What OT primary for KT88 SE?

[tubby]

hi all,

i'm designing a SE kt88 amp at the moment and i'm not sure what
primary to use on the OT. i'm going for about 350V on the plates, cathode bias, tetrode mode.

in the tube data sheets for kt88s there's only the load value
for push-pull / class ab. is there a way to calculate the correct value for SE use from those?

[tubeswell]

The KR Audio site says 3k for SE

The AX84 site has a design for SE power amp running about 370VDC on the plate with 2k5 20W OT

Somewhere in that ballpark I guess

[tubby]

just did a test run with 2,3k primary. not good....

after i had finished playing full throttle for 30mins or so (awesome tone),
there was an intense blue glow in the upper part of the kt88 and a hum coming
from the speakers. then i saw sparks through the cooling vents in the plates.
turned off the amp immediately. tried it 5 min later. no sound, but literally fireworks in the kt88.


i've never seen this before, but i'd guess i just fried the screens because of a too light primary load
there was no red-plating or anything to give off some warning sign, happened all in a few secs.

can anyone confirm this from experience?

[Steve Conner]

Yikes! That really shouldn't happen. 2.3k is a light load (equivalent of almost 10k plate-plate) but 350V isn't much voltage at all. I think your KT88 might have been gassy.

Was it new or used? What kind was it? Do you have any other tubes you could try in the circuit?

One more thing, did you use a screen resistor?

[tubby]

Hmmm...it was a brand new JJ KT88 biased at around 90%.
B+ is actually more like 370ish, but still...

So far i've tested Sovtek 6L6s and a Svet EL34 in the circuit without any problems.
However i've used a different primary load with those tubes (4,5k).
I had the KT88 running on that primary load at first too, but then got the
(as it seems wrong) impression from the datasheets that 2,3k are suited better
to maximize output and tone. I like the sound of the KT88 in that amp,
it just sounds bigger. I can't spare another KT88 to see if it happens again...

The screen supply comes off the B+ node, through a 10k R, 33µ filter and a 1k screen grid stopper.
I might try doubling the value of the screen grid stopper,
if that does not change the compression characteristics too much.

Also, i've come across a SE design which uses a screen grid stopper as large as 25k
in combination with a 50µF cap connected directly to the screen grid in order to prevent extreme screen dissipation.
I've seen this only in hifi and bass amps so far.
Any idea what that would sound like for guitar?
And wouldn't I lose power because of the much lower screen voltage?

[GibsonLover]

That 2.5K / 5K primary OT question is really a toss-up. I've noticed with 2.5K the second order distortion (the good kind) is a little higher, power is a little higher, and damping factor a little lower. Depending on your mood, the 5K has slightly less power, slightly less second order, and slightly better damping factor. That was with a 6L6GC in triode mode, but your KT88 will have the same generalization. For more in-your-face sound > 2.5K. The difference is not huge, though one might think (the reason I did the experiment) with a doubling in value that something may go horribly south. The output section is a good place to screw around though, especially with transformers and speakers.

[tubby]

that's interesting, i didn't notice any difference in sound. it was only
when the flashes occurred that i realized something is different
with the lower impedance setting. well, since i couldn't make out
anything soundwise, i guess i'll just leave it at 4,5k then. it's
cheaper that way....

[Guitarist]

I've been thinking of building similar amp too. What brand and model of OT would you recommend? KT88s have such a HUGE sound!

[6267]

I'm thinking the load line is too low & the screen supply too stiff for that tube. That's weird though, because that sounds about right for a KT88. No accidental shorts between B+ and screens that might be putting the screens up too high? Are you sure the correct load is on the transformer secondary?

[Merlinb]

Hmmm...it was a brand new JJ KT88 biased at around 90%.
B+ is actually more like 370ish, but still...

So far i've tested Sovtek 6L6s and a Svet EL34 in the circuit without any problems.

