Will a 2n5089 vs. a 2n5088 be quieter loess noisey in my Ross Comp

[rosssurf]

I have built several Ross comps some with the rise time toggle, a small cap across the output pot to allow highs to pass and bigger caps on the input and out put .

Will changing the 2n5088's to 2n5089's make any audible difference in backgroung noise or hiss?

If so which ones are worth changing?

Will it affect the sound in any other way?   

Thoughts and suggestions appreciated.

[John Lyons]

Probably not.
Technically the 89 is quieter but it's a low gain circuit so....

John

[Paul Marossy]

Probably not.
Technically the 89 is quieter but it's a low gain circuit so....

John

The 2N5088 is a medium gain general purpose NPN transistor. The 2N5089 is a high gain transistor that is made specifically for use in audio applications or other applications where circuit noise must be minimized.

IMO, in the case of the Ross Compressor, it is a low gain circuit but there are five transistors so there might be a cumulative effect if you used "quieter" transistors.

[ayayay!]

Any impact on noise will be minimal at best.  MXR's use MPSA13's which have much more gain than 5088/9's, and I think they're silent enough.  The circuit design is what it is.  In this case, I'd go for all metal film resistors & caps first to make it the quietest, and NO tantalums.

[composition4]

I would say probably not.  Most of the noise in a Ross comp comes from the CA3080 OTA I think.  So you could use theoretically perfect noise-free transistors but as the noisefloor is already at a certain high level because of the OTA, it won't matter what happens below that floor anyway.

Also, compressors are inherently noisy because of the job that they are asked to do.

I would just use the 5089s if you have them anyway, using then won't hurt at all even if the difference is minimal/negligible.  Shouldn't make a difference to the circuit in any other way.

Jonathan

[R.G.]

I would say probably not.  Most of the noise in a Ross comp comes from the CA3080 OTA I think.  So you could use theoretically perfect noise-free transistors but as the noisefloor is already at a certain high level because of the OTA, it won't matter what happens below that floor anyway.

... and we have a winner.

Almost all the noise in the Ross/Dynacomp circuit comes from the OTA; and from the fact that the signal must be reduced down to 10-25mv at the input of the OTA to keep distortion in bounds. After this, it must be re-amplified back up to the output level again. It's even worse when the signal level at the input is low. In this case, the compressor tries to amplify something back up to the limiting level, and it's own noise is all it has to get there with, so the noise comes up all it can.

Paradoxically, noise is least in the Dynacomp style circuits (Ross is a minor tweak on this) when the incoming audio signal is biggest and most able to mask any noise the compressor makes; and noise is highest when the incoming audio signal is lowest, and least able to mask any noise.

[Paul Marossy]

Doh! I didn't notice that the CA3080 was on that schematic. I agree, that's got to be the culprit in this case.

I built that Anderton "Volume Pedal De-Scratcher" circuit, which uses a CA3080. While I think conceptually this circuit is pretty cool, I was disappointed by the noise floor in the circuit. IMO, it's too noisy to use either live or in a recording setting when compared to other volume pedals I have. The quietest one I have by far is the DOD FX-17 wah/volume pedal. That thing is dead silent.

[Quackzed]

Paradoxically, noise is least in the Dynacomp style circuits (Ross is a minor tweak on this) when the incoming audio signal is biggest and most able to mask any noise the compressor makes; and noise is highest when the incoming audio signal is lowest, and least able to mask any noise.

So does that mean that a relatively low noise booster, placed before a dynacomp type circuit, would allow you to get more compression out of the circuit at lower compression settings, and therefore less overall noise from the 3080? Seems plausable that a louder signal might prevent the 3080 from needing to amplify SO MUCH, and therefore lower the overall noise level. ?

[rosssurf]

Thanks for all the wisdom folks! This forum has taught me so much. Is there a substitution for the 3080E that might be worth trying in this circuit?

[R.G.]

Paradoxically, noise is least in the Dynacomp style circuits (Ross is a minor tweak on this) when the incoming audio signal is biggest and most able to mask any noise the compressor makes; and noise is highest when the incoming audio signal is lowest, and least able to mask any noise.

So does that mean that a relatively low noise booster, placed before a dynacomp type circuit, would allow you to get more compression out of the circuit at lower compression settings, and therefore less overall noise from the 3080? Seems plausable that a louder signal might prevent the 3080 from needing to amplify SO MUCH, and therefore lower the overall noise level. ?

Could be, but the reason most people like so-called "optical based" compressors is that the typical compressor with that name first amplifies up the signal with a low-noise amplifier, then only lowers the signal by loading/dividing it down with an LDR or similar. This makes the noise just dependent on the noise qualities of the input amplifier, not the gain-changer.

It's funny - compressors done with the NE570/571 are much quieter. It's the use of the 3080 and its follow-ons that make the average guitar compressor noisier than the simpler hifi/studio ones. This is yet another place where we're trapped by history. The Dynacomp compressor was the first good compressor for guitar, so it got ensconsced in history as THE compressor to emulate, and beginners, even apprentice effects hackers only want to do that one. There are other, better ways, but they don't have the vintage cachet of the Dyna and the presumably better Ross. The Ross compressor has a few more resistors and filter caps to cut hum, and that's all. It's in no way an evolutionary improvement.

[ayayay!]

What about building a Ross/DC w/ a BA6110?  Reason is I saw a CS-2 using a 3080, so I'm intrigued.  

