Fixes for the Median compressor

[jubal81]

I was looking at the Median schematic and I see a few changes that would make it perform a lot better and be easier to bias. Bonus: It would only take a couple more resistors and caps.

There's no reason you'd need matched JFETs or have to select ones with the VP and IDSS Wampler used. The 5457 isn't really ideal here, either. In this case, a FET with higher VP and IDSS would perform better as a variable resistor.

First job is fixing the feedback on the gate of the compressor JFET. Feeding back half the drain signal helps tamp down distortion, but taking it directly from the drain is not ideal. If you look at IC2.2, it's a 100% pointless buffer. Let's put it to work buffering the feedback.

Next, let's fix the JFET buffer on the source of that first JFET. We can add caps so their DC voltages don't interact, and getting rid of the trimmer on the source means we can use equal resistors for feeding back exactly half the signal.

Now, we can move that trimmer to directly bias the compressing JFET, which opens up using any JFET in there. I drew it up taking the voltage from VREF, and if you do it that way, you'd need a JFET with a VP less than 4.5V. I think ideally it'd be best to use a 5V regulator so the bias would never be impacted by input voltages and you could happily plug in an 18V supply.

That keeps the parts count almost the same. If you'e still getting distortion with humbuckers, you could add a switch to cut the signal in half going into the compression side and make it up on the mixer side.

Anyway, here's the original schematic and my drawing showing the changes.

 

[Gordo]

This one never sounded right to me, although I've never played the original. Looking forward to tearing into it.

 

[jubal81]

That JFET buffering the source of the first one is still bugging me, especially how the buffer signal goes straight into that big cap to ground.

I looked at the 1176 schematic and it feeds back half the drain signal to both the gate and the source, but there's a 10K resistor on the compressor JFET source. I think that's another one to add to this design to get that feedback signal to the source and not spilling into that cap to ground.

 

[mckillacuddy]

Following along here. Don't know much about compressors. Don't use them, don't know how they work, but I like to learn. Someone posted up a Morris Klein video around here and I quickly went from: cool accent guy in the background, to: I need to learn everything I can about this shit

 

[jubal81]

Spotted something else on the 1176 schematic that could help a lot.

To make the compressor JFET happy, we buffer and feedback half the drain signal to the source and the gate. There's another trick in the 1176 that helps tamp down distortion a lot. It's another feedback path (negative this time) to the drain of the VCR JFET. This makes a big reduction in the voltage difference the JFET must deal with.

In the 1176, the stage following the VCR (voltage controlled resistor JFET) is a transistor that does double duty by giving negative feedback to the drain of the VCR and positive feedback to the source.

First, we see the 1176 bipolar transistor feeding half the positive feedback to the source through C2. The inverted signal from the BJT collector is buffered by Q3 and the negative feedback reaches the base of Q2 through R11, further reducing the strain on the VCR (Q1).








Looking back at the Median, we can move the "buffer" JFET to take its signal from the drain of the VCR instead of the source (like the 1176). A cap in series with a resistor from the drain back to the gate gives the negative feedback.

I haven't breadboarded any of this, so the schematic I'm posting is just conceptual and would need the details worked out. This change keeps the parts count the same, but adds another trim pot to bias the second JFET.