DynaWatt Power Amp

[GainFreak]

Hi,

I'm looking at the schematics of those Mesa amps like Studio caliber, DC-3 and 20/20 and I can't get it how is this phase inverter supposed to work? It doesn't look like any that I know or you can usually see in a guitar amp? One side is driven by the input stage but what about the other phase?
I'm not very much into theory so would someone explain please how it works or it's just another wrong schematic?

[teemuk]

It's basically just drawn confusingly, as well as deviating from the "basic" LTP circuit by having negative cathode voltage reference.

Still, the PI's cathodes are connected to each other and the 15K resistor isolates them from the DC bias supply. Basically, the schematic confuses you because it shows the cathodes hooked to -2.5 DCV, while in reality there's a resistor in between, allowing AC signal swing at the cathodes. (The voltage polarity is also indicated incorrectly but looking at the bias circuits reveals the error).

So, it still works like an ordinary LTP phase splitter: the cathode of the "input" half drives the cathode of the other half, which in turn acts as a common grid amp. The "input" common cathode side of the LTP is inverting, the common grid side of the LTP is non-inverting, so as a result you get phase inversion.

Normally a common grid circuit would ground the grid for AC signals but since negative feedback is used its inserted to the grid.

There you have it. It's not so strange after all and really not that much different from the generic LTP circuit.

[GainFreak]

Thanks for the quick answer. Now it makes sense.
I have an EL84 amp that I often modify and I'll try this arrangement. I have individual bias controls for the tubes like the one from the schematic attached. I'll use them for the power tubes but what about the PI? My PI is the regular LTP found in most amps. Should I add one more bias control with the taper unbypassed or just simply use a larger resistor to ground and how to determine its value?

[GainFreak]

Here's what I did. I added one more bias leg similar to those from the schematic and attached a 15k resistor to trimpot's taper. Dialed +2.5V at PI's cathodes. It worked without problems.
However there was a nasty low level distortion when the amp was cranked up a bit. I don't know why but that 1k5 looked suspicious to me. As far as I can see it forms a voltage divider with the 56k setting up the amount of feedback going to the PI. I used a trimpot instead. It turned out that there's a sweet spot around 1k where the distortion in question disappears and the sound becomes very "round".
With -11V bias at power tubes grids current draw is ~31mA of which ~3mA from screens. Anode voltage was 358V, screen 307V which makes ~10W dissipation (83%) which seems a bit too much. I rebiased for ~8W dissipation. I suppose with 380V as per schematic and -11V at the grids the dissipation will reach 100% maybe more.
What about the sound? I can't notice anything special about this amp at least for now. I'll have to play with it some more time. There's a "deep" mod I found while reading in the net about this amp and it consists of making the input stage capacitor switchable. Maybe I'll try this one later.

[teemuk]

Dialed +2.5V at PI's cathodes

I suppose you mean -2.5VDC not +2.5VDC. I don't know why Mesa omits the negative marking from the schematics but the cathode bias should be negative, that's the entire point of that kind of biasing arrangement. Also, a quick glance of the Mesa's bias circuit should reveal the correct polarity.

As far as I know the point of the negative cathode voltage bias on PI is to improve the symmetry of the phase inverter. With +2.5VDC the symmetry would be very skewed resulting into distortion.

What about the sound? I can't notice anything special about this amp at least for now.

Frankly put, I don't know why you even should. It's a pretty basic EL84 amp and it even has quite a high damping factor so its likely less "tubey" sounding than many other EL84 amps. The whole "Dyna" thing is just a fancy name for adding series resistance to the rectifier to increase power supply sag. The most impressive thing about the design is that Mesa actually managed to patent that age old idea.

[GainFreak]

I suppose you mean -2.5VDC not +2.5VDC.

No, I dialed +2.5V. Even when resistor is directly connected to the -20VDC supply it won't go below 0V. With another 15k in parallel (7k5 total) voltage drops to +1.5V. With a 4k7 resistor in parallel (~3k6 total) it goes below 0V but it starts humming. Changing PI's bias influences plates' voltages which now (+2.5V) are 267/274V and below zero they drop down to 90V. At 2.4V for example anode voltage goes down with 10V. It looks like it should be +2.5V. I also looked at Mesa's other amps schematics that have this power amp and they are the same with the voltages slightly varying.

As far as I know the point of the negative cathode voltage bias on PI is to improve the symmetry of the phase inverter. With +2.5VDC the symmetry would be very skewed resulting into distortion.

I don't have a real scope except on my PC but I'll take some measurements to see how it behaves.

The whole "Dyna" thing is just a fancy name for adding series resistance to the rectifier to increase power supply sag.

Considering Randall Smith mania for patents I already had that in mind but was curious to confirm it for myself.