It seems that the new SVT like to eat their 12AU7 Drivers

[bob p]

PS: The version of the story I heard was that Ampeg made the output transformer a bit smaller in the SVT2-Pro than the original SVT, because they felt that a little transformer saturation improved the sound. It's still hardly "lightweight".

The version that I heard was similar, but a little different. Users were comparing the SVT-2 to the SVT-2 Pro. Every discussion focused on the weight differences of the amps, and none of the conversations approached any technical issues. Someone made a knee-jerk response was to imply that the transformers were responsible for the difference, people accepted that idea, and it became SIL (sticky internet lore).

I think that the actual difference in gross weight was attributable to several factors: I think that the sheet metal in the original SVT-2 was of a heavier gauge; the faceplate of the SVT-2 was heavier and of uniform thickness, while the SVT-2 Pro faceplate is scalloped in relief; the SVT-2 Pro has additional metal removed for the EQ; and the transformers didn't seem to be significantly different in size to me. Personally, I think that the box itself makes up for a lot of the difference in the weight of the 2 vs. the 2 Pro, but you know how the internet works -- one guy says that the transformers are smaller and that the original NJ Ampeg amps sound better, and to hell with the facts, you've got a sticky Internet story that's impossible to refute.

With that said, the first time that I plugged into either one of these amps, it was instant gratification -- the result was instant unmistakable Ampeg SVT Tone. In spades.

R.G.. I'm all ears.

[exclamationmark]

Apologies if I'm reading this wrong, but there appears to be 425 volts from anode to cathode at idle on the driver tubes (380 + 45), which I assume is not a good thing (and it probably increases to 500v++ on negative peaks). Usually on my source followers (or sand followers as I like to call them ) I run the positive rail from the lowest B+ I can get away with, so I don't have to heatsink the thing. I imagine dropping the plate voltage down from 380 to the 200's would probably help the poor 12AU7 live longer.

[bob p]

No doubt about it, there are lots of volts across those poor little AU7 -- more than their 300v maximum ratings. That's one reason that the BH7A looks better to me -- in oscillator & deflector type circuits It's rated for 450 volts or so.

[R.G.]

Hmmm. Good point about the voltage. I wonder what happens if the plate supply to these things is dropped to about 150-200V. Might be doable with a series zener string, something like two 100V 1W zeners in series from the +380 supply. That would leave it 200V to go up and 180V to go down. What's cutoff on the output tubes?

[jazbo8]

I think we all agree that lowering the plate voltage could make the CFs operate safer, but does anyone know why the Ampeg's were designed that way in the first place? The Dumble SSS also has pretty much the same configuration with >400V across the 12AX7 PPICF, and the same issue with heater-to-cathode voltage at turn on, etc. I guess I am asking, does putting higher than spec voltage across the tubes provide any sonic benefits?

[bob p]

I agree -- it'd be good to understand why the voltages are that high before we go tampering with them. The only problem is that I don't have a clue why they designed the voltages to be that high. For all we know that's where the SVT vibe comes from and if we go tampering with it the amp might end up sounding bad.

[R.G.]

I think we all agree that lowering the plate voltage could make the CFs operate safer, but does anyone know why the Ampeg's were designed that way in the first place? The Dumble SSS also has pretty much the same configuration with >400V across the 12AX7 PPICF, and the same issue with heater-to-cathode voltage at turn on, etc. I guess I am asking, does putting higher than spec voltage across the tubes provide any sonic benefits?

Good question.

Using higher plate voltage on a triode lets you use a higher plate resistor, which gives higher gain. That may matter to the gain side of the 12AU7, but I don't think it matters in this circuit to the follower. It lets the tube drive more current out of the cathode, which may matter here, since the output is the pull up on the follower. It sure makes them run hotter.

In this circuit, the real phase inverter is a split load type, in V1's second section, pins 6,7, and 8. As a side note, the schematic is incorrect. The grid of that section cannot be sitting at 5.7V. The 68K plate resistor is dropping 95V, and so the 68K resistor in the cathode circuit has to be dropping the same amount. The grid has to be at about +95V.

The signal from the PI is run into the two amplifier sections of the two 12AU7 driver tubes. The first sections amplify the signal, and the second sections are followers for driving the output tubes from a low impedance.

The output signal into the output tubes is twice the bias voltage, since the bias voltage is the zero-signal DC on the grids, and a peak voltage equal to the bias voltage drives the output tubes just to the edge of grid conduction. So the signals on the follower outputs is not much more than twice the bias voltage, about 90V peak to peak, plus whatever they intended to use to drive the output tube grids into conduction. That can't be much, maybe another 10-20V. The schematics say "30V" in a box; maybe this is the AC rms value. If so, that corresponds to a peak value of 42.42V, which matches nicely with the guess about the peak being equal to the bias voltage.

That taken together tells me that they didn't intend to drive the grids positive much, if at all. And the 47K grid stoppers - are they REALLY 47K?? - mean that there's not much current being used to drive the three output tubes, even if you do manage to abuse them positive. Still, it's a 16K load, about, if you do. My inclination from a quickie read of the schemo is that they used the CFs for lowering the impedance of the wires to the output tubes.

We're pretty sure that the cathodes won't go much above 0Vdc, at which point the current in the tube will be 180V/47K = 3.8ma, plus whatever is being used to charge the stray capacitance of the cables and/or the input capacitance of the output tubes. I eyeballed the plate curves of the 12AU7 in my Sylvania data book, and it looks like a decent bias point for it to be moving 4ma is about -10 to -12V on the grid and 200V on the plate. More plate voltage makes it more linear.

