With a little help... KT88 PP UL Amp

[ray_moth]

You may notice that there is a step-network in the circuit in Hafler's article. This will deal handily with the problem of LF instability. However, it does complicate the biasing requirements.

The circuit by Franz Wichlas is deficient, IMHO, because of limited headroom of the concertina splitter. There needs to be intervening amplification, as in the Williamson. The only OP tubes that can reasonably be driven directly from a concertina splitter are low-power pentodes (e.g. EL84, 6V6).

 

[Herr Grau]

The Plitron is a gapless core: the Sowter is conventional E&I. Swapping to the Plitron in a Williamson can be done...

Thanks for the answer. This far I can follow...

...but one effectively alters the Bode map so the loop gains when global nfb is used have to be re-worked to avoid LF instability. Reminds me of the Eckland amp:may be worth examination.

(Although the discussed schematic ist from John Eckland...)

...and here we're leaving my territory.
Can this re-work be done from afar? (w/o the circuit)
Cause I somewhat get the funny feelings actual problems cause...
Ordering Plitrons is out of question. They would charge far over 100$ just for shipping. I'm still a student after all.

Eli D. pointed out that the Williamson circuit requires top notch transformers to prevent unstable phase shift oscillation. He says that the Mullard circuit is more stable in this regard.

I had read the thread an didn't worry too much about that, since I understood Sowter made some of the best OPTs available. Dont they?

"[...]Separation of the time constants permits less phase shift for the same frequency loss. Increasing one pair of coupling capacitors from .05uF to .25uF gives a five-to-one ratio of time constant for the two pairs of networks and increases the low frequency stability margin at nominal increase in cost."

In this schematic the coupling capacitors are the 0,22uF between the tubes, right? How would you modify them? Decrease the one pair or increase the other one?

 

[SY]

He's done that for you by staggering the R part of the time constant. There's a 5:1 spread in this design, just like in Hafler's modification. 10:1 would be even better, but probably not necessary. The Eckland circuit is an almost-exact copy of Hafler, with a standard KT-88 UL output stage. No trick circuitry, just standard, well-established stuff. Which is a good thing, IMO.

 

[Herr Grau]

Thanks SY, one thing less to worry about.

 

[Herr Grau]

Leaves the question of the OPT.
I'd realy like to use Sowter, but I also have access to Rondo-Müller and Amplimo to get toroidal OPTs. Would that make things easier?

 

[jwjensen356]

I don't think you can go wrong with this one:
http://vacuumtubebrasil.profusehost.net/Adapting the Ultra-Linear Williamson to 6550 operation.pdf

This was after Herbert Keroes and David Haller parted company, Hafler founding Dynaco and Keroes Acro Products. This amplifier uses the then-new 6550 and the Acro 330 OPT. Note that it has a vacuum tube rectifier.

I built two of these 50 years ago (!!!) and they are still very solid. I've variously used 6550s and KT-88s. Right now there are pairs of the Russian Gold Lion KT-88s in place. I used WWII surplus oil capacitors for the 500V KT-88 plates and electrolytics for the 6SN7 circuitry. Construction was with military-style terminal boards. I substituted the Dynaco 430 for the Acro 330 (The 430 was $30 then, the 330 was $40, big savings for a student at the time), The power transformers are surplused replacement transformers for the RCA 630 TV chassis. Surplused HV capacitors and power transformers were purchased at the original Radio Shack store in Boston (I'm estimating total cost for all to be $10 maximum)

Reliability through the years has been very good with minimal problems. Output tube replacement of course. Several years ago the electrolyics were replaced. Also, the inter-stage coupling capacitors were replaced (also stabilizing the output tube bias setting).

I'll never replace them.

John

Which design would you recommend?
I'm searching for the definite amp for the next many years, so I'm more than willing to consider every option, of course searching for the best value. I just need a complete, working schematic which I can 'cook' the amp from.

