How to Bias SF Twin Reverb (135W)?

[matt h]

(deleted)

[matt h]

(deleted)

[pdf64]

Apologies for not being clear, my experience is that when pushed, these amps draw heaps of screen grid current and screen grid resistors are very necessary, especially given the higher VB+ (dissipation being a product of current and voltage).
I don't perceive that 470 ohms is sufficient, rather at least 1K, and 2k2 if it's intended to be cranked.
Sorry to labor the point, but my hunch is that it may be due to the OT being, well, a basically pentode DL type, rather than a regular UL, which is moved more towards triode conditions.
Though it may be that UL wouldn't be much better, as I understand that normal practice with pentodes wired for triode operation is to still include screen grid resistors. And even ~40% UL taps may still be way closer to pentode than triode operation.

[matt h]

Apologies for not being clear, my experience is that when pushed, these amps draw heaps of screen grid current and screen grid resistors are very necessary, especially given the higher VB+ (dissipation being a product of current and voltage).
I don't perceive that 470 ohms is sufficient, rather at least 1K, and 2k2 if it's intended to be cranked.

What is your experience with the UL fenders? I'm not trying to be a dick about this, but do you have one? Can you take measurements? Do you also have a non-UL twin running similar voltages? It would be a terrific learning moment to see these differences documented rather than cobbling together what tenously grasped theory and limited and fractured practice has shown. I'm curious about this because, as rightly pointed out by potatofarmer, that since the screen voltage tracks with the plate voltage, even though screen voltage may be at some portion above the plate, that while the plate voltage drops down to 50V or whatever on the low part of the cycle, that you don't have the screen remaining fixed around its idle voltage (and since out of phase, increasing during that period. So a 50V plate would not exist at the same time as a 550V screen).

Since screen voltage informs screen current draw, it follows that if the screen's upper limit of voltage is lower in UL/DL/BaberahamLincoln mode than it is in "pentode" mode (screen tracks with plate, screen is out of phase from plate)-- it must then follow the screen would draw less current than in pentode mode. The peaks are pulled down, the troughs are pulled up. This has the effect of limiting dissipation through proportional/tracked manipulation of both voltage and current.

[pdf64]

There are a few 135 TRs in my area, and I seem to work on one or the other at least once a year.
Regular TRs are much less common, I only know one guy with one and he was looking to sell it.
I've got a 75 http://schems.com/manu/fender/fender_75_schem.pdf and service a friend's, but I modded both to regular pentode with a choke / extra screen grid B+ node.
Reason being that although they seem to work fine with the old Sylvania STR387, modern tubes tend to become so weak after a few hours of being pushed as to be useless, even with 1k screen grid resistors.
I suspect that's mainly due to the high stiff B+ on the screen grids, rather than the tapped primary, as a pair of EH 6L6 didn't last long after the mod to pentode operation. I've since added a 1k dropper in series with the choke to get some sag on the screen grids, which so far seems to be successful in taking the damaging stress off the tube, see http://s963.photobucket.com/user/pdf64/ ... sort=3&o=0

My 75 is across town in the band room currently, but I can get it back in a few days; what tests were you thinking of?

Since screen voltage informs screen current draw, it follows that if the screen's upper limit of voltage is lower in UL/DL/BaberahamLincoln mode than it is in "pentode" mode (screen tracks with plate, screen is out of phase from plate)-- it must then follow the screen would draw less current than in pentode mode

I'm confused by the 'screen out of phase with the plate' part of that?

My understanding is that in pentode mode or with screen grid taps, the plate and screen grid signals will be in the same polarity.

Regarding the significant of the primary taps, I think that the voltage ratio between screen grid and plate will be related to the impedance ratio, rather than winding ratio.
So for 43% of winding taps, the screen grid will be at 18.5% of the primary CT-P impedance.
So with a loaded VB+ primary CT at ~500, with the plate at ~50V the screen grid will be at 417V

With 12.5% of winding taps, the screen grid will be at 1.6% of the CT-P impedance, so CT VB+ ~500V, plate ~50V then screen grid ~493V.

Hence I perceive that this range of Fenders are operating basically in pentode mode, the small degree of linearising negative feedback from the screen grid primary taps probably being insignificant.
The screen grid dissipation, with them being basically at VB+, being little different to a regular pentode connection, but with the additional stress (compared to a regular amp) of the B+ node being super stiff, due to no sag across the choke.

[potatofarmer]

But it seems it's still based on an assumption that UL must be the most-optimal % of DL, and that because fender is clearly using DL, but maybe not most optimally (for least distortion), it's therefore not UL. I don't buy that. The same articles linked in that other thread use UL and DL interchangeably--the history of UL arose from someone else giving a different name to the same principle, and the marketing teams running with it. UL and DL are interchangeable, UL being a cutesy description of a more optimal form.

If you read Hafler and Keroes' original publication (and/or patent) on this, yes absolutely they stress the importance of how amazing this one particular tapping point is. That oestex link has excerpts of the patent's text. F Langford Smith also went into detail analyzing the "optimal" tapping point.

See the attached.

This is nothing like Class A vs Class AB, and Vox Class-A marketing bullshit. There were no audio wars wherein one designed trolled another designer for "Using DL but non-optimally configuring the screen the tap for absolute least distortion and calling it UL, lulz".

Wait, that happened? I figured it was just marketing BS that had built up over the years, like "Peak Watts" versus "Watts RMS." Morgan Jones had a great quip about that: "There is no such thing as an RMS watt. Please review the definition of RMS again."

Anecdotal (and I'm not arguing it's untrue, just that I haven't seen that number published before and the only sources I'm finding on the web are your own posts and your blog) measuring of Fender UL OT shows a 12.5% tap.

