Push/Pull Pair Removal Theory & Questions

[trip]

So I found this majorly-conflicting-with-everything-i-know post



Can someone give me a basic run down of the "what happens" when you pull 2 tubes and half the impedance switch on a typical push/pull w/ 4 power tubes? Like.. the theory of what happens?

Voltage and current draw on the 2 tube plates left running go up, the overall impedance doubles, the overall resistance goes down (I think?). Things like that. My understanding is fuzzy but would love a quick professional and fast to-the-point of what exactly happens here and what if anything additional needs to be done. (Parts that get more stress/less stress?)

I was under the impression of puling two, half the impedance, and you're good. You might have to adjust bias but as bias is mostly a preference within acceptable range thing you can get away without having to adjust after pulling two given that there is no red-plating happening. Is this inaccurate? Not should you still adjust the bias (we know we "should") but do you have to? Does anything negative happen to the amp with the change, specifically, will it "actually" cause any problems by consistently running 100 watt heads w/ only 1 pair?

I was told at an expo event that this will always damage an amp because of the change in current and stress on some parts. I have also been told the opposite, that it's fine, you don't need to adjust bias (but you should) and that it will not fail because of running at half power. Then, I see a post as above, which also conflicts.

I'm hoping to get a solid understanding of what happens when you pull a pair in a push/pull and what changes if any are necessary after the fact (and to avoid misinformation).

Thanks guys!

[eschertron]

After a brief glimpse of the rant you linked - where the author claims 'that's bogus', '[they're] wrong', 'dead wrong', etc. - I'd say that you're better off taking the argument with a grain of salt.

There are circuit-dependent factors that can affect the usefulness of any thumbrule. Any determination aught to be done with a specific amp as an example, only then can any general case be extrapolated. Some amps can operate in the envelope with two tubes pulled, others may be over-dissipating. A cathode biased amp will change the bias operating point (not sure if there are any examples of a cathode-biased 4-tube amp?) and need compensation.

As a general rule (after advising you to be suspicious of general rules) pulling two tubes and moving the load to the next lowest impedance (thereby doubling what the reflected primary Z sees) should keep the amp inside the designer's operating envelope.

If your objective is to optimize the amp's behavior with 2 tubes versus 4, then all the technical suggestions come into play. Bias, check plate and screen dissipation, check output signal, etc. But you'd likely take some of those steps with any simple tube change.. if it's the tone you're chasing.

[glebert]

So you are asking strangers on the internet to comment on a comment by a stranger on the internet in response to another stranger on the internet?

I actually think it would be great to get some data on these questions, but any experiment you set up would just cover that one circuit design with that transformer, those tubes, etc. Whenever you use something differently than it was designed there is a chance that things break faster than they would normally. But there seems to be enough overdesign usually to let you get away with such things.

[Justin Thomas]

A cathode biased amp will change the bias operating point (not sure if there are any examples of a cathode-biased 4-tube amp?) and need compensation.

Anyone care to offer up their AC30 on the altar of amp science?

Justin

[glebert]

Anyone care to offer up their AC30 on the altar of amp science?

Justin

Crate VC30 would be less of a sacrilege, but I like them anyway and wouldn't do it.

[eschertron]

A cathode biased amp will change the bias operating point (not sure if there are any examples of a cathode-biased 4-tube amp?) and need compensation.

Anyone care to offer up their AC30 on the altar of amp science?

Crate VC30 would be less of a sacrilege, but I like them anyway and wouldn't do it.

So let the OP beware that I clearly don't know all the amps

[Justin Thomas]

Sorry, no offense was intended on my part.

Justin

[The Dude]

The language in that .png rant in the first post looks vaguely familiar. Just enough knowledge to appear to know what he's talking about mixed with an "I know everything" attitude. Hmmmmm, there used to be a guy around here that sounded like that. Can you tell us the source of that screen shot?