With no more knowledge, I'd guess that they used the high voltage to allow the follower to drive enough current. Flashover isn't much of a problem, as the 12AU7 is rated to 1200V peak in vertical deflection amplifier use.

I think this reinforces my opinion that a power MOSFET would make a good substitute for reliability reasons. It could handily run from the +380, although that would make it sit at about 1.7W of idle dissipation, so a heat sink is probably needed. Its gate would sit at about 15-20V more positive than the grid of the follower, so some good looking at the bias voltage string would be needed to be sure there was enough room there to adjust it in. A quick calculation puts the bias pot adjustment range at -51V to -89V.

If I were subbing in MOSFETs for the follower, I'd parallel that 68K to ground in the bias divider strings with 130k or 140k, to open up the bias range to about -38V to -90V. That will let the bias voltage get positive enough to keep about -45V on the grids of the output tubes, and it's conservative on the negative side. Be careful if you try this - the tubes you save could be your own.

I might put 100V to 200V of zener in the drain side of the MOSFET to eat away some of the power dissipation. A 100V/1W zener would dissipate about 0.4W, and that comes right out of the power dissipated by the MOSFET; two of these gets the MOSFET dissipation down to 0.9W, and you're getting into the range where a TO-220 doesn't need a heat sink.

A MOSFET would need a gate stopper of about 100 to 1K right at the gate, and a 12V zener from source to gate to protect the gate.

It's worth a try. I might try this device: STMicroelectronics STF5NK100Z
It's insulated in a full-pack, with 2500V of insulation, so a metal heat sink bolted onto it will not be hot with the drain voltage. A couple of square inches of aluminum and some heat sink goo would probably make it golden.

[R.G.]

I agree -- it'd be good to understand why the voltages are that high before we go tampering with them. The only problem is that I don't have a clue why they designed the voltages to be that high. For all we know that's where the SVT vibe comes from and if we go tampering with it the amp might end up sounding bad.

Fortunately we have your delicate, educated ears to do an opinion...

[bob p]

> The schematics say "30V" in a box; maybe this is the AC rms value.

If you look at the proper schematic that Enzo posted, rather than the screen capture snippet that I posted, you'll find the following statement in the legend:

"Voltages in rectangles are RMS signal voltages with 0.4v in. All other voltages are DC in conditions stated."

[R.G.]

That makes me more inclined to try a better buffer with more current available.

[Steve A.]

For longer life (and extra geek points) you could try replacing the driver 12AU7 with a JJ ECC99. It's just a cathode follower, so changing to a different tube type won't affect the bias point too badly.

Tell me more about the JJ ECC99 tubes- are there any good applications replacing the 12A_7 tubes in the common guitar amp designs? Can they work well as preamp tubes?

Thanks!

Steve Ahola

[bob p]

Steve, here's my take on it: the ECC99 is close to a 12BH7.

You should be able to use a 99 anywhere a BH7 would work.
I guess a more accurate thing to say is that their plate characteristics are very similar.
It's not really considered an A_7 substitute, although in this case we're talking about subbing it for an AU7 in a case where an AU7 was probably subbed for a BH7 due to availability.

http://www.jj-electronic.com/pdf/ECC99.pdf

[bob p]

> That makes me more inclined to try a better buffer with more current available.

Well, the schematics get confusing now.

Looking at the amp's input jack, the boxed voltage is 0.011 V.
Looking at the preamp out, the boxed voltage is 0.400V.

Looking at the legends, the explanation is not consistent:

The legend on the preamp says: "Voltages in rectangles are RMS signal voltages. Other voltages are DC in conditions stated."

The legend on the driver board says: "Voltages in rectangles are RMS signal voltages with 0.4v in. All other voltages are DC in conditions stated."

The legend on the power tube board says: "Voltages in rectangles are RMS signal voltages with full power output. Other voltages are DC with no signal."

So it's unclear if 0.4V means 0.4V at preamp input jack, or 0.4V at power amp input jack. I'm trusting that the boxed voltages all refer to identical conditions, which should mean 11mV input at guitar jack, 400mV output at preamp / input to power amp, and that is sufficient to drive the amp to full power output.

Clear as mud?

[Enzo]

Once again, we already established they draw these board by board. The preamp schematic says 0.011v input signal and 0.4v output. The power amp shows 0.4v at its input, which also says preamp out. SO clearly the input signal to the power amp is 0.4v, whether we stuff 400mv into the power amp directly or use the preamp. The output from the power amp board is marked as 30v at those cathode followers, as well as the approx. -45v legend. and on the power tube board we see the same 30v with -45v bias notation on the grids of the power tubes. And we see 372v of signal on the power tube plates.

All that seems quite consistent and clear to me. The schematic clearly shows 400mv at the power amp input. It seems very unlikely to me that they would also intend to have 400mv at the preamp input to result in the same voltage leaving the preamp.

[Steve Conner]

The cathode followers in this design have two purposes as far as I know. First, they stop the high capacitance of the power tube grids and wiring from forming a pole that might destabilise the feedback loop. Second, they help prevent thermal runaway by providing a low impedance sink for gas current coming out of the power tube grids.

I don't think it is intended to be a Class-AB2 design that runs the power tubes into grid conduction.