So I managed to wire up most of the output section between clients lol. I almost forgot about old school soldering, it's been so long. It's a labor of love, you gotta cram lotsa wires into those little terminal strips, and you end up with loose strands and little bits of solder all over the chassis.
Gotta work this week, the stimulus checks must have arrived, suddenly there's a bunch of starving artists beating down my door. But I'm getting it done. Waiting for some #6 hardware to arrive in the mail to mount the cap cans, after that I'll post some more pics. As planned, the DC stuff comes first, for measuring and tweaking voltages. So no coupling caps yet. You can only see the "underneath layer". (One would think 23x10 is a big roomy chassis, but no, things are getting tight already).
As this build proceeds, we can talk about the sections. The output tubes are self explanatory, not much going on there. I use 5.6k grid stoppers on the KT-88's like Marshall does, and 250 ohm screen resistors in series with the ultralinear taps from the output transformer, those are important to prevent current rushes when you push the amp into overdrive at full volume. I also prefer Marshall's bias circuit to Fender's, because if the bias pot gets dirty or the wiper lifts you want full bias on the tubes, otherwise they red plate and you probably lose your OT too.
As already discussed I'm using a 12AU7 driver stage like the Marshall Major. The interesting thing here is Marshall uses a 1.5k cathode resistor, which is about double what it should be, and it turns out to be a sensitive adjustment.
We want the driver's output voltage to swing about 100 volts from the midline in either direction, which means ideally we'd like the 12AU7's plate voltage to be sitting around 200 volts (say, between 200 and 220). You can assess the component values and see what the load line looks like by using the AmpBooks vacuum tube calculator:
If you plug in 450 volts and a 47k plate resistor and push the green button, it tells you the baseline cathode resistor is 1.5k. But that only for one section, if you're sharing the resistor between two sections you get double the current which means you want half the value. If you recalculate with 5 mA plate current and -10 V grid bias your plate voltage goes to 205 but your cathode resistor becomes 2k, and half of that by sharing is 1k. So Marshall's value is too high, it looks like it should be between 750 ohms and 1k. We could try 820 ohms, that's a standard commonly available off the shelf value. (Sharing cathode resistors is why you find 820 ohms in the Twin Reverb, that's the closest off the shelf value to "half of 1.5k").
So after setting the supply voltages and biasing the power tubes, that's the next thing we're going to look at. It's important that the driver stage be working correctly. We don't want any distortion here, least of all blocking distortion, and it has to handle a clean bass guitar sound as well as a dirty lead guitar sound at full volume. This stage has a gain of around 10, so to test it we run a 5 volt sine wave through it and make sure it doesn't clip on either side. Total current draw through this stage is 2 x 5 mA give or take.
This is an inverting stage that sits between the PI and the output tubes, so that means we have to invert the negative feedback signal that feeds the PI from the output transformer. The easiest way to do this is simply switch the coupling caps between the PI and the driver.
Also since the gain from PI to OT is so high in this circuit, we'll have to adjust the value of the NFB resistor. You can see how this works by playing with the AmpBooks negative feedback calculator.
www.ampbooks.com
You can enter the following values, and push "recalculate" after each entry.
LTP voltage gain = 60 (6 for the PI times 10 for the driver)
RMS power output = 100 watts
Transformer feedback tap = 16 ohms (that's the only thing 16 ohm taps are good for, no one uses 16 ohm speakers anymore)
Power amp input headroom = 50 volts
This gives us a feedback resistor of about 315k, so 330k is probably the closest off the shelf value. This is for a nominal 50% reduction in gain through the output stages, which is supposed to (in theory) give you pretty good linearity.
We can verify this as follows: begin by using a 1 meg pot in place of the feedback resistor. Begin with the pot all the way to the right, and a sine wave input of 1.6 volts which will push the output stage into overdrive. (Do this with the dummy load connected to the 8 ohm output, it'll be handling a full 100 watts). Now back off on the NFB pot until the output signal barely stops clipping. This is about 50% negative feedback in theory (since full power is achieved with a .8 volt input in the absence of feedback). Now turn the amp off and disconnect the feedback circuit and read the resistance value of the pot with a meter. If theory is correct it should be in the neighborhood of 315k. This is the value you want for your feedback resistor. With this value you should get good performance from the presence control.
Okay? We can talk about the PI in more detail tomorrow, it's getting late and I have to be in the studio at 5am in the morning. (A chick singer wanted an early morning slot, go figure). We want the PI cathode voltage not to exceed 90 volts, which is a bit of a discussion relative to a long tailed pair. (The Twin Reverb exceeds spec by 20% or so, which is why the PI is usually the first tube to go).
