What is this circuit doing?

The FETs I presume are there for muting, what is the purpose of R154 and R155 please?

Many thanks.

The Fets are not ideal switches, so R154 and 155 feeds the same residual signal amount (when the mute is engaged) to the negative opamp's input. The opamp is wired as a differential amplifier.

Ah, so it adds addition attenuation? Cool, cheers.

I seem to recall Steve Dove writing in his classic series of articles about console design that ran in Studio Sound last century, he covered several similar circuit tricks to improve fader kill, pan pot kill, and the like. Tweaky console customers, and therefore console designers can lose a lot of brain cells trying to deliver GOP (good on paper) specifications for everything (like that mute kill example).

As I recall Steve described some exotic tricks like twisting wire leads together to make small value variable caps, to use in canceling stray capacitive coupling, etc.

JR

looks like a polarity switch.

But the -ve polarity has much less gain than +ve or maybe not cos on +ve, there is leakage to the -ve i/p.

I've used a much simpler version to get clean polarity swap.

JohnRoberts said:

As I recall Steve described some exotic tricks like twisting wire leads together to make small value variable caps, to use in canceling stray capacitive coupling, etc.

Click to expand...

"Gimmick" caps, good for single-digit pF. I recall reading an article about early Cray supercomputers, which also used gimmicks to tweak path timing.

I'm sure that they took Nigel Tufnel's admonition, "Don't touch it. Don't even look at it" seriously.

--a

ricardo said:

looks like a polarity switch.

But the -ve polarity has much less gain than +ve or maybe not cos on +ve, there is leakage to the -ve i/p.

I've used a much simpler version to get clean polarity swap.

Click to expand...

Yes at first glance, but note the 4.7ohm  resistor to ground in the inverting path..

The 4.7 ohm is a crude approximation for the JFET on resistance.

In consoles the customers don't like mutes that leak signal when muted. 

JR

> note the 4.7ohm  resistor to ground in the inverting path..
> The 4.7 ohm is a crude approximation for the JFET on resistance.

Note the one side is *10K* and 4.7r, the other side is *two* 3.3K and two FETs.

So we figure 4.7/10,004.7 is 0.000,47. The FET side must do the same with two dividers, so take square-root, 0.022 in each divider.

I figure the JFET is more like 71 Ohms. Which is a more believable value than 5 Ohms (a VERY fat JFET or a power MOSFET).

PRR said:

> note the 4.7ohm  resistor to ground in the inverting path..
> The 4.7 ohm is a crude approximation for the JFET on resistance.

Note the one side is *10K* and 4.7r, the other side is *two* 3.3K and two FETs.

So we figure 4.7/10,004.7 is 0.000,47. The FET side must do the same with two dividers, so take square-root, 0.022 in each divider.

I figure the JFET is more like 71 Ohms. Which is a more believable value than 5 Ohms (a VERY fat JFET or a power MOSFET). > note the 4.7ohm  resistor to ground in the inverting path..
> The 4.7 ohm is a crude approximation for the JFET on resistance.

Note the one side is *10K* and 4.7r, the other side is *two* 3.3K and two FETs.

So we figure 4.7/10,004.7 is 0.000,47. The FET side must do the same with two dividers, so take square-root, 0.022 in each divider.

I figure the JFET is more like 71 Ohms. Which is a more believable value than 5 Ohms (a VERY fat JFET or a power MOSFET).

Click to expand...

yes correctamundo....

My apologies if I mislead Ricardo but he thought it was a polarity switch. 

The significance of  mentioning a single digit ohms shunt is that the alternate signal leg is seriously attenuated tens of dB.  In practice JFET on resistance is not very repeatable batch to batch unless using graded parts so that actual value may get selected for whole batches of JFETS.

Consoles are not very high volume manufacturing so values probably don't get tweaked that often.

JR

To me the 10K/4R7 divider into-the-inverting-leg simply looks like a trial-error tweaked nulling circuit for the after-two-fets shunt mute.

Perhaps this was cheaper than adding yet another (or two or three) fets in series.

... but of course it was cheaper ...

Thanks all, much appreciated.

So, presumably for this to work the impedances for + and - inputs need to be pretty well matched, can't quite see how that works out but will assume it was done this way as a result of trial and error, as mentioned above.

Any ideas why R94 is there? putting it before the FETs would give the same series resistance but would increase the muting attenuation, no?

I was wondering that myself. Perhaps it just "sounded better" or performed some other sort of "matching" when tweaked.

If you look at the schem from a certain angle, the 3 x 3k3 in "+" leg add to 9k9, which should be a good-enough match for the R154 10k in the "-" leg. (when the mute isn't in fact "muting", IOW when fets are off).

Could be something else entirely... ?

It's 10k +2.2k though for the inverting leg though isn't it?

Nope. It's actually (f.e. at vol. pot at "0"): ((10k in parallel with 4R7) in series with 2k2) in parallel with 5k1)

That's what the "-" input sees. So actually, something roughly like 1k5.

So the balance in "legs" to my eyes ends with the 9k9/10k ratio that jumps out when you look at the schematic.

To unriddle this one, perhaps you should hunt down the actual designer.

To me, it looks like someone started with a "de-balancer" circuit, and then by trial-and-error got to where it stands as-is.

This is what used to be called a Pooge or a tweak...

They started with the 2 jfet shunt mute and then looked for practical ways to improve the mute kill using the following op amp as a differential to subtract out a small error voltage derived from the mute input.  Any one of the three resistors in the subtract path could be varied for deepest kill. I suspect at least two of those values were already in use else where in the circuit, perhaps all three.

Starting with a blank sheet of paper, I would have made the 2.2k R larger to reduce noise gain of that stage, but suspect that combination delivered good kill using standard parts already in use and the resulting kill and noise floor was adequate .

JR

Had another look at the schematic, they use  this circuit elsewhere but with different values for the 10k, 4r7 combo!

Ah well, time to forget about this circuit me thinks! Cheers all.

FETlife said:

Had another look at the schematic, they use  this circuit elsewhere but with different values for the 10k, 4r7 combo!

Ah well, time to forget about this circuit me thinks! Cheers all.

Click to expand...

Different JFETs will exhibit different Rds on, so require a different ratio on the tweak to provide deepest mute.

JR