Multi-Channel Preamp for Stereo Output instruments


2010-10-04 5:25 am
I finally settled on a basic design for my next build. Effectively a "swiss-army-knife" stereo instrument preamp designed with signal flexibility in mind.

This originally started as a balanced headphone amp concept based on Ti Kan's JISBOS (J-Fet input stage Bi-fet output stage) line driver. I already have the JISBOS boards, and have had a desire to build a stereo preamp that would work with my NS/Stick. You know what they say about curiousity... rolleyes:

The Block Diagram (click here for large version):


The Design Brief:

-2 discrete instrument input channels with EQ and an Aux/FX send.

-2 Line in inputs for FX return or other input sources.

-Stereo bus input stage architecture for maximum flexibility.

-Switchable to balanced direct inputs for use as a balanced line-driver/dedicated headphone amplifier.

-Transformer interstage accepts various input sources, and provides the negative phase signals for balanced or dual-output.

-Premium Penny & Giles 3000 series stereo faders for the mains level control. Just because I bought them really cheap on eBay, and they're cool. :cool: Actually, they're just my way of controlling a balanced headphone output without resorting to pricey stepped attenuators or 4-gang pots. In this configuration they give you balanced level control as well as individual stereo unbalanced control.

-4-board JISBOS line-drivers, configured with a little extra feedback to give a gain of 2 (unbalanced) or 4 (balanced).

-Neutrik Combo output jacks to drive XLR-terminated balanced headphones or other balanced line equipment. 1/4" outputs can be used for the same purpose, or, when the mode switch is engaged, can provide two separately buffered, single-ended stereo outputs for driving two pairs of headphones

With this design, you could run a stereo instrument into a stereo P.A. with complete balance and level control, run a stage amp/monitor and send a separate signal to FOH. Mix in other instruments, two completely discrete (and separately pannable) effects loops....there's almost too many applications to list.

It's a bit of a "trick" using the negative phase signal from the transformers to provide two separate single-ended signals. The cost is that the second stereo signal is out of phase with the first. This is fine (if not necessarily "ideal") for driving two pairs of headphones, but obviously I have to work out some sort of phase-reversal in order for any other on-stage applications to work.

I was thinking of using Rod Elliots parametric circuit:


It already has an input buffer (my NS has active pickups), so I'm thinking this should work just fine.

For the effects send/returns I was thinking of adapting this little circuit, but maybe using two discrete sends and changing the blending pot to a panning control.

The selector switch changes bypasses the instrument inputs, enabling the unit to function as a high-fidelity headphone amp, or alternately drive two unbalanced pairs of stereo headphones, each with it's own volume control. The last (unwired) position on the selector functions as a "mute" switch.

I'm thinking of adding an opto-compressor between the input buffer and the eq, and have toyed with the idea of also adding overdrive capability, though I'm unsure if the JISBOS buffer/amp is really suitable for this purpose.

Thoughts? Am I totally crazy? :boggled:


2010-10-04 5:25 am
Fun with jfets...

I found this great Nelson Pass article:

I'm thinking of mocking up the ESP parametric on breadboard, using a discrete jfet opamp (along the lines of fig. 16 in the article) in place of the input buffer, and perhaps as the make-up amp after the parametric as well. A few questions:

1.) I'd like to keep to 15v rails on this project (that's the recommended voltage for the JISBOS buffers, and several other circuits I'm considering). Mr. Pass mentions in the article that reducing the rails (with the appropriate bias corrections) from the 32v tested in his examples has a negligible effect. However, since the article's fairly old and I need find a substitute jfet anyway, does anybody have a recommendation for a preferred jfet that will spec out nicely at the lower rail voltage?

2.) (Angels-on-a-pin question ;) ) ...Just how large a batch of jfets do I need to test to be able to get one (or two) reasonably matched pairs?