Power Tube load sharing

[nevetslab]

Question: In the case of a Fender Twin Reverb, who's Sovtek 5881WXT Power Tubes measure 27mA/11.9W for V7 & V8, and 20mA/8.8W for V9, and 23mA/10.2W for V10, would you leave this as is, or swap them, moving the two 27mA tubes to both sides, to mix with the lower current tubes? I have one on the bench, and after swapped V7 & V10, I then bumped the bias up for reading 28mA/12.8W for V7, 32mA/14W for V8, 25mA/11W for V9 and 34mA/15W for V10. While this averages out to around 60mA per pair, it also places the higher current tube doing most of the work, and the load sharing isn't as good. With the original balance, the two tubes running 27mA vs the other two running 23ma & 20mA, the load current was better, though the balance of the two halves wasn't equal. What makes more sense? Besides replacing the set with a new matched set. The date codes on these tubes would warrant that, being 0601 and 0801 (2006, 1st 2k, 2008, 1st wk).

[Enzo]

What is your goal? Hum balance at idle? This isn't a hifi, so absolute balance for signal might not yield the "best tone". And I wonder how much of that most of th work occurs at audio, rather than DC idle.

[Mick Bailey]

I also wonder how a static match shifts under dynamic conditions. The only way to check is to run the amp into a dummy load and measure the respective dissipations at (say) 25, 50, 75 and 100% output with a fixed frequency input signal. That would be interesting because the shifting screen voltages would have more influence than at idle.

[Helmholtz]

See if any of the tubes redplates between half and full power.
Otherwise I'd go for equal sum currents on both sides of the primary.

[nevetslab]

This morning, I checked my notes to see which tube I found having the silk screen discolored. It was one of the two tubes having the higher plate current. I left off yesterday with mixing the lower current tubes with the higher current tubes. I plugged in my 4 ohm 200W dummy load, with AC Voltmeter across it, as well as the distortion analyzer/scope, and turned up the output as Fender states for full power measurements: 1khz 5% THD into 4 ohm resistive load, Normal Ch, tone controls fully CW, with Vib ch controls all CCW.

With this idle plate current balance being:

V7 28mA
V8 32mA
V9 25mA
V10 34mA

I turned up the output to where I found 5% THD. 5.9VAC, 8.7W! Forget the 5% THD figure. I reset the output to where I see Visible Clip on both sides of the waveform, which was about 10% THD. That was 17.1VAC, 73W. Just to see the difference, I powered down, then swapped the tubes back to how they were positioned when the amp came in (though I had increased the bias from where it was):

V7 27mA
V8 27mA
V9 20mA
V10 23mA

I turned up the output to visible clip, which was 15.1VAC, 57W. So, that current imbalance on the Primary of the Output Xfmr does somewhat limit the output under similar distortion figures. I don't have a fresh matched quad set of 6L6GC's here on hand to verify my suspicions of only getting full power with all four tubes matched.

I watched the tubes for around an hour between half power and full power, though never saw any evidence of red-plating.

I restored the tubes to the first balance shown, each half having one higher current tube and one lower current tube.

[Helmholtz]

Well, a bias balance pot like in a SF Twin would be nice here...
What model of Twin is it?

[nevetslab]

Well, a bias balance pot like in a SF Twin would be nice here...
What model of Twin is it?

65 Reissue Twin Reverb

[Helmholtz]

If you care about accurate output power measurements, you should measure RMS voltage at the output jack lugs, not at the dummy load.
Reason is that each phone plug contact involved can have a contact resistance of maybe 0.1R.
There are 4 contact points involved, which in total might steal around 6W at a current of 4A.

[nevetslab]

If you care about accurate output power measurements, you should measure RMS voltage at the output jack lugs, not at the dummy load.
Reason is that each phone plug contact involved can have a contact resistance of maybe 0.1R.
There are 4 contact points involved, which in total might steal around 6W at a current of 4A.

The RMS meter WAS across the output jacks, not across the cable end of the load.

[Helmholtz]

I plugged in my 4 ohm 200W dummy load, with AC Voltmeter across it,

Sorry, wasn't clear from your statement above.

[Leo_Gnardo]

Another way of assessing whether load is evenly distributed in a group of output tubes, laser infrared thermometer. I'm sure I've mentioned it before. Very affordable and handy as a shop tool say $20-40. Also reveals whether any output tube is hogging current or running cold. I let a set of tubes warm up say 15 minutes or so, handling a nominal signal, whether it's 100% or a fraction, it's easy enough to tell who's running hot & who's not. Only hassle is if tubes are set up in an uneven heating/cooling scheme, such as Mesa's power amps that have a fan blowing at one end of a row of tubes.

[Randall]

The way I look at it is, what would Leo do? He didn't think matching tubes was a thing, or even from side to side on a Twin as far as I know. And who complains about the sound of those amps? I just match them like a see saw. I put the heaviest and the lightest kid on one side, and the other two on the other side. Then I get on with my day. If Leo Fender and Jess Oliver didn't worry about if one tube was taking more of the load than the next, than neither should I, I should think.

[DrGonz78]

I think tube matching and the sort leads to further OCD. I am guilty of it when I finish up an amp where everything is perfect but the tubes are not perfectly matched. Then after the compulsion lets go I just play a guitar through the amp and the amp sounds great! job Done!!

edit: There is another process when dealing with SVT amp and other behemoths with 6 output tubes. In those cases it is very important to balance the load.

[Enzo]

A 100 watt amp, 15 watts either way won't be a noticeable difference, seems to me anyway.

[Helmholtz]

The '65 Twin RI isn't even a 100W amp.
It's rated at 85W like a BF Twin.

Actual full power output before clipping depends on:

1) B+ (squared) at full output, and B+ depends on actual mains voltage. A 10% mains voltage variation can change power output by up to 20%.

2) The residual plate voltage (aka "saturation voltage") of the power tubes, which lowers available plate voltage swing.
Saturation voltage can vary considerably (say between 50V and 100V) between individual new tubes and tends to increase with tube age.

3) Symmetry. With unbalanced tubes, one side will clip earlier than the other, thus lowering available "clean" power.


Regarding THD:

Each preamp triode might add something like 1% THD.
The Vibrato channel can be expected to have a little more THD than the Normal channel.
Fender specifies the 5% THD with the normal channel.