This is a common prblem amongst guitar amp DIYers. The crux of the matter is that the KT88 isn't like the other valves we're used to - it has way more gm for a given screen voltage. Ultimately this means that you can't just run the screen voltage at roughly the same voltage as the anode, unless you use a MUCH lower load impedance, which isn't suitable for SE. With the screen voltage so high, you get enormous screen dissipation when overdriven.
With 370V anode, and 2.5k load, you should really be looking for about 200V on the screen.

Also, i've come across a SE design which uses a screen grid stopper as large as 25k in combination with a 50µF cap connected directly to the screen grid in order to prevent extreme screen dissipation.
I've seen this only in hifi and bass amps so far.
Any idea what that would sound like for guitar?
And wouldn't I lose power because of the much lower screen voltage?

That is simply a sepatate B+ supply for the screen, bringing the screen voltage down to a more appropriate level. No, you won't lose power (unless you used a stupidly low screen voltage of course). If you expect any more KT88s to survive, then that's what you shoudl be doing too!

[tubby]

I've been thinking of building similar amp too. What brand and model of OT would you recommend? KT88s have such a HUGE sound!

I've used a custom OT wound to my specs by a german tube electronics guy.
It's reasonably priced and seems good value for the money. You can check it out at:

http://www.roehrenendstufen.de/

[tubby]

With 370V anode, and 2.5k load, you should really be looking for about 200V on the screen.

I used a 10k dropper for the screens and a 1 screen gris stopper so the screens are not at full B+ voltages. I'll measure the voltage and adjust the Rs if necessary to drop the voltage to around 200V.

That is simply a sepatate B+ supply for the screen, bringing the screen voltage down to a more appropriate level. No, you won't lose power (unless you used a stupidly low screen voltage of course). If you expect any more KT88s to survive, then that's what you shoudl be doing too!

Ok, sounds reasonable.

Thanks for the input!

[tubeswell]

With 370V anode, and 2.5k load, you should really be looking for about 200V on the screen.

Hi Merlinb

I hope you don't mind me taking a learning opportunity here, but I am curious to know how you derived the screen voltage from the anode voltage and the anode load resistance?

[tubby]

This is a common prblem amongst guitar amp DIYers. The crux of the matter is that the KT88 isn't like the other valves we're used to - it has way more gm for a given screen voltage.

gm would be the amplification factor? i always get that messed up with transconductance...

Cuz I was wondering: even though I'm using only half a 12ax7 to drive the KT88, the amp really starts to roar with power tube distortion on the last third of the master volume dial. The same happens with el34s and 6l6s. And I always thought you couldn't drive a power tube like a KT88 not even near saturation with just one preamp triode. Is that maybe because it is a SE design and thus needs only half the driving voltage?

[Merlinb]

Hi Merlinb

I hope you don't mind me taking a learning opportunity here, but I am curious to know how you derived the screen voltage from the anode voltage and the anode load resistance?

Quite simply, from a load line. With pentodes/tetrodes, you want the load line to pass through, or just slightly below the 'knee' of the curves. If it passes WAY below the knee, you run into problems when overdriving- the screen tends to burn out!
Looking at Tubby's load line I could see that the loadline did exactly that, so immediately we see that the screen voltage is too high, or the load impedance is too great. In this case we don't want to reduce the load impedance too much, because you'll end up hitting cut-off really soon.

The Genelex datahseet happens to provide a graph for 200V screen, and the same load line is about perfect then; it passes right through the knee. Even if a graph isn't available, there are other ways to figure out the screen voltage indirectly, with a little info from the datasheet.
Desiging power stages is simpler than you think!

gm would be the amplification factor? i always get that messed up with transconductance...

gm is transconductance or mutual conductance ('g' is the symbol for conductance, and 'm' stands for 'mutual').

I always thought you couldn't drive a power tube like a KT88 not even near saturation with just one preamp triode.

Common misconception. Think about it, a typical 12AX7 can deliver a swing around 2/3HT, which is at least 200Vp-p in most cases. A power valve begins to clip when the peak input voltage exceeds the bias voltage. If the KT88 has a bias voltage of -35V, then anything over 2x35V=70Vp-p will begin to clip. Now imagine 200Vp-p!!!