Replace 4xEL34 with 2xKT88

[SlashLP97]

Hi guys, I'm a new poster here, but I've been lurking and learning for about a year now. I built a guitar amp that has 4 EL34 power tubes in the output section. The output transformer is 1.9k rated for 100W. I want to replace the EL34s with KT88s, but my EL34s are physically placed too close together to safely run all 4 KT88s, as they are much bigger. Instead of attempting to move the tube sockets I figured I would try getting 100W output from 2 KT88s, just using the outer tube sockets (so there is no physical interference issue).

Technical info:
Plate: ~460V
Screen: ~458V
OPT: 1.9k 100W push pull from Magnetic Components, Inc.
4/8/16 Ohm taps
NFB: 47kOhm + Depth control

My idea was to place the two KT88s in the outer tube sockets and bias them to ~60mA each, thus having idle dissipation ~30W. Then I was going to change the NFB resistor to 100k + Depth control, roughly doubling the actual amplitude of voltage placed on the grids, since there are only two tubes and they have roughly the same transconductance.

By this logic I should not have to double the impedance of the cab I am using (aka 8ohm cab, 16ohm impedance selection as if just pulling two of the EL34s). Is this correct enough to attempt an experiment or should I expect explosions?

The current am design has been tested through and through, there are absolutely no problems and it has been gigged with numerous times, played while dimed for decent periods and the like, so there are seemingly no major electrical problems (underrated transformers and the like).

I would greatly appreciate any feedback you guys have!

[es345]

The config you want to try may kill the KT88 because you will overshoot its maximum limit of the cathode current (230 mA max ) compared to 2*El34 (2*150=300 mA max).

[pdf64]

Why bias to 60mA - it seems rather high?
The data sheet http://www.mif.pg.gda.pl/homepages/f...086/k/KT88.pdf p3 indicates that there may be insufficient B+ to get 100W, 560V being needed for that. Pete.

[TransLucid]

I think it will work fine, although you may not like the sound of 88s vs. el34s. The one thing I would check is the relative gains of the two tubes. If the 88s are significantly different from el34s, you will need to adjust the feedback factor. The characteristic curves of the two tubes are not that different, however.

Your plan for changing the feedback resistor sounds a little sketchy to me. I would measure the output stage gains with each tube complement and no feedback, then determine the feedback factor with the el34s by reconnecting the feedback network, then determine what feedback resistor is necessary to give you the same closed loop gain with the 88s. As long as that gives you a reasonable amount of gain reduction, you should be ok.

[SlashLP97]

@es345: The El34s running at max current of 75mA each produce about 120W, if I were to keep the current to the max of 230mA that would be almost exactly 100W with a 460V B+/1.9k OPT. Is this incorrect?

@pdf64: The 60mA bias current is to keep the OPT impedances correct, if I were to bias to the standard 30-40mA I would have a gross impedance mismatch, fixable by using double the speaker impedance as my output impedance. Also if you look at the Genalex KT88 datasheet it says that 2x KT88 biased at 60mA each with 560V B+ produce 100W output. Granted my B+ is not that high, so I may come in a little short of 100W, but the impedance would be much closer than with a lower bias.

@TransLucid: Yeah, the idea for the NFB network was just a starting point, in fact I may just pull the EL34s, put in two KT88s, keep the feedback network the same and measure the max output power and adjust that way, also noting power section breakup as my "Maximum PI output" point. Actually the reason I am switching to the Kt88s is to have a less pronounced mid-range with more low end punch and some more high end definition. This particular amp is a high gain kind of number and the EL34s make it sound too tame IMO.

Thanks for your input guys, I'm gonna give this a shot (in steps of course) I'll repost with the results.

[pdf64]

The 60mA bias current is to keep the OPT impedances correct, if I were to bias to the standard 30-40mA I would have a gross impedance mismatch, fixable by using double the speaker impedance as my output impedance

? What is your basis for correlating bias and output impedance?
Re B+ discrepancy between the datasheet and your amp, 460/560 = 0.82, and as power is derived from V squared, P = 0.67, eg your 100 watt amp is only 67 watts. Your slightly lower than datasheet OT primary may make that up a little, but it will still be way short of 100W.
Re bias, bear in mind that the limits of a current production KT88 will be different / lower than those from Western Europe made in the heyday.

[SlashLP97]

Point well taken. Would I be better off doubling the OP secondary impedance and running 2x KT88 biased at 35mA (Standard 2 output tube 50W bias) if I were hellbent on evaluating the KT88s?