There was someone else at MEF who measured their Super Twin at around the same percentage, so at least it's not just me.

If UL operation actually worked with a different mechanism, I'd bandwagoneer onto the "Duh, not UL, lulz" troll-train, too. But it doesn't.

"Pedantry and pederasty--only one of them is illegal, but both are a pain in the ass."

Calling me a troll and making a molestation joke. Lovely.

Fine; you don't like me. I'm not sure how this is relevant, but thanks for the heads-up.

Increasing them is great; I put 2k/5W flameproofs on mine

Definitely; however, this action seems to be in contradiction to the previous statement? eg why (if they aren't necessary) is it great to increase the value, what's the benefit?

I think I installed them before I'd done any measurements of the OT after I saw someone (you?) measure 20+W of screen dissipation. Largely I prefer the tone with the larger screen stoppers, the slight drop in headroom, and I'm not entirely willing to bet the farm on both my math being precise enough to guarantee the screens are safe... or that current production tubes won't suffer from reduced life with screens always hovering at 4.5W.

I've also seen reference to larger-than-typical screen stoppers recommended in UL/DL connection to suppress oscillation, though I can't remember a source.

In addition, I haven't done any measurements at overdrive.

With 12.5% of winding taps, the screen grid will be at 1.6% of the CT-P impedance, so CT VB+ ~500V, plate ~50V then screen grid ~493V.

Hence I perceive that this range of Fenders are operating basically in pentode mode, the small degree of linearising negative feedback from the screen grid primary taps probably being insignificant.
The screen grid dissipation, with them being basically at VB+, being little different to a regular pentode connection, but with the additional stress (compared to a regular amp) of the B+ node being super stiff, due to no sag across the choke.

No, the distinguished gentleman from New England is correct. Since the screen voltage tracks the plate voltage you don't have the Vg2 >>> Va condition that you do with pentode connection. The effect isn't huge in these Fender OTs, but it's still enough that the old chestnut "you don't need screen stoppers in DL/UL" is true, at least for dissipation purposes.

Not exactly sure where this guy got it, but there's a forumla for figuring out the corresponding Vg2 for a given Va here: http://ayumi.cava.jp/audio/ULKF/node1.h ... 0000000000

The formula is:
Eg2 = Ep0 + Bsg(Ep - Ep0)

where Bsg (actually Beta sg) is the percentage of the turns at which the screen taps are taken.

With the screen taps at 12.5% of the winding, the screen drops ~64V from idle, which isn't phenomenal but it's preferable to the 7V you estimated. With the proper tapping point (whose name I shall never utter again under penalty of mockery) the screen voltage would drop roughly 200V when the plate bottomed out around 50V, which would be fantastic. Fender put these exactly on the edge, and apparently they're fairly reliable because there hasn't been a massive rash of shorting tubes. I'm just glad the PT was wound for a 120V primary at least... and we'll see what happens when the wall voltage creeps up even further than it is now.

[pdf64]

I think I installed them before I'd done any measurements of the OT after I saw someone (you?) measure 20+W of screen dissipation

Yes, that may have been me. Merlin advised that the current measurement method I'd used (Fluke's (Vdc+Vacrms) / 470) may have over estimated, and that a simple Vdc would have been better.
It's still over 5W.

Eg2 = Ep0 + Bsg(Ep - Ep0)

where Bsg (actually Beta sg) is the percentage of the turns at which the screen taps are taken.

Ah, so my assumption about the voltage across the primary being related to the impedance was incorrect?
Thinking about it, that seems reasonable, as a transformer's voltage ratio is the same as its turns ratio.
Thanks for the info!

[martin manning]

Here are traces of pentode (Vg2 = Constant) and 12.5% DL operation for a 6L6GC. I can't make a full trace much above 250V (for now), but this is representative. I happened to have a 6L6GC all set up, so it was easy to run these. You can see that in DL the screen current is throttled down about 17% relative to pentode operation at the minimum anode voltage.

[matt h]

(deleted)

[pdf64]

the AC signal at the screen is always 180(degree symbol) out of phase with the plate

Darn it, I'll have to go and check!
Are you sure that you aren't getting the control grid mixed up with the plate?
I think that weird self splitter push pull amp used the inverse polarity signals at the control grid and screen grid to achieve its effect.

[martin manning]

When the plate voltage is pulled down the screen voltage comes down with it, so they are in phase; for half of the cycle anyway. In cutoff, the screen voltage is flat (in pentode operation) while the plate rises to something approaching 2x B+.

[matt h]

(deleted)

[pdf64]

See p7 lower chart http://www.mif.pg.gda.pl/homepages/fran ... /6L6GC.pdf
As g1 voltage rises towards 0, plate and g2 current increase.

[matt h]

I said 'tube current' is constant in a pentode. (Not plate current). The effect of a screen is that when plate current goes down screen current goes up by the same amount, so the overall tube current remains constant. And it is precisely because screen current does that, that screens really should have some screen grid resistance to dissipate the current.

I'm terrible at formatting-- Tubeswell, who posts here, posted that on another forum-- I went googling for a quote I believe I saw Merlin state elsewhere, which says essentially the same thing but can't locate it at this moment.

Here's a link to where I ganked that quote, in the interests of ass-coverage http://www.tdpri.com/forum/shock-brothe ... ation.html

[martin manning]

And you believe everything you read on the internet? ;^)

At constant Vg1, as Va goes down that is sort-of the case, i.e. plate current goes down and screen current goes up as Va is reduced, but the total current is not a constant. You can verify that by looking at data sheet curves. In practice, when the tube is connected to a load Va and Vg1 change together, and plate current and screen current are in phase.