[glebert]

The language in that .png rant in the first post looks vaguely familiar. Just enough knowledge to appear to know what he's talking about mixed with an "I know everything" attitude. Hmmmmm, there used to be a guy around here that sounded like that. Can you tell us the source of that screen shot?

Did a google search for the text.
https://www.seymourduncan.com/forum/...age&styleid=28

[Justin Thomas]

But I think That Guy used the same name everywhere... And there's no mention of hating Crate-Peg or having invented sliver mica caps or using an ice axe to bias your Mesa...

Justin

[g1]

Trip, your understanding is mostly correct as far as I can tell.
The 'responder' makes an argument based on an error, that "the impedance of the primary winding does not change".
The impedance of the primary actually depends on the load on the secondary, as the ratio stays the same.
The power delivered is dependent on the primary impedance being matched to the power tube impedance.

[eschertron]

Sorry, no offense was intended on my part.

Justin

Oh! None taken!

I could claim that I was thinking about the AC30, but didn't know for sure...

[trip]

Trip, your understanding is mostly correct as far as I can tell.
The 'responder' makes an argument based on an error, that "the impedance of the primary winding does not change".
The impedance of the primary actually depends on the load on the secondary, as the ratio stays the same.
The power delivered is dependent on the primary impedance being matched to the power tube impedance.

Thanks! That was my understanding of what I remember, that's it's based on a ratio so removing two tubes would alter that ratio, so his response made me question everything I've been told haha I thought the info didn't seem correct, but I just wanted to lose any misinformation and gain a better understanding.



Appreciate the responses from every one guys, I know there is no one-answer-fits-all but just a general rule of thumb as to what happens internally when you pull the two.

& Yes, I guess I am asking strangers on the internet to comment on a comment by a stranger on the internet in response to another stranger on the internet lol

To be a little clearer, thinking along the lines of fixed bias 100 watt. Most of what I own are Peavey's, but I own way too many tube heads, most are the typical push pull 4 power tube variety. (My VK doesn't fall in line w/ the power tubes in series) If it costs me half as much to tube everything than it's worth always running half power. I'm not too concerned on tone as I don't hear too much of a difference in most (though the JSX sounds more saturated to my ears), it's mainly a money / wanting fresh tubes without spending 10x more than what I spent on the amps in the first place lol

For reference, this is more or less what i'm working with (as applicable) when I have these thoughts in mind. Ideally I kinda would want to run everything at half power except maybe the 5150 or 6505 (and BV, just because it makes me stuck to 16 ohm cab):

B52
AT-100 - Likely a bias issue happening but running 2 currently
AT-100 - Modded for el 34 - running all 4

Bugera
333XL Infinium - Running at 2

Crate
BV120 - Running all 4 - Set the bias but it still feels like this guy runs too hot, as the other one runs with barely any heat
BV120 - Running at 2, kinda a b*** because I can only run it into 16 ohms at half power, my good cabs are 8, never risked a real mismatch
Palamino V32 Custom - 4 el84s, never attempted to pull or even looked at the circuit but of course thinking about it

Line 6
Spider Valve HD100 Mki - Worked great w/ 4, lost a little punchy-ness going to 2 but have been running at 2 for most of its life, concerned as lots of forums talk about how crazy in spec SVs have to be or else they fry something

Peavey
Triumph 120 - All 4, would attempt two, love the tone and saturation on this amp, but with the layout of the tubes it's questionable which to pull (I don't remember the schem)
JSX - 2, loving how the tonal color changes, fave amp, would love to always run at 2 tubes, the idiot before sent it to me red-plating, this is the only amp I haven't ever actually measured and set bias for, I just set it arbitrarily to not be red-plating like when I bought it and have left it as is (for $250 I wasn't THAT concerned).
3120 - All 4, 6l6
5150 Signature - Something in the bias chain is failing but when I fix that I'd like to switch to 2, have always ran 4
6505 Import - Have to run this guy through a 110 to 220 converter, all 4 tubes, not sure of any potential issue with it being imported spec
Valveking VK100 (Can't, Tubes in Series, would love to, see lots of mods for a half power switch similiar to mesa, kinda wanting to implement that because I really don't have a need to run 100 watts)

