Bogen CHB-35A Conversion

Okay, so I've gone through a ton of revisions on this thing. It's been quite an undertaking to design an amp from scratch, but once it's built it's going to sound fantastic and the next one will be much easier.

Six knobs? Plenty.

I got this amp for $30 off of Craigslist. I don't know if it was turned on without a load or what, but the OT was fried. Shouldn't be a huge problem finding a 6.6k primary OT that can handle a pair of 7868s, right? Turns out I found an 8.4k primary OT  (from Musical Power Supplies) which was much cheaper and looking at the datasheets that just means it'll be a hair louder and a hair more distorted. Oh no!
But what about the increase in screen grid current at overdrive? Bigger screen grid stoppers, of course. It did take me a while to wrap my head around sliding screen operation, but it turns out 1k screen grid stoppers should be fine. From my limited experience frankensteining a single ended EL84 amp, I know I'm not particularly fond of the change in overdriven tone from excessively large screen grid stoppers, but I think these will be reasonable. 

Original guts.

Oh, if only I'd listened.

There are a handful of oddities already in the stock schematic. Well, not "oddities" so much as "deviates from Fender/Marshall." The rectifier is a voltage doubler, the bias supply is capacitor-coupled, there's a lone 7-pin triode as a cathodyne phase inverter and the choice of output tubes is unconventional.

The PT really calls for a voltage doubler and that hasn't been unreasonable to work with. I changed the novar sockets for the 7868s (9-pin the size of an octal socket) to octal sockets for 7591s. Same tube, just with a different base. The biggest difference is I have a pair of UOS 7591s.
Old stock 7591/7868 tubes are rare and expensive and there are a handful of urban legends about why. Some say there were whole warehouses just dumped because these tubes were developed late in the tube era and transistors were taking over. Others say speculators bought up every tube they could find for the Japanese audiophile community.
It doesn't matter, really; what does matter is that these bad boy beam tetrodes can put out up to 40W a pair yet only require about -20V bias. They're often described as a cross between a 6V6 and a 6L6, and apart from some Ampeg amps they're largely overlooked in the guitar amp world.

The bias supply has been giving me trouble, so I decided to switch to cathode-biased, at least for the time being. When they say capacitor-coupled bias supplies can turn into charge pumps if they're not sufficiently loaded, they aren't kidding! God bless lightbulb limiters.

The latest version. Pretty sure there have been more than 7 of these.

For a while I was thinking about making this a two-channel amp, but I decided I'd rather just have one great channel instead of two compromised ones. Also I haven't messed around with relays and optoisolators yet, so all channel switching would have to be done externally. If I could start with a different PT, maybe add a couple more triodes for the dirty channel and mixing the two channels, I'd consider it.

I find a whole lot about this design exciting (as well I should, considering I'm designing the damn thing.)

• So far I have an irrational love of the cathodyne phase inverter. It's in the 5E3 Tweed Deluxe, the old Orange graphic amps, the Fender Princeton - anywhere this PI goes, awesome seems to follow. Super easy to design, and Merlin's warned to stuff it (and the output tubes it pushes) with huge grid stoppers, so that's what I've done. Now that I've got an oscilloscope I might jumper them to see some nipple distortion, but I'll probably just take his word for it. 
• I'm totally sold on directly-coupled cathode followers when it comes to overdrive. I'm somewhat afraid that this bootstrapped arrangement might make way too much gain, but I can always put that on a switch later. 
• The tone stack is a tweaked FMV with no mid pot because I've only got six knobs and that thing is useless. It's got much less mid cut, and the "shift" control moves the mid notch to a lower frequency as well as making the Q wider. This stack won't get the blackface Fender tone (who'd want it?) but it should be very useful for both clean and overdriven tones. 
• Parallel input stage! Smaller grid and plate resistors for (theoretically) less noise right where the S/N ratio is at its worst! Two different bias points (warm & cold) for "complex" harmonics! Two different bypass caps for a multi-level mid & treble boost! Below ~100 Hz the gain is lowest, slightly higher between 100 Hz and 300 Hz, and highest above 300 Hz. Should be pretty awesome. Why not just give each triode its own input and add a switch to bridge them, you say. Well, I'm building this amp for myself and my pedalboard output is stereo so I want two identical inputs if need be.

Heater wiring!

I've used 22 AWG teflon-insulated 630V silver-plated wire (from Apex Jr.) for everything.  The math says this gauge is actually more than sufficient for my heater draw, and the heaters were originally wired with 24AWG anyway. Teflon is a bit of a pain to strip and it doesn't like to hold a bend very well. I intended the hot glue to be a stopgap but I think most of it will stay put even when the amp gets hot so I'll probably leave it.

The heaters work!

 

Progress so far.

 Gotta love point-to-point - with only the power supply wired up it's already looking like a rat king. I've actually abandoned this "true star" grounding system in favor of a "local star" system which makes a whole lot more sense and cleared out a lot of wire.

Since I took this picutre, I got the power tubes, bias supply, and phase inverter completely wired but I'm going to take the bias supply out (or at least disable it) in favor of cathode biasing. I went through the trouble of load lines and, y'know, ALGEBRA to figure out what the bias resistor should be and it's actually the same as what the data sheet calls for.

Now I'm just waiting for parts. I still need to order a bunch - all the pots, the bias resistor, the heater elevating resistors - I should check that cathode follower again; I might be okay with the 21V of elevation I'll get from the bias resistor. And then, of course, I have to finish building this thing... and then debug it... and then tweak it. GONNA BE A WHILE.