its pre-filtering) of the line amp.

Yes I have seen that. R7 does not appear to me to be involved in feedback or forward. The opto cell resistance can reach quite low levels and with the switch shorting R7 the amount of attenuation that can be achieved is very high. The original spec says 40dB so the opto cell resistance must fall to about 680 ohms in limit mode (R7 shorted). In compress mode the 2K7 in series simply limits the maximum gain reduction that can be achieved to about 26dB. You could probably make a fairly smooth transition from compress to limit simply by making the 2K7 a pot.

Cheers

Ian

 

[Kingston]

R7 does not appear to me to be involved in feedback or forward.

When R7 is shorted, the photo resistor (compressed) output that goes to line amp, also goes directly back to the side chain input. This makes the compression design inherently feedback based through the light emitting foil and photo resistor. If you break this link (high R7 value), there is no longer (or significantly less) this form of feedback. Compression response changes. 2K7 is almost negligible, 20k does quite a different thing already.

I don't know who came up with those terms "compress" and "limit" for this function. They do not describe what is happening.

 

[ruffrecords]

I don't know who came up with those terms "compress" and "limit" for this function. They do not describe what is happening.

The original manufacturer called them that. I don't have them to hand right now but if you search the net you can find input/output  curves for each position. The compress curve looks very much like a soft knee compressor with a 3:1 ratio and the limit curve looks much the same but with an ultimate ratio in the region of 10 or 20 to 1.

Cheers

Ian

 

[ruffrecords]

Well Ive seen on some hot rodded La2a designs they have a pot that goes from feedback to feedforward. Dont know if this is a trade secret, but just curious how theoretically this would be implemented.

well there's R7. At the compress/bypass position it's "feedback" around the optical stage.

Increase the value of R7 and it sends less and less of the photo resistor affected "output" of the optical stage back to the sidechain (actually the light emitting foil).

In practise fiddling with this value makes surprisingly little difference. I wouldn't call this "feedforward". Just means you send less (or none if you like) of the photoresistive/compressed output back to the sidechain/lamp.

The feedback in this circuit, as in most compressors is via the side chain. The surprising thing is that changing the amount of this feedback makes no difference whatsoever to the compressor characteristic. All it does is alter the threshold. The compressor characteristic is entirely determined by the opto and the pot divider it makes. So including R7 for compress response not only alters the characteristic but also lowers the threshold which what you would expect.

Cheers

Ian

 

[KevinOConnor]

Hi Guys

A few years ago a friend built an LA-2A to hear how it sounds.  It is okay when used for very mild compression but sounds awful when pushed, which is why most engineers cascade two units each set for a mild effect if they want a deeper one.

The bulk of sonic problems are in the side-chain. The EL-84 driving the opto adds quite a bit of distortion to the control signal, but enclosing it in a feedback loop makes it work much better. In the end, we might have replaced the pentode with an ECC-99 or something similar to provide a lower output impedance - I can't recall for certain what tube we used finally, and the notes are buried.

For the time it was built, the LA-2A was impressive and the electroluminescent cell is a cool device - too bad it is single sourced. The output gain block can be made much cleaner, as well, as it impedes tone if you want more accurate sounds  - an oxymoron since compression by definition is distorting the signal.

Have fun
Kevin O'Connor
londonpower.com

 

[Biasrocks]

A few years ago a friend built an LA-2A to hear how it sounds.  It is okay when used for very mild compression but sounds awful when pushed, which is why most engineers cascade two units each set for a mild effect if they want a deeper one.

Your La2a is broken.

I'm continually amazed at how transparent mine sounds, even with huge amounts of gain reduction.

The reason you chain a 1176 and la2a together is because the la2a is relatively slow and the fast acting 1176 catches the stuff that the opto misses.

If you want clinical compression you need to look elsewhere.

Mark