[pdf64]

That's what I'd do (it is what I'm doing regarding an EL34 amp that I'm using for KT66).
Sorry, I hadn't taken in that you had been planning on leaving the OT primary at 1k9. I've no idea how well KT88 would work with such a low OT primary. Pete.

[SlashLP97]

I'll givethat a shot first. If i like the soynd I'll get really brave and try to get the two to crank out more power. Thanks for your help!

[Wilder Amplification]

You're not going to get 100 watts out of a pair of KT88s in a 100 watt amp that was designed around 4 x EL34s with a 1.9K primary. While the power transformer definitely has the power available, the load impedance is too low to allow for maximum power transfer to the load at that B+ and screen voltage.

If you were to draw a load line of this combination, you would see that the load line would cross the Vg1=0 grid curve well above the knee. What this means is that at maximum swing, plate voltage will not drop low enough to provide maximum power transfer. Furthermore, since the plate voltage at the peaks of the swing will be higher, the power dissipated at the peaks will also increase as well as the power dissipated during any clipping duty cycle, which causes your average dissipation to increase and hence the valve overheats.

In order to get 100 watts out of a pair of KT88s, you would have to have a dual rail design with something like a 4.5K Zplate-plate, a B+ supply of around 00-700V with a screen voltage of around 300-350V. Or it could happen in an ultralinear design with about 600V B+, but UL designs don't like output clipping at all.

And no, this is not something you can correct for via biasing hotter. You will just overheat stuff.

You can run a pair of KT88s in that amp. But that amp does not have the correct B+/Vscreen/plate load to allow you to get 100 watts out of a pair of KT88s. It would require completely different transformers and power supply redesign to even make that happen.

[SlashLP97]

Wilder, thanks! Great explanation.

Assuming I were to put a pair of KT88s in, bias to 50W and then double the output impedance how hot could I get away with biasing them before I started to have OPT problems?

[MWJB]

You just said you were biasing to 50W, 25W per tube?

25W/460vdc = 54mA per tube...they'll probably take 60mA (28W per tube) at idle happily enough as long as they are not Shuguang KT88...and will probably "work" OK from mid/late 30's up...so I'd bias to say 40mA have a listen, try 45mA have a listen...and so on. You may not need to run the highest possible currents to get a good sound.

If the OPT is designed for 4xEL34 you won't have problems with 2xKT88 in fixed bias at regular idle currents (<60mA per tube).

[SlashLP97]

I have EH KT88s. You aren't supposed to bias the tubes at 100% of the total dissipation you want though. I know that tube amp power rating is more of a "number of tubes" kind of number than an "actual power output" number, but it sounds like if you were to get up into the 50-60mA range you would be in class A almost. I do want to bias higher than the standard 35mA as EL34s would, as I have heard that KT88s sound great in the class AB (but getting close to class A) style configuration. So no problems biasing up to 60mA? I wouldn't start there, but I wasn't sure if there was some sort of recommended maximum.

[MWJB]

"I have EH KT88s. You aren't supposed to bias the tubes at 100% of the total dissipation you want though." Aaargh! Nooooo!

In an efficient design the "power rating" is pretty well the "actual power output" rating....but you're not running the KT88, or even the EL34 at max efficiency type voltages.

In fixed bias the plate current will rise far above idle current under signal it is typical to bias fixed bias output tubes to 50-70% plate dissipation. In cathode bias, maybe 70-90% as signal current may still rise, but not to such high peaks as in fixed bias. That 60mA I mentioned was the reasonable maximum.

Forget class...most guitar amps run B+ voltages in excess of what would be required for class A operation, without burning up the tubes. Bigger tubes hold more heat (single ended 6V6/EL84 can sometimes take more than 100% plate dissipation in SE mode), heat kills tubes & OPTs.

[SlashLP97]

Was the painful yell because I have EHs or because I said you shouldn't bias to 100%?

And in efficient designs, that is correct. But look at a 100W Marshall. It's 100W because it has 4 tubes. They actually output 120-140W if you look at the output with an oscilloscope. Some companies actually report power ratings (Like ENGL and sometimes Peavey) but it is usually a number of tubes measurement (aka this amp has 4 EL34 or 6L6 power tubes, therefore it is 100W).

Thanks for the insight on the biasing. I'll probably start at around 40mA then go up from there and see how I like the sound.