KT88 with chub clem

[justinsweber]

I built a Chub Clem. it wired with the strap between 1 and 8... I think ? Ive used EL34s no problem. Is it true we can use KT88s too?
What would they be biased at? aside from output any real benefits? The circuit was designed for 6L6s? (IE the circuit is designed to get the most from the characteristics of the 6L6?

My 6L6 STR chub chem no feedback was getting 3watts RMS before clip. Checked the EL34s, if I recall it was 4watts or so. Maybe the KT88 is good for 5watts ?


Thank you

[Shannon Parks]

Plug'em in - no re-biasing is needed.

Shannon

[justinsweber]

which woudl you recomend?
Does using the higher output tube create a larger sweetspot/ less distortion?

Thanx Shannon :-)

[Shannon Parks]

Just looked back at my early KT88 vs EL34 data on the Clementine. The EL34 is a little easier to drive to max output (625mV vs 800mV), but otherwise they look similar out to their max output: 4.75W.

Shannon

[Fiddler]

JJ KT88's sound fine in my Clem; but if you are tube rolling don't neglect the KT77. I'm currently using Gold Lion reissue KT77's and they sound better than anything I've heard to date.

[justinsweber]

Did I understadn this correctly... KT88s and El34s have very similar max output in this circut? I perfered 6l6Gs to the EL34s.

Justin

[Shannon Parks]

Did I understadn this correctly... KT88s and El34s have very similar max output in this circut? I perfered 6l6Gs to the EL34s.

Justin

Correct. You could adjust the biasing more aggressively for the KT88s, and your linearity will increase a little, but the max output will be the same.

Shannon

[justinsweber]

I asked a friend for help. He looked up the KT120 which needs 1.8amps x 2 and the 6SL7 needing 300mv so ~3.9 amps...
The Edcor XPWR105 has 2 6.3 windings @ 2.5 amps... so there is enough juice to light of a Clem KT120. I think the better choice if you want to do it is the Edcor XPWR178 which has 3.5-0-3.5 @ 6 amps and then there is a 5v 3a for a rectifier :-). I might get a custom one made 330-0-330 @ 200ma and 1 6.3 @ 7amp.

On my Clem with Russian 6l6GC I get 2.5 watts RMS, with KT88s I get 4watts RMS. I hope with KT120s 6 or 7 watts?

[Shannon Parks]

Checkout the SE Amp CAD shots below. Note that max dissipation rating is basically not a problem with the Clementine voltages. We are looking to adjust the bias resistor to where our linearity (the tube curve linearity) is below 5% at max output power (i.e. clipping). The bottom pic shows a KT88 (or KT90 or KT120) operating in a standard Clementine circuit (the stock 720 ohm bias resistor network). The middle photo shows the KT88 operating with a 380 ohm resistor. This seems about optimal. The top circuit shows us at 50W dissipation, like we are running a KT120. But we've lowered the voltage differential between the grid and cathode so much, we are limited in the driver voltage. This operates the tube in a more linear operation, but is poorer than the middle pic operating point as we are not optimized (i.e. we start clipping at a lower output power than pic #2).

The previous paragraph seems a little confusing, but when we set up a operating point for SE amps, I like to see the distortion slowly increase until I know I am clipping at 5%THD. Triodes will normally do around 3-5% THD at their max output. If I get to >5%THD before power tube clipping, I know that I am overdriving the power tube and we have grid current (clipping from the driver). This means I've not maximized my output power and I need to adjust the circuit. I always want the clipping to happen at the power tube first (though never really!).

So switch to a 390 ohm or 400 ohm, 5W bias resistor for KT88s and above. For even more output power, switch the the ULtra circuit with feedback. The UL tube configuration has more mu, thus more voltage swing at the same operating point. The feedback is needed on the ULtra to lower the output impedance back to a value similar to the triode value.

Shannon

[justinsweber]

Forgive the ignorance... Which resistor to replay R?. What do you think the change in output will be for a KT88. I was getting a solid 4watts RMS just as it was.
Should I assume its the resistors below?
R17,R20 1.2K, 3W
R18,R19 1.5K, 3W

Chnage these to 800ohm 3W?




