Two preamp channels in parallel, WITH NFB?

[Badside]

So I'm gearing up for something I've wanted to do for many years: an amp with both a 2204 (Marshall MV aka JCM800) preamp and an AB763 (Fender Blackface) Normal Channel preamp.

I want both sides totally parallel and on at all time (dedicated inputs) so I can use an A/B switch (as if I was using two amps, but they share a power amp)

The trick is I want to use NFB. So they have to mix together just before the PI's input cap (can't use the PI's 2nd input).

Fender mixes channels on the AB763 by simple mixing resistors (220k's).

I could do that...

I'm just worried the AB763 channel might interact with the 2204 channel's tone stack and MV, or vice versa. E.g.: reducing the MV on the 2204 side will drag down the AB763 side, and the AB763's AC path to ground via the plate resistor will affect the 2204's tone stack response. Also, the mixing resistors interacting with the PI tube's capacitance could reduce highs (more of an issue on the 2204 side since it will be in series with the MV's resistance when not at full... something I could address by making it a smaller value pot)

So I might be overthinking this

My options as I see it right now:
- Stick with the simple tried and true approach of mixing resistors, interaction won't be that big of a deal
- Put a MOSFET in there (after the MV) as a pseudo cathode follower
- Throw a tube at it (as an actual cathode follower after the MV, a bit overkill though)

Thoughts?

[Badside]

Just saw that at least one version of the AC50 (Mk3 I think) mixes two channels with full tone stack before the PI by simply adding a 470k resistor after the treble pot wiper (no MV).

So I'll probably just do mixing resistor, start with 220k's, along with a lower value MV on the 2004 side (either 250 or 500k).

I don't like PPIMVs (don't find PI distortion that enjoyable), so the regular 2204 MV is a must for me.

[tubeswell]

Yes, where the two channels mix back together it is simplest to use mixing resistors, or if you have an LTP inverter, you can feed one channel to each LTP grid and use the LTP as the mixing stage.

Meanwhile at the input ends of each channel, the easiest thing by far would be separate input jacks for each channel that can be jumpered with a patch cable or a switch.

[pdf64]

For me, a key part of the Marshall thing is that the presence control be set fairly high; have you tried the Marshall pre-amp into a BF Fender type power amp, with its flatter response?

Note that the tone stack response is affected by the loading on it, eg the suggested 250k master vol and mixing resistors. Consider modelling things in TSC or on a test bed amp, prior to committing yourself.

[Badside]

Yes, where the two channels mix back together it is simplest to use mixing resistors, or if you have an LTP inverter, you can feed one channel to each LTP grid and use the LTP as the mixing stage.

Meanwhile at the input ends of each channel, the easiest thing by far would be separate input jacks for each channel that can be jumpered with a patch cable or a switch.

Can't mix at the LTP because the 2nd grid is used for NFB (negative feedback).

Yeah, the plan is to have separate input jacks and use an A/B switch pedal to alternate (jumpering them might sound pretty bad since they'll be out of phase with each other). This way I can also do stuff like put the overdrive pedal past the A/B switch in the dirty (2204) chain.

[Badside]

For me, a key part of the Marshall thing is that the presence control be set fairly high; have you tried the Marshall pre-amp into a BF Fender type power amp, with its flatter response?

Note that the tone stack response is affected by the loading on it, eg the suggested 250k master vol and mixing resistors. Consider modelling things in TSC or on a test bed amp, prior to committing yourself.

I like my Marshalls with less NFB than stock, but some NFB. I tend to replace the Presence control with a global NFB control.

Good call about the tone stack response. I did some simulations in TSC of the stock 2204 (1M volume straight into PI) vs a 500K volume in parallel with the AC path to ground via the AB763 channel (the 2 220k mixing resistors plus the 100K plate resistor) and it's fairly similar except for some loss of gain and like 1dB loss of low-end (which can be compensated by using a bigger capacitor).

[Smokebreak]

If your goal is to switch channels with a pedal, why not just use relays and mike it a true channel switcher? Then you don't have to mess with mixing, which will be a compromise IMO.

[Badside]

If your goal is to switch channels with a pedal, why not just use relays and mike it a true channel switcher? Then you don't have to mess with mixing, which will be a compromise IMO.

Call me old fashioned but I like an amp with dual inputs and no foot switching of anything.

I'm reconsidering making it a 2 preamps design though, will perhaps go for a classic Plexi-type preamp (but with MV)u with an extra tube for extra gain on one of the channels (meaning mixing will be ahead of the cathode follower driver). The shared EQ never bothered me on the DSL50.

The compromise in this version will be the shared MV, but I've been working my volume knob on single channel amps for so long, I can deal with that.