If you don't think the design is 'the thing', than the PSU-work isn't necessary now.

I don't have too many wishes, I want a KT88 PP Amp with non-rare pre-tubes (personaly I do like the 6SN7, but thats not so important)(yeah, it shouldn't need an external pre-amp...) and tube rectification. It would be most terrific if the design would work with one of the standard supplied Sowter OPTs...

 

[Herr Grau]

It looks good to my amateur eyes. Although I don't get the thing with all the potis and it's not good that it uses a third transformer. Every transformer is a major cost factor. Whats with the taps marked "x" and "y"?

Any other oppinions to which design should be preferred?

 

[SY]

T3 just generates the bias voltage by upconverting 6.3V to 120V. It's very low current, so the transformer costs maybe $4 or so. Not a major deal. It can also be done with a separate winding or a tap on the main power transformer.

For European voltages, you'd want a 220/12.6V to get the same output.

 

[jwjensen356]

It looks good to my amateur eyes. Although I don't get the thing with all the potis and it's not good that it uses a third transformer. Every transformer is a major cost factor. Whats with the taps marked "x" and "y"?

Any other oppinions to which design should be preferred?

In those days (the 50s) preamps were often powered by B+ available from the power amplifier (see octal socket on chassis skirt in the photo). The last few sentences in the section 'Amplifer Circuit' addresses this and the 'x' and 'y'. Eliminating this simplifies the circuita and eliminates one pot. The other pot provides for balancing of the output tubes.

John

 

[Herr Grau]

You're right, if I had read more carefuly...

Can I simply substitute the Selenium Rectifier with any other 50mA Si or Ge Rectifier?

Aaaand of course: Are the changes I made correct? (The removal of the pre-power-part...)
He says in the article that you should ground the middle tap of the 6,3V-winding if you remove the pre-power-part; should I order a transformer with middle tap in this winding or just ignore this information?

An externally hosted image should be here but it was not working when we last tested it.

 

[SY]

A silicon rectifier will give you a slightly higher voltage, but it's not like mains voltages are all that accurate. If necessary, you can slightly alter the bias resistor divider string values to get the right bias voltage adjustment range.

I would definitely try to get the heaters ground-referenced. A tapped winding on either transformer is a perfect spot. Bias it up to +60V or so with a resistor divider (or if you're Mr. Fancy Pants, an emitter follower like Morgan Jones's THINGY), bypass to ground, then bypass each heater pin to the chassis with small (1nF) 500V ceramic caps (the old fashioned disc kind).

 

[TheGimp]

T3 just generates the bias voltage by upconverting 6.3V to 120V. It's very low current, so the transformer costs maybe $4 or so. Not a major deal. It can also be done with a separate winding or a tap on the main power transformer.

For European voltages, you'd want a 220/12.6V to get the same output.

Any reason not to use a voltage multiplier instead? Caps and diodes are cheap and small. 6 caps and 5 diodes for a voltage pentupler.

Just a thought.

 

[SY]

Yes, you could do that, too, but since the bias string draws a little bit of current, and you want low ripple, you'll have to size the caps accordingly. But certainly do-able.

Checking some catalogs, low power 6V transformers go for about $3. Is the multiplier cheaper than that?

 

[jwjensen356]

Yes, as SY says you can substitute a Si rectifier (that is what I did in my original builds).

Here is another version of the same circuit, taken from an Acro brochure (look for the 60W 6550 Williamson circuit):
http://www.clarisonus.com/Archives/Trans/Acro55.pdf

You'll notice the preamp supply circuit is not there. There is also a pot shown for bias adjustment. This is necessary for the different bias voltages between the 6550 and KT88 tubes. I precisely set my bias for 70mA. At 500V that is 35W dissipation which I believe is the rating for the 6550; I believe 42W is the rating for the KT88. That sounds like the tubes are taking a beating but I have never had a problem. The tubes do not glow red. Say what you will about Russian technology, I believe their metallurgical technology is very good. Besides, they help warm the house in the winter

John

You're right, if I had read more carefuly...