Thank you Shannon

[Ty_Bower]

Should I assume its the resistors below?
R17,R20 1.2K, 3W
R18,R19 1.5K, 3W
Change these to 800ohm 3W?

R17 and R19 are in parallel with each other, as are R18 and R20. 1.2K in parallel with 1.5K gives an effective resistance of 1/(1/1200 + 1/1500) = 667 ohms. Your amp is already setup partway between Shannon's second and third simulated examples. If you study the numbers in the "Load" section of each simulation, I think that gives you an idea of what you could theoretically expect to achieve from the given circuit. I say theoretically, since if I'm reading these charts correctly it is predicting an output power between 8.42 watts and 9.25 watts. You say you are measuring 4 watts from your amp. Perhaps your measurement technique is imperfect, or the model is optimistic.

Anyhoo, there are other numbers to study besides just the predicted load power. There are also numbers for output impedance (Z out) and distortion figures (2nd and 3rd). Plus, as Shannon points out, you need to consider the driver stage's ability to swing sufficient voltage and/or current into its load. In short, if you want to follow Shannon's suggestion and optimize this circuit for KT88, you would replace R17 through R20 all with 800 ohm parts. I would not expect to hear a dramatic difference.

If it were my amp, I would probably just leave it as is. I've played quite a bit with another single ended, octal tube, circuit board amp. It started life with 560 ohm cathode resistors, but ran just slightly too hot with EL34 or 6L6 family tubes. I ended up increasing the cathode resistors to 810 ohms. It still runs KT88 just fine, but no longer cherries the plates of the smaller tubes. Another option for you to consider might be to keep the 1.2K parts you already have and replace the other two with something like the 620 ohm Xicon 286-680-RC. They are five watt rated - check first to make sure the physical dimensions will fit the board. Put them in parallel with the 1.2K parts, and you get an effective 430 ohms which is better optimized for your KT88. If you want to swap out your tubes and go back to EL34 or 6L6, remove the 1.2K parts (and maybe install a DPST switch in series with them, so you can toggle it later) and you are back to 680 ohms.

[Shannon Parks]

Just back from a little Christmas visit with relatives. Indeed, I second everything Ty said.

Shannon

[justinsweber]

Its my intension to run this amp with only KT88s.
I asume that this means R17,R20 shoud be 800ohm, 3W and R18,R19 should be 800ohm, 3W. This will prob yield a output @ ~8watts?
I noticed the modeling shows a 5k load. KT88s perfer 3.5k or 3k loads ?
Would the same parts be used for a 3.5k laod? Is that what youd sugest?

[justinsweber]

My friend Tom did some modeling for me.
with 430ohm cathoid bias, idle is 29.2watts and rms is 4.05watts which was 1 sugestion for optimizing for KT88
with 380ohm cathoid bias, idle is 32 watts and rms is 4.21 watts which was a 2nd sugestion for optimizing for KT88.
The stock config which allows for 6l6GC, KT66/KT77 EL34 and KT88 tubes is 660ohm cathoid bais, idle is 21 watts and RMS is 3.5watts. In this config, I was getting 4watts RMS.
We tried modeling this with 3.5K loads and distortion was up and oupt was down... go fig.

So, Im using the brlient thinking of Shannon's stock circut with a 5k load.
Meaning 1.2k + 1.5k =660hm per side.

[Shannon Parks]

Would the same parts be used for a 3.5k laod? Is that what youd sugest?

I had a stock Clem 6L6 Edcor in my hand and didn't even think of this, but yes, I would use a 3.5K output transformer for a Clem only using KT88s. The measured Rp on my eight KT88s of various old stock manufacture averages 1100 ohms at this operating point. I've attached new SE CAD shots, with a 667 ohm, 400 ohm and 195 ohm cathode bias (equivalent parallel) resistors.

So I would suggest the ~400 ohm bias resistor (72-RWM410-680-5 & 72-RWM410-1K-5 in parallel, or use a pair of 800 ohms 3W ones of your choosing) and 3.5K output.

Shannon