> my box just suffers from the "original sound" and needs to be hit by disaster.
Looking back over the decades:
I don't think there is anything very wrong with the forward path. Like any other FET limiter, you have a resistor/FET voltage divider followed by a gain amp, buffer, output driver.
The 100K-123K divider top resistor is reasonable, tho a bit high compared to many studio boxes.
The amplifier looks like gain of 150, but NOTE it has both Voltage and Current feedback, is really intended to give 600R output and be loaded in 600R, and the actual voltage gain under nominal load is 75.
The amp looks old-fashioned, but has ample gain and linearity for non-audiophile work. I dimly remember messing with it, and did not find any obvious problem. The OT was perhaps a bit lame for the levels it worked at. You could bypass the OT, but you'd want to re-consider the NFB, and the amplifier has no short protection (originally the OT resistance and normal pad loss protected it against abuse).
BTW: that bridge-T output pad is useful.
If output threshold (before pad) is 3.8V, then voltage across FET at and above threshold is 0.052V, and 50mV is likely to be well within the linear zone of any common JFET. The 2N5459 is ~4V Vgs and ~~200R minimum resistance. Note that it is P-channel, not N-channel as we usually see.
It even has drain voltage compensation, the two 1Meg resistors at Q202 Gate. No, I have no idea why Q203 is there. This and other aspects make me suspect it is a quick re-design of older products.
The SLAM as it goes into limiting must be some transient in the rectifier sidechain. The time-constant network is 0.47uFd and 2.2Meg, very reasonable long-decay values. The attack time is not simply defined. It could be hundreds of mA (dumped into ground!). For more fun, everything is cap-coupled and will bob around as things get wild.
The base idea is: set the wiper of R224 just high enough to cut-off Q202. Then large signal rectifies in CR202/203, forces gate negative, FET turns-on and reduces signal.
There is considerable gain and uncertain (to me) threshold in the sidechain.
Putting 2K in series with each diode might tame the attack. (Better if you can bring both diodes together, insert resistor, then drive the node with C211 R223 and FET gates.)
You could redesign the sidechain. But single power rail frustrates many modern tricks.
If you like optos, it begs to have an opto where the FET is, and a LED-driving sidechain.
It is a very nice early-1970s box and knobs and meter.
The +45V can be changed to +36V by replacing CR105; that's how I ran chip op-amps in this box.
T1, mounted upside down behind left knob, is 600:10K, mildly shielded, moderate level. It is not very wideband, which is why I built a transformerless diff-amp for the second input.
The noise-gate is IMHO not worth touching. Lift one end of R202 to be sure the NG FET never does anything.
