Ralph Power amplifier

[stigla]

Bernard;

I've studyed the Quad II schematics, and the output stage reminds me a bit of what you are talking about. Do you know what this configuration is called? (is the quad II class A?)

What about PPP (paralell Push-pull)? I've seen a schematic over a PPP KT-88 amp, and it had an impressive distortion diagram. limited at mostly 3'rd... This was even in class AB!! Very close to a good SE amp!

Both of these configurations have something to do with what you are talking about, I think... Well, its just a thought...

[dice45]

Stig,



1st, have to stress that, i read a lot, i figured out many configurations, this one appealed to me. But at the moment i am talking from theory and from observations i made with other amps. No practical experiance yet with the circuit. I come back with results as soon as i have them.



The Quad II never attracted me particularly, so i do not have the schematic in mind. But, interesting to know they did soething similar, those amps have quite a rep.



The biasing is important as the tubes age. They change their plate resistance and their transconductance. I would be extremely gateful if they wold consider to that synchronously, but i am afraid, they won't do me that favour.

The CCS keeps the total constant, but the tubes get unbalanced with aging. Which is not welcomed by the output transformer at all.



LL1668: please do not waste them as CCS-replacement in the cathode, their high OPT-like core quality and inductance is neeeeeeded as plate load. As they can be wired in a centertapped, balanced mode, they are perfectly suited as plate loads in differntial pairs and, both coils being on the same core, add further help to balance the stage!

I have LL1667 and LL 1668 as plate load in preamp stage 1 and 2 and in power amp stage 1. Quite costly, and i would not consider them if i ha to buy them now.

I certainly do not waste them in the cathode.



The virtual ground of a differntial pair is quite low Z. So impedances do not have to be so big to achieve proper low frequency behaviour. I would go with PS chokes of 10-20 Hy for a 1st try (they also can handle the power).



Coloration of the CC device, be it SS or choke: i have to try this myself, i am a newbie here probably more than you

[Mohan Varkey]

Stig,

The Quad II (Class-A PP) and McIntosh amps (Class-AB PP) use cathode windings. This is a form of feedback. The well known ultralinear is also feedback. You may also call it Distributed Loading. In the McIntosh the load is split 50-50 between the cathode and anode. Bifilar wound transformers were used. In the Quad II, there are separate windings for the cathode and anode.

I am a stickler for transient response and dynamics. I have experimented with local and global feedback in many designs. None of them appeal to me even though they look good on scope. In other words, if you can live with a laid back presentation without the hellfire and brimstone that I like, then, by all means go for feedback.

If you are not a purist like me, then consider parallel single ended amps. I am at present developing a SE amp for Quad ESL57. There two designs; the parallel single ended topology (15W output) uses 300V on the anode and the single ended design (30W) uses 930V on the anode. Both are two stage designs and will use a specially developed output transformer that will replace the input transformer in ESL57. It will be interesting to see how these perform.

Mohan

[yves]

Stigla,
The ultrapath I was thinking of was to shorten signal
path. It enables the output signal to get to the cathode
directly. I think that the beauty of the ultrapath
(earliest use of it was in WE amps!) is that it references
the output to the input directly. Remember that the
input voltage is really developed across the grid and
cathode.
If the ultrapath cap is made very large, as in
tens of uF, then I think the hum cancellation is not
effective. It is there only as a bypass to ground, no?

rgds
Yv

[stigla]

Mohan;

Thanks for the quad info! Hmmm, quite different to what I thought
Well, I like to think of myself as a purist, but regulary find that my wallet have other plans....

I DID (in the beginning) consider a PSE design with 2x 2A3 (andrea Cuffolini's). I designed a heck of a driver (chokeloaded 26 into a IT loaded 31!!! All DHT!!) Then, I learned som more about drivers, found that the 31 perhaps wassn't as good as I thought to drive the 2A3 (but still a good tube). And my wallet also had a few words with me

Then I read the Lynn Olson / Aurora article and thought... "Hmmm, maybe PP ain't that bad after all..." Ran over the Ralph Amp and now I am here.

One thing I do know; NO NFB!!! (gimme class A!!!!!


Yves;

So you mean like; adding a cap in the IT stage? How to calculate the value?
I know that, as I have used in the Outputstage, the cap is 3 times smaller than the BP cap. This effectively reduses ripple, as well as shorten the signal path.
But I think I will try out your advise. The Electraprint uses a 40uF... and a strange 100R resistor in the PSU end... Just to burn of Volts or has it other issuses?



Dice 45;

Have you seen the CCS thingy at Aloha Audio? Used in the Aurora amps? Interesting stuff!! (but not really a purist thing though...)



Stig

[yves]

Stigla,

Yes, I mean to connect the cap from the bottom of the
IT to the cathode directly.

I hv a "cutting" of a post in the Audio Asylum that explains how
to calculate the value for the Ultrapath cap, but this is assