Difference between 2 x EL84 and 4 x EL84 board architecture?

[bob p]

I've seen amps with no screen resistors, screen R's on some but not all the power tubes, and screen R's on all power tubes.
I thought it was basically an evolution but I could be wrong.

I've seen amps where each side of a dual pair amp shared common screen resistors, I've seen amps where each tube had a dedicated (symmetric) screen resistor, and I've seen amps without any screen resistors... but I've never noticed amps that had asymmetrical screen resistors in the parallel sides of the PP pair.

In the case of the PV Classic 30 I can't tell if they did that on purpose or by accident. I can imagine the case where someone said, "OK let's scale this push-pull pair amp up to a dual pair configuration." Then somebody either drew the schematic wrong, or did the assembly wrong, and an amp that had screen resistors on one pair but not on the other made it to production. I'm guessing it could be a chicken-egg thing. If that topology was derived from a misteak, was it a mistake in drawing the schematic that caused the assembly guys to get it wrong, or was it an error in assembly in the prototype lab, and someone just drew up the schematic to match the prototype? Serendipity has a lot to do with discoveries, and I'm not sure whether the Peavey design was intentional or accidental. What we do know is that they decided to keep the design.

We also know that Randall Smith filed for a patent for his fixed-bias "Simul-Class" design, where one PP pair would be run in "Class A" [sic] while the other pair ran in Class AB. This was allegedly done for tonal reasons, to allow one pair to begin breaking up while the other pair was still clean, in order to give a fatter sound that was both distorted and clean/large at the same time.

During the duration of his patent he enjoyed exclusivity on this design. If people wanted to thicken up the sound by running two pair of tubes with different fixed-bias parameters, then they had to license his Simul-Class technology or develop a work around.

The Peavey design shows that there is more than one way to skin a cat. If you want two pair of triodes to distort at different levels, you can change where grid current limiting occurs and you can change where cut-off occurs. If you're working with a tetrode or a pentode then you also have the option to change where screen current limiting occurs. I'm thinking that PV did the screen resistor thing as a cheap way to circumvent one of Smith's patents by implementing differential clipping through differential screen-current limiting. You'd have to check the dates on the PV amp and the Smith patent to be sure.

[potatofarmer]

I've seen the "only 2 of a quad get screen stoppers" on both the Peavey Heritage VTX and Deuce 2, so it's something they've been doing for a good 40 years. I've always figured they only added those two because they had problems with oscillation or some such caused by the leads to the added tubes that didn't show up in the "one pair" models. That's really just a guess though.

Really want to add the missing screen stoppers to my Heritage VTX but the sockets are mounted on a board that makes that difficult at best. One of the drive transistors went on the previous owner and the PCB above the tube socket actually warped from the heat of a melting-down tube, so I'll probably just chassis-mount 4 new sockets someday, but as-is the missing screen stoppers aren't causing any oscillation problems... not that I've busted out the scope looking for them.

Just had a thought and looked up the Mace VT schematic - there they have a screen stopper on every tube (albeit an inadequate one). Actually it's a combined schematic for both the Mace VT and the Deuce VT - http://www.thetubestore.com/lib/thet...-Schematic.pdf

And both the 4- and 6-tube models apparently have a stopper for each screen. But they didn't do that before or after so maybe it's just a production cost decision more than anything deliberate.

[bob p]

... I won't argue that the grid load should be adjusted for the driven load. I completely agree. I've even said as much already. I was only noting that there are some coveted designs that failed to recognize this design element and the resulting manifestation of "improper" design has been well received tonally by players. Not sure how much more discussion that deserves, but it is something that could go unnoticed (or at least unmentioned otherwise).

I alluded to this in the EL34 reliability thread. I think it's worth discussing.

One function provided by the grid leak resistors is to provide a leakage path away from the grid to the cathode when the grid gets heated, emits electrons and becomes positively charged. As the grid gets hotter it emits more electrons and anode current increases. This can create a vicious cycle in which heat causes more anode current, which causes more heat. When the bias drifts out of control we end up with thermal runaway and failure.

