Well, since my last update, I have tried about 35 new circuits or modifications to this amp, and I have aquired a second one as well. So now I am rocking two of them.
My expereince has changed quite a bit since my last post. For starters, running E34Ls in this amp became a problem for me. I did the conversion correctly, modifying the bias circuit, jumping the proper pins on the tube sockets to accomodate the E34Ls, upgrading the screen grid resisters, etc. But I started having trouble red plating tubes. I initually thought it was because the added current of the E34Ls were causing doides to reach thier threshvold voltage and fail, but to be honest, I am really just not sure. I was going to try to run KT-77s instead for the similar sound with less current draw, but then I thought better of spending big bucks on new tubes when I have tons of 6L6s everywhere. So I went back to the drawing board, re-installed the stock Sovtek 6L6s, and then just tuned the preamp to get a similar sound to the E34Ls but with the stock tubes instead. That ended up working out great, now I have a killer sounding amp, with the stock power tubes which are about half the cost of E34Ls.
Some of interesting items of note: As mentioned, through pspice modelling, tuning by ear, and just personal experience, I have tried about 35 changes to this amp, both in the power secion and the preamp section. Some of those changes were entire new circuits, some of them were just component value changes, and some were "system" changes - meaning changing several complonent values based on a model to produce a desired result - the general idea might be changing everything possible to get the correct vaules so that the frequency response "should" match that of a desireable amp like and Engl or Uberschall. Well, of all of these changes, I have reversed all but three of them, in both of my amps. I find that modelling a circuit in a computer program is cool, but still does not usually match up to what you are trying to pull off, because I am not familiar enough with modelling at this point to capture nuances like compression based on plate voltage, etc. Mostly just frequency response is what I use modeling for. Anyway, the point is, I have tried lots of different things, but in the end, the stock circuit topology has produced the best results, with just 3 small exceptions.
1) I added a high frequency roll of circuit, just like I do to just about every bright amp I have. But, this circuit has evolved. I used to just use fixed vaules and break it out on a switch, but I now control it with a potentiometer so it truly acts like a presence control. This amp NEEDs this circuit and greatly benefits from it.
2) I added a little low end through some component value changes in the preamp.
3) Gain 1 on this amp sounds different from Gain 2. Gain 1 is thinner sounding, and more classic rockish with humbuckers. Gain 2 is thicker and more modern sounding, but also sounds great at low gain with a strat for thick blues tone. Gain 1 was just too thin to run single coils through to my ear, so I really wanted Channel 1 to sound just like Channel 2 (thicker) so I could kind og assign channel 1 to be my "strat" channel, and channel 2 to be my "Les Paul Hi Gain" channel. If I can make the channels close to identical, I would still have the Gain control for each channel to dial in the tone I wanted. So, I made a modification to get these channels to be the "same", which involves the same optofet that Bruce employed in the design of the MTS modules. THis made a big difference and gave me the thickness I needed in Gain 1, but I still came up a little short. The channels are 40% more identical than they were stock, but not exact yet, so I think I am still missing a gain limiting resistor or something somewhere that is different between the two channels. Actually, they are not really channels, they are more like 2 modes within the same channel.
Outside of these three circuit changes changes, both of my AT-100s are back to stock (except tubes - running JJs in the preamp and a NOS RCA 5U4GB rectifier tube in one, and a NOS WInged C 5U4G in the other). I play them everyday and love them. They meet the spec for me, which is, the amp must sound great for texas style blues with a Strat, and high gain stuff with a Les Paul. This amp delivers. I would like to someday invest in an upgraded output transformer for one of them, but I just can not justify the cost since i just use these amps as fun projects and learning tools.
Other items of note:
It has been a struggle to get more low end out of this amp without totally reworking the entire power section and filter caps. I have tried to "open up" the preamp to let as much usable bass pass through as possible, but I think the power section is the bottleneck, especially since this amp only uses one rectifier tube. Alot of people would disagree with me on this due to the "Low Res Control", but that controls damping and "stiffens things up" when adding more low end, so I try to keep the res control set to 0 to keep the amp as loose as possible. It sounds so good loose, and tube rectified, but you sacrifice a little low end in doing so. I have corrected this to 90% satisfaction with just small preamp changes, but one of these days I might add some real low end to this sucker, without leaning on resonance control.
I notice that Madison Amps are almost an exact clone of the B-52. I have never actually heard one, but I can tell due to the layout - it is the AT-100 in a different skirt. I see that the Madison also sports EL-34s. Anyway, interesting.