OK, I'm done! The one on the left sounds like the one on the right. I've concluded that the Dr. Robert is mainly an EQ and a boost. Sure, you can get some dirt in Mach Schau mode, but most of the sweet distortion you hear in Dan's demo comes from Dr. Robert overdriving his amp.
I figured a few things out, thanks to Dan loaning me his Walrus pedal. End-to-end gain and output levels are very similar for both units. The TNK actually goes louder than Dr. Robert, particularly when BOOST is cranked. I took a few parts out, changed several parts and added two parts: TR1 & C40. C40 is silver mica, all the rest are film or aluminum electrolytic. No more charge pump, no more aluminum caps in the signal path. Q1-Q3 can be your favorite NPN Si transistor. 2N2222 can get noisy. MPSA18, 2N5088/89, BC549C or BC550C are good subs. Or go with BC109C if you want the vintage look of a metal can transistor. Use the JFETs of your choice, just shoot for Vp around -1.5V. I adjusted the trimmer to get 4.5V on Q7-S with BOOST at noon. Dr. Robert v1 had two trimmers that are not present in V2. I suspect they did what this trimmer does. Dr. Robert uses 1M resistors for R32 & R33. Those interact with Q6's gate-drain capacitance to roll-off the top end. That capacitance varies a LOT from one JFET to the next, so I installed C40 and reduced R32 & R33 so that the treble roll-off is more predictable. The first stage gain in the Dr. Robert is low, about 9dB, so I left that alone. I cut the gain of the 2nd stage by 6dB so we don't run out of headroom. I cut the gain of the 3rd stage by another 6dB for the same reason. The lost gain is recovered after the output stage because unlike the Dr. Robert, I don't cut the signal down by 11dB between Q7 and the VOLUME control.
I'm not suggesting you do all this stuff to the Dr. Robert; it's a great sounding pedal. This mod is something I did for Cooder because he wanted a modd'ed Dr. Robert design. A big part of the Dr. Robert's marketing is it's sold as a carbon-copy of the Vox UL730 preamp's clean channel. What I did kinda deviates from that, although the differences are subtle and mostly have to do with reducing the tone control interactions. Since the JFET output stage makes the lion's share of the distortion, I think it's entirely possible to replicate the freq response using opamps for the first two stages and let the JFET output stage do the rest.
Here's the signal with GAIN, TREBLE, MIDS & BASS at noon, VOL dimed, BOOST at zero. Green trace is the input from the bridge pickup (Fender Tex-Mex) on my Strat. Low E chord with a moderate strength strum. Blue trace is the output. This is pushing the limit of the max clean signal you can get out. 4.25Vp-p is more than enough to overdrive your favorite amp! Both TNK & Dr. Robert make 3Vp-p without a hint of distortion. Above that, you start to get some 2nd and then 3rd harmonic in the output stage on both units.
Here's the most signal and distortion you can get out of the TNK. TREBLE, MIDS & BASS at noon, GAIN, BOOST & VOL dimed. Clipping is very evident. This is almost twice the output you can get from a Dr. Robert. Backing off of the GAIN and/or BOOST smooths out the distortion. If this is not enough dirt, then lower R35 to 220Ω or 100Ω.
