B-Format buffer preamp, Mk.II

Here's a later version, with changes made as per suggestions from PRR and bcarso:


One channel shown, of a matched-gain 4-channel version.

I tried an 072 as the first stage, and it seemed to want to oscillate rather too readily. Currently I have 5532's in there and it's not too noisy, though I have some investigating to do.

Anyone see anything easy ways to improve this? -The 39K resistors have been scaled to match each gang of the pot's static end-to-end resistance, for improved inter-channel tracking.

Keith A.

Probably the oscillation tendency is associated with parasitics in the switch wiring? This could be alleviated with lower Z's of the R's, or possibly some distributed small feedback C's. Or similarly, excess shunt C and relatively high feedback R at some settings of the gain pot, which again might be corrected with a small feedback C. I remember a Baxandall tone control I did once, I think with an 072, that had a similar problem if the op amp wasn't right up against the pot and other parts. And it still needed a feedback C besides the other components.

It would be nice to preserve the 072's just because of the wiper noise reduction in the second stage. I guess you could use a 553X for the first stage and an 072 for the second. Then you could make the gain switch R's a good deal smaller and not worry about parasitic poles from strays and amp input capacitance.

[quote author="bcarso"]It would be nice to preserve the 072's just because of the wiper noise reduction in the second stage. I guess you could use a 553X for the first stage and an 072 for the second. Then you could make the gain switch R's a good deal smaller and not worry about parasitic poles from strays and amp input capacitance.[/quote]
That's how I have it at the moment, (half a 5532 for the first stage, half an 072 for the second) since originally I just built the second stage, then added the first stage when it became apparent that operationally I might need to switch in the 20dB pad from time to time. I have two IC's doing the first stages for 4 channels, then two more ICs doing the gain stages for the second channels. The wiper noise is currently well down below -60dBFS, pretty much silent even making adjustments during quiet passages.

Thanks for looking it over again. -What reduction in size should I make the resistance ladder? -A factor of ten, you reckon? -or more?

Keith

For that first stage, you're probably better off switching input resistors and leaving the feedback resistor constant at, say, 22k. Something like:



That gets the switching away from the - input, so strays will be less of a problem.

For the second amp, I'd put 30k as the input resistor, and have the feedback a 24.9k resistor in series with a 50k pot rigged as a variable resistor, with the pot toward the output end of the string.

Peace,
Paul

The pot is the only value that I can find in a decent quality 4-gang, reasonably matched single unit, so I'm sorta forced to use it.

Switching gain in the first stage does currently momentarily 'slam' the output agains the rail, as all NFB was lost: this evening, I was going to try clipping a 1Meg resistance from the output to the wiper of the switch, to see if that should provide a constant DC feedback path and also momentarily drop the gain to unity as the switch breaks, before re-making...

Keith

Paul's suggestion is a good one as now the feedback C has roughly the same effect at all switched gains.

For the slamming problem, get a make-before-break rotary selector switch if possible.

[quote author="bcarso"]Paul's suggestion is a good one as now the feedback C has roughly the same effect at all switched gains.

For the slamming problem, get a make-before-break rotary selector switch if possible.[/quote]

Actually, if you use my circuit, you don't want a make-before-break switch. With a break-before-make, the switch goes open momentarily between settings, and on the circuit I diagrammed, that gives you unity gain, lower than the other settings.

On the other hand, if you're switching the feedback resistor, then make-before-break is definitely in order.

Peace,
Paul

I might well re-do it according to Paul's circuit, but tonight is a rehearsal in a particularly nice auditorium for the 35-piece brass band, so I wanted to use it there. Tomorrow is another and this weekend is a gig, so I'll use it as-is for now, though I may well add the 1-meg resistor to the current design as this should also drop the gain momentarily to unity in between settings... (I think: signal to + input, feedback through 1Meg to -input) but the 1Meg should be plenty high enough to not significantly affect the 10dB-per-step gain settings.

Keith

Hehe that reminds me of some occasions when I was just completing gear in time for a recording. Nothing like deadline pressure.

Or as an (ex)-friend put it, "deadlines are useful even when you don't make them."

Hey keith, re your pot availability, can't you roll your own 4 gang whatever (10k, 47k, 100k) from all the clarostats you must have lying around?
I've got a bunch of ex SSL aux pots with failed schadows, but they keep me in decent dual 10k and 100ks by switching the pot sections over when I need them. Just a thought.