28 watts from a pair of el84's. Is this even possible?

[Gainzilla]

Hi guys,

Low priority post here, but I'm curious. Victory amps recently released a couple of lunchbox claiming to produce 28 watts from 2 el84's. Martin Kidd is a well respected designer with a long track record of shooting straight. Giving him the benefit of the doubt, how is this possible (if at all)? The amp is a non-master volume, btw.

http://www.victoryamps.com/rd1-rob-chapman.html

One guess: it's all in how you measure it. If you typically measure the output in clean watts, maybe he's getting his measurement by cranking it wide open?

Another guess: I would think you could increase the plate voltage for more headroom, but from my limited experience, el84s are not nice sounding when punished with excess voltage. That said, there are examples of some iconic amps that run at the high end of the range. So maybe he's designed the circuit to combat any brittle or harshness. He's certainly a much more knowledgeable guy than I. 😆

Data sheets are like Shakira's hips. They don't lie. Ive never seen one showing more than 12w max plate dissipation. Amps I've seen on the market top out at 20w for a pair in p/p configuration. If going by the datasheet, with max voltage, he could get 24watts clean (har har), going downhill in a hurricane. So what's the deal here? Is he measuring outside of accepted practices (with the amp running full tilt)?

He's making some damn nice amps, and this one looks really cool! The skeptic in me really wants to understand the whole 28w claim, though.

Cheers!

- Bryce

[Stevem]

Not RMS power in class AB1, but in AB2 yes!

[jazbo8]

No need for AB2, 28W is no sweat in AB1, Modjeski even managed to get 35W from a pair long ago - the ingrediants are high plate voltage and high Ra-a.

[Gainzilla]

Thanks for the reply! Huh. Interesting. I'm Afraid I don't know much about that (no surprise there. Lol).

From Duncan's amp pages:
... AB "2" indicating that the valve output stage grid voltage is being pushed above the 0 volt mark and into a positive grid voltage, causing the grid to draw some current from the preceding driver stage.

So how does that work exactly? The grids are getting more voltage than the plates? Any examples you know of that I can track down?

Cheers!

[David Root]

No, the grids aren't at higher voltage than the plates. The easiest way to do AB2 is to use a driver tube as a cathode follower to feed the power tube grids direct connected, ie no blocking cap, after the PI. The driver cathode follower allows the power tube to draw much more current than in AB1, so more power.

AB2 requires fixed bias and typically a raw bias voltage > minus 130V. Any Dumble Steel String Singer with 6550s does this. I built one and it definitely works. It requires biasing both the driver grid and cathode. See schematic below.

[Gainzilla]

No need for AB2, 28W is no sweat in AB1, Modjeski even managed to get 35W from a pair long ago - the ingrediants are high plate voltage and high Ra-a.

Aaaaaah, ok. I had to do a little googling but does that mean the higher summed plate-to-plate impedance requires a different transformer ratio, which provides more signal to the speaker?

Apologies for the newb questions. This is actually something I've really been wanting to learn more about. I've learned so much over the last couple years, but this is one area I've not been able to find much info on. Or maybe because the it's so theory heavy, I have struggled with it in the past. I've recently been diagnosed with adult ADHD, and now that I'm on some meds I feel much more focused and able to consume dense technical information. Prior to that, I would read like 3 lines and then follow a squirrel somewhere else. lol

Anyway, thanks for helping me understand this.

[Gainzilla]

No, the grids aren't at higher voltage than the plates. The easiest way to do AB2 is to use a driver tube as a cathode follower to feed the power tube grids direct connected, ie no blocking cap, after the PI. The driver cathode follower allows the power tube to draw much more current than in AB1, so more power.

AB2 requires fixed bias and typically a raw bias voltage > minus 130V. Any Dumble Steel String Singer with 6550s does this. I built one and it definitely works. It requires biasing both the driver grid and cathode. See schematic below.

Thanks for the info. Interesting. Going out on a limb here, but does running the driver as a cathode follower keep the impedance higher, allowing the power tube input grids to get more signal?

Clearly I need to go dig up some reading material on this. Thanks for the schem, and the info!

Cheers,

- Bryce

[lord preset]

This is not really so different than what Music Man did with their amps. They used a voltage doubling PS and ran the plates around 700vdc and the screens at 350vdc to get ~100 watts out of 2 - 6l6GC. It would appear that Modjeski did something similar with EL84s http://www.ramlabs-musicreference.com/rm10design.html

The trick is getting an output transformer suitable for this application without going custom. You can get OTs designed for MusicMan and Hiwatt (DR200) but Modjeski makes his own. I am not even sure how one would spec one for 2-el84s at ~30 watts.

[David Root]

Running the driver as a cathode follower makes its output impedance to the power tube grid very low, only a few hundred ohms typically, and the grid is a high impedance input.
Not sure how much if any difference this would change signal voltage transfer but it definitely enables greater current draw by the power tube grid.

[LeeMo]

wasn't the Traynor Bassmate from the early seventies about 25 watts?

[pdf64]

Running the driver as a cathode follower makes its output impedance to the power tube grid very low, only a few hundred ohms typically, and the grid is a high impedance input...

The power tube control grid is high impedance until Vg>Vk, above which it conducts, behaves as a forward biased diode and drops to a low impedance, eg ~2k.

Hence the phenomena of bias shift etc, see text https://www.ampbooks.com/mobile/amplifi ... excursion/

The amp's spec is 28 watts for a sine wave at the onset of clipping, see of http://www.victoryamps.com/uploads/2/3/ ... ssue_1.pdf
See photo of amp gubbins p7
It can't be operating in the B2 grid current region, as there's no dc coupled buffer (eg transformer, tube or MOSFET) between the LTP phase splitter and power tube control grids.
Bear in mind that linear operation in the grid current region can't be achieved using capacitive coupling to the power tube control grids, as bias shift will ensue.
However, the HT caps, despite being 450V types, look to be in a series totem pole arrangement, indicating that the HT is likely >400V.

The 28 watt power output is feasible in regular AB1, provided the normal limiting plate and screen grid voltages are ignored.

Have a play with http://bmamps.com/Tech_tds.html to see for yourself, eg plates @ 400V, plate load of 8k, screen grid @ 380V, -20V grid bias
gives just over 32 watts output with max plate dissipation slightly over 12 watts.
That's allows for >10% losses, eg OT, in a real implementation.

[jazbo8]

Slightly OT, but it has often been said that high plate voltage makes the amp sound hard, I have not build anything over 450V, so have no clue if that is true or not. What's your experience?

[Leo_Gnardo]

However, the HT caps, despite being 450V types, look to be in a series totem pole arrangement, indicating that the HT is likely >400V.

The 28 watt power output is feasible in regular AB1, provided the normal limiting plate and screen grid voltages are ignored.

Two things concern me with such a design, one is the tendency for arcs to develop between tube pins closely spaced, as on EL84. I know the spec sheet says "up to 800V" and they have been used in TVs that way, but still, why invite arcs.

And yes "provided the normal limiting plate and screen grid voltages are ignored." So be prepared to swap in fresh tubes often. No MIC "firecracker" EL84's either... and don't waste your collector-item Telefunken, Valvo, Mullard EL84's on this amp either.

[pdf64]

it has often been said that high plate voltage makes the amp sound hard, I have not build anything over 450V, so have no clue if that is true or not. What's your experience?

Yes, my perception is that for most any CC stage, pre or power, the higher the idle Vp-k, the harder the tone. But for power tubes, maybe it's the Vg2-k that matters?

Something I'm uncertain about regarding limiting voltages on tube info is that 2 levels are often specified, eg p8 of http://www.mif.pg.gda.pl/homepages/fran ... e/EL84.pdf
Vao = 550V
V1 = 300V
P3 of http://www.mif.pg.gda.pl/homepages/fran ... e/EL84.pdf
calls it Va(b) etc.

I assume that the higher voltage is acceptable if it's under idle conditions and there's minimal current flow?

[jazbo8]

I assume that the higher voltage is acceptable if it's under idle conditions and there's minimal current flow?

Yes, there have been many tests (and actual products) that use very high voltages - well above the datasheet limits without issue, even over long period of time, e.g., the Musical Reference uses around 750V for the plates of the EL84. The key is to keep the idle current low with proper bias and use higher-than-normal load to keep the peak current in check as well. But just because it works, does not mean it sounds good...