Essentially, a properly sized grid leak resistor will allow sufficient leakage path to the cathode that it pulls the voltage of the grid back to a stable value.***

When you add an extra pair of output tubes the effective parallel impedance is cut in half, so you can cut the value of the grid leak resistors in half without loading down the previous stage. Decreasing the grid leak resistance has the beneficial effect of allowing more current to leak to the cathode, neutralizing the charge on the grid and helping to prevent thermal runaway.

Based on the chart above I see that Fender didn't bother to make these adjustments in the Tweed era, but they were smart enough to make these adjustments in the BF/SF era. Marshall on the other hand, never made these adjustments and we see Marshalls having the nasty habit of going into runaway and sacrificing tubes. If I had one of those tube-eaters, I'd be adjusting the grid leak resistors by cutting their value in half. Who knows, it might fix the EL34 reliability problem. Worth a shot anyway.

*** It would be interesting to know if the grids are being carried positive in the Marshalls that fail.

[nickb]

When you add an extra pair of output tubes the effective parallel impedance is cut in half, so you can cut the value of the grid leak resistors in half without loading down the previous stage.

I have to be misunderstanding you. If you have twice the number of tubes and half the grid leak resistance the impedance seen by the driver is halved i.e the loading on the driver is increased.

[tubeswell]

I alluded to this in the EL34 reliability thread. I think it's worth discussing.

One function provided by the grid leak resistors is to provide a leakage path away from the grid to the cathode when the grid gets heated, emits electrons and becomes positively charged. As the grid gets hotter it emits more electrons and anode current increases. This can create a vicious cycle in which heat causes more anode current, which causes more heat. When the bias drifts out of control we end up with thermal runaway and failure.

Essentially, a properly sized grid leak resistor will allow sufficient leakage path to the cathode that it pulls the voltage of the grid back to a stable value.***

When you add an extra pair of output tubes the effective parallel impedance is cut in half, so you can cut the value of the grid leak resistors in half without loading down the previous stage. Decreasing the grid leak resistance has the beneficial effect of allowing more current to leak to the cathode, neutralizing the charge on the grid and helping to prevent thermal runaway.

Based on the chart above I see that Fender didn't bother to make these adjustments in the Tweed era, but they were smart enough to make these adjustments in the BF/SF era. Marshall on the other hand, never made these adjustments and we see Marshalls having the nasty habit of going into runaway and sacrificing tubes. If I had one of those tube-eaters, I'd be adjusting the grid leak resistors by cutting their value in half. Who knows, it might fix the EL34 reliability problem. Worth a shot anyway.

*** It would be interesting to know if the grids are being carried positive in the Marshalls that fail.

We were also discussing this in this thread

http://music-electronics-forum.com/t44631/

There is an older version of a Marshall 1959 schematic floating around which seems to show 120k for the grid leak for each EL34 pair (although its difficult to read). This makes more sense than 220k/pair, but does mean the impedance bridging from the PI is more lossy. I've heard some guys go down to 150k.

With the EL34 design being such that g1 is in very close proximity to the cathode, some of the emissive material that is applied to the cathode (in order to help emit electrons) would tend to migrate to g1 after a while. Steve Conner has posted about in this thread from 2013 http://music-electronics-forum.com/t32043/. This could make older EL34s harder to bias properly. However, even new EL34s don't seem to last long in these amps unless there is some kind of modification.

I've had to change the 27k AC bias supply resistor to 15k to get the bias to stabilise.

Also in the 1959 amp only 2 of the EL34s have a (signal) grid stopper. Go figure.

[bob p]

I have to be misunderstanding you. If you have twice the number of tubes and half the grid leak resistance the impedance seen by the driver is halved i.e the loading on the driver is increased.

I probably should have worded that better, with something like "without unduly loading down the previous stage."

I think you are understanding me. Sure, lowering the impedance seen by the driver will increase the load on it. But the question is whether the change in load on the preceding stage is going to be bad enough to cause a problem. In most cases the typical LTPI driving the output stage doesn't have a problem driving the lower impedance load of 2 pair of tubes instead of one. Fender Twins are a good example of this. Driving one or two pair of tubes isn't hard, but when the impedance falls a lot, like in the case of driving three pair of tubes, then you have to deal with that by changing the drive circuit. An example of how to address the loading when it gets bad would be Fender's Super Twin, using CF drive between the Pi and the 6x6L6.