Can I simply substitute the Selenium Rectifier with any other 50mA Si or Ge Rectifier?

Aaaand of course: Are the changes I made correct? (The removal of the pre-power-part...)
He says in the article that you should ground the middle tap of the 6,3V-winding if you remove the pre-power-part; should I order a transformer with middle tap in this winding or just ignore this information?

An externally hosted image should be here but it was not working when we last tested it.

 

[crispycircuit]

The extra transformer for the bias is a 6.3 heater xmer hooked up in reverse to get your 120 vac... It works because the bias supply draws very little current.... You also need a bias pot for adjusting the bias. I only see a bias balance pot. I'm enjoying this thread as I would like to build the same amp. If Sy and other fine knowledgable members give it the OK, I'm going for it. I do understand the feedback will need to be tuned to the xmers. Cheers.

 

[Bandersnatch]

Thanks for the answer. This far I can follow...

I had read the thread an didn't worry too much about that, since I understood Sowter made some of the best OPTs available. Dont they?


In this schematic the coupling capacitors are the 0,22uF between the tubes, right? How would you modify them? Decrease the one pair or increase the other one?

hey-Hey!!!,
Check out the Heathkit W5m schematic( The Free Information Society - Heathkit W5M Electronic Circuit Schematic ). It has a 2:1 spread of coupling cap-grid resistor time constants. More is better as has been discussed. It is *STILL* a very long FB path. IME, shorter is better; consider returning FB to the second stage and leave the first run open-loop. It is a Class A triode stage running at fairly small signal level;there won't be objectionable distortion from that...
cheers,
Douglas

 

[SY]

The Heathkit also uses the 12AU7 for voltage amplification. Curious what the distortion spectrum looks like...

 

[Bandersnatch]

The Heathkit also uses the 12AU7 for voltage amplification. Curious what the distortion spectrum looks like...

hey SY,
IMO, getting rid of the 12AU7 is first order of business. The 6CG7 is a fine substitute, and since this is a ground-up we can get a bit more creative. 14N7 comes to mind... as well as a host of other interesting valves. 6DN7 with its sections split between stages for one.
cheers,
Douglas

 

[navin]

You're right, if I had read more carefuly...

Can I simply substitute the Selenium Rectifier with any other 50mA Si or Ge Rectifier?

An externally hosted image should be here but it was not working when we last tested it.

The Heathkit also uses the 12AU7 for voltage amplification. Curious what the distortion spectrum looks like...

Well, if there are no rudimental objections against Williamson....

Q: With a 6SN7 for both input and driver stages the amp will be very quiet but it might not have enough gain. If one needs more gain cant one use the 6SL7 for the input stage instead of the 6SN7?

here is a simplifier schematic (for tube newbies like me)
CayinTA35

 

[Herr Grau]

I've pretty much settled on the Keroes circuit by now.

With exeption of the output stage / bias, the Acro and the Keroes circuit are basicaly the same. I think, it's both necessary to adjust the general bias voltage and balance it to the tubes, but that implementing a poteniometer for each of the tubes is more senseful than doing one for overall voltage and one for balancing. If I understood correctly, the bias voltage is dispensed to the point between coupling capacitor and grid resistor.
Based on the Keroes schematic I drew up, what I thought might be about right:

Klick!

Could you tell me
1. Is that right?
2. If so, how do have R19, R20, R21, R22 and R23 to be?
3. I'm pretty certain, that the bias voltage supply can't stay this way, can it? What would I have to change?
4. Brian Sowter answered my question if he could supply me a fitting OPT:

[...]According to the article the spec is [...] 10Hz to 100 kHz ± 2 dB.
The U070 certainly is the closest we can get to this spec. The HF cut off
does not quite reach -2 dB at 100 kHz.[...]

Is that anything of a problem?

Thanks so far,
Herr Grau