& that's my cluster of stuff. I care about all of these and want to keep them operating forever long as I own them (and not blowing up components) so understanding the why and logic behind what happens would be pretty helpful. And honestly I'd prefer to run everything at half if I can so I can always have fresh tubes and not spend 2x more on retubing everything I own. (And also before I find the issue with the b52 and 5150 bias I would rather not fry 2 tubes in troubleshooting than not fry 4 lol)

[bob p]

I thought the guys name should have been "Brown Spot" instead of "Brown Note." But I'll try not go go down the path of bullying somebody who isn't even here to defend himself. To that end I'm not going to make snarky comments from the sidelines without at least making a useful contribution to the thread:

Getting back to the OP's question:

Things are always more complicated than we like to admit, and people like to work with rules of thumb because that's easier than deriving the correct answer to these types of questions. That's not all bad, because the rules of thumb can be very useful. As useful as those guidelines may be, it's still worth understanding the potential pitfalls of following those rules so that you won't get into trouble.

Can someone give me a basic run down of the "what happens" when you pull 2 tubes and half the impedance switch on a typical push/pull w/ 4 power tubes? Like.. the theory of what happens?

Voltage and current draw on the 2 tube plates left running go up, the overall impedance doubles, the overall resistance goes down (I think?).

Suppose that a pair of 6L6 are going to be run in AB1 at an anode voltage of 450 VDC. The GE data sheet suggests Raa of 5k6 and a bias voltage of -37, and a peak load of 232 mA for the pair.

Going to a quartet at the same B+, the recommended Raa will be halved to about 2800 (5k6 in parallel with 5k6) and the peak load will increase to 2 x 232 mA = 464 mA.

Now suppose that you had an quartet type amp, like a Twin Reverb. It might have an OT with a turns ratio of about 700:1, so that when you place a 4R load on the amp (two 8R speakers in parallel), it reflects a primary Z of 2800 to the quartet of 6L6: 4R x 700 = 2800R. That puts us right in the area of the design center values for running the quartet.

If you pull two tubes out of the amp, that doubles the target Raa for the tubes design center values, taking you back to an optimum value of 5k6 for a pair. A problem pops up though, that you've still got a 700:1 turns ratio in your OT. Your optimal load would be 5600R/700 = 8R; in order to reflect a "proper" load impedance through the transformer to the pair of 6L6, you'll need to connect an 8R load. If you have two 8R speakers wired in parallel, disconnecting one of them would solve that problem.

As a matter of practice, most people don't bother to do this, and instead they just rely upon the fact that the amp's engineers designed the amp to match design center values, which provide a fair amount of leeway for mismatch errors. In that case you've got some safety margin to work with, so most people won't worry about changing the speaker load, even though doing so would be optimal.

Unfortunately, there are a lot of people who develop a false sense of security about the safety of proceeding to pull tubes without worrying about Z-matching, because they've gotten away with sloppy practices for a long time. Because amps like Fenders were designed around optimal design center values, and are built with robust components, many people have developed a cavalier attitude about mismatching. They haven't been burned (yet) so they're inclined to recommend what I would not consider good engineering practices.

Those rules of thumb work great for amps that were designed around centered values and were built with robust components. They don't work so well though, if you've got an amp that was deliberately designed away from the tubes' design center recommendations. An example of this could be a boutique amp, which could be deliberately designed to alter the amp's distortion characteristics by deliberately moving it to an off-center load resistance. In cases like that the mismatch could push things a little too far, and being cavalier about mismatching could result in getting burned. In that scenario Z-matching when pulling tubes would be a good idea.

That covers the Z-matching question. The voltage/current/bias problem still needs to be addressed. More coming...

[glebert]

Palamino V32 Custom - 4 el84s, never attempted to pull or even looked at the circuit but of course thinking about it

FYI, this is Class A.