As we know, dealing with the parallel output tube scenario is a lot like playing whack-a-mole... you fix the problem in one place and it moves somewhere else.

[bob p]

With the EL34 design being such that g1 is in very close proximity to the cathode, some of the emissive material that is applied to the cathode (in order to help emit electrons) would tend to migrate to g1 after a while. Steve Conner has posted about in this thread from 2013 http://music-electronics-forum.com/t32043/.

I had missed those threads, but it sounds like Steve and I were on the same page on the subject of maximum grid leak resistor values and thermal runaway. Although he described it the mechanism as grid contamination, I think that cathode proximity heat by itself is sufficient to cause the grid to get hot enough to give up electrons.

Also in the 1959 amp only 2 of the EL34s have a (signal) grid stopper. Go figure.

I saw that. It can't help the situation. If it were my amp I'd "fix" that. (More heresy.)

[frus]

Fender Twins are a good example of this

But Fenders also used a 12AT7 as a PI, which has a lot lower internal plate impedance, so it can more successfully drive harder loads (lower power tube grid resistors) than 12AX7

[bob p]

Arguably, a superior design. Is this where we fork the thread to start listing everything that Marshall did "wrong"?

[Dave H]

As we know, dealing with the parallel output tube scenario is a lot like playing whack-a-mole... you fix the problem in one place and it moves somewhere else.

I feel like playing it for real. There are two 12 foot long 'snakes' in my lawn. I keep treading them down but the next day they are back. He has dug them up again! Where's the spade?

[vintagekiki]

To determine the Difference between 2 x EL84 and 4 x EL84 some standards must be observed when testing, because only this can be the difference in the sound.

It is assumed that in both cases the concept of schematics (or board architecture) that is compared is identical, with only the difference with the stronger PT and OT transformers (4 x EL84)

Recommended standards for testing are
- The same guitar
- The same tone combination and guitar and amplification
- The same loudspeakers
- The same test power

If some electronic measurements are made for the purpose of the comparison, considering that in both cases board architecture preamp is an identical, the measurements are performed on the power amp.

The 2 x EL84 amplifier provides a lower power of 4 x EL84, 4 x EL84 tested in the power resonance provided by the 2 x EL84 amplifier.

Translation by https://translate.google.com

[Mike Sulzer]

- The same test power

If the reason for using 4 EL84s is to get more power, maybe testing it with twice the power in a larger room would be better. Or perhaps it could be tested with two speakers, one located outside of the listening room and sufficiently isolated so that only the speaker in the room is heard..

[vintagekiki]

Looks like we do not understand.
Or the discussion takes place in the wrong direction, or the title should change in Difference between 2 x EL84 board architecture, because board architecture can be different from model to model, from manufacturer to manufacturer.

It is logical that the amplifier with 4 x EL84 gives more power than an amplifier with 2 x EL84. In order to get some useful results between amplifier with 2 x EL84 and amplifier with 4 x EL84 it is necessary that in both cases the board architecture be identical.

The same is true for the test power, because the same speaker gives a different subjective picture at different power (2 x EL84 or 4 x EL84)

[pdf64]

Arguably, a superior design. Is this where we fork the thread to start listing everything that Marshall did "wrong"?

Bear in mind that Marshall settled on the EL34, which has a 500k g1-k resistance limit, so 220k resistors should be reasonable, even with 2 tubes paralleled.
Whereas 6L6GC g1-k limit (in fixed bias) is 100k, so with 2 paralleled tubes we're down to 50k; when we bring in the bias supply output resistive impedance, that's pretty low! Fender didn't make any move to compliance with that until the late 70s.

[Enzo]

Re: the Peavey two of four tubes have screen resistors. They have done this for decades, it isn't a mistake or accident. They do it for stability. The tubes basically have no screen resistors, but they put the 100 ohm resistor between screens on each side to prevent potential interaction of the tubes. The design does not assume matched tubes.