Dang I seemed to have misplaced that 6110 datasheet again.  Grrr....

edit:  Ahh, here it is  http://urekarm.tripod.com/synth/BA6110.pdf

[Mark Hammer]

The key limitation of the CA3080 is that you can't observe the normal rules of optimizing S/N.  The 3080 distorts at such a low input voltage that you can't feed it a nice hot signal.  So, it not the 3080 itself that is so noisy, but rather the fact that the noise it does contribute is allowed to overshadow everything.

As for the trannies, only 2 of them are actually in the audio path.  The other 3 are part of the sidechain, so variations in the noise specs of those trannies will make no audible difference.

[Paul Marossy]

The key limitation of the CA3080 is that you can't observe the normal rules of optimizing S/N.  The 3080 distorts at such a low input voltage that you can't feed it a nice hot signal.  So, it not the 3080 itself that is so noisy, but rather the fact that the noise it does contribute is allowed to overshadow everything.

As for the trannies, only 2 of them are actually in the audio path.  The other 3 are part of the sidechain, so variations in the noise specs of those trannies will make no audible difference.

Maybe someone out there needs to design a new and improved version of the CA3080. Are there other OTAs out there besides that one? That's the only one that I've really ever heard of...

[composition4]

Turns out Don Tillman is good for something else as well as a JFET buffer... 

http://www.till.com/blog/archives/2005/06/last_of_the_ota.html

[Mark Hammer]

The key limitation of the CA3080 is that you can't observe the normal rules of optimizing S/N.  The 3080 distorts at such a low input voltage that you can't feed it a nice hot signal.  So, it not the 3080 itself that is so noisy, but rather the fact that the noise it does contribute is allowed to overshadow everything.

As for the trannies, only 2 of them are actually in the audio path.  The other 3 are part of the sidechain, so variations in the noise specs of those trannies will make no audible difference.

Maybe someone out there needs to design a new and improved version of the CA3080. Are there other OTAs out there besides that one? That's the only one that I've really ever heard of...

Plenty.  The CA3280 is better than the 3080.  The LM13600 and 13700 are "better", as is the Rohm BA6110 and the BA662 that Roland has made specially for them.

[Paul Marossy]

The key limitation of the CA3080 is that you can't observe the normal rules of optimizing S/N.  The 3080 distorts at such a low input voltage that you can't feed it a nice hot signal.  So, it not the 3080 itself that is so noisy, but rather the fact that the noise it does contribute is allowed to overshadow everything.

As for the trannies, only 2 of them are actually in the audio path.  The other 3 are part of the sidechain, so variations in the noise specs of those trannies will make no audible difference.

Maybe someone out there needs to design a new and improved version of the CA3080. Are there other OTAs out there besides that one? That's the only one that I've really ever heard of...

Plenty.  The CA3280 is better than the 3080.  The LM13600 and 13700 are "better", as is the Rohm BA6110 and the BA662 that Roland has made specially for them.

Cool, I didn't know that. Are those straight replacements for the CA3080 or is tweaking required for a circuit that would normally use a CA3080?

[vondran]

Almost all the noise in the Ross/Dynacomp circuit comes from the OTA; and from the fact that the signal must be reduced down to 10-25mv at the input of the OTA to keep distortion in bounds. After this, it must be re-amplified back up to the output level again. It's even worse when the signal level at the input is low. 

I didn't know it lowered the signal level before the OTA.  Is that a result of the 10k resistor before the input transitor?  I can't seem were else it is done. 10k does seem high as I typically I see 1k on the input of the input stage.  I run my R/D comp at 18v, so am wondering if I have more headroom to drive the OTA at full input level.

[R.G.]

I didn't know it lowered the signal level before the OTA.  Is that a result of the 10k resistor before the input transitor?  I can't seem were else it is done. 10k does seem high as I typically I see 1k on the input of the input stage.  I run my R/D comp at 18v, so am wondering if I have more headroom to drive the OTA at full input level.

All 3080/3094 compressors and many LM13600/13700 compressors lower the signal level before the input. The input of these devices is a raw NPN differential amplifier, and the linear range on a bipolar diffamp is only between 10 and 15 millivolts. They are driven to full clipping with 25millivolts, and this is a factor of the semiconductor physics of silicon bipolar transistors, not something a better transistor will fix. So all circuits that don't want distortion from the inputs will put some kind of divider ahead of them to get the signal down to well below 25mV. In the case of the Dynacomp, it is the combination of the 10K and the 1k/1k resistors keeping the signal level down.

Running the compressor at 18V does not a single thing to change this. All it does is change the output dynamic range and increase the power dissipation on the chip. The input level limitation is a limit imposed by the junction physics of the input differential amplifier.

Notice that this is fundamentally different from the diffamps in operational amplifier front ends. In opamps, the 10-25mV limit is still there, but the feedback of the opamp forces the negative and positive inputs to be within a few millivolts of each other at all times, so input distortion is essentially non-existent for inputs which are within the input common mode range. OTA applications like the Dynacomp do not use feedback to keep the voltage difference off the inputs.

[DDD]

R.G. - respect !
The best explanation of the difference between the " input behaviour" of the usual OpAmp and the OTA I've ever read!

[vondran]

R.G. - respect !
The best explanation of the difference between the " input behaviour" of the usual OpAmp and the OTA I've ever read!

Yes, this is why I love to ask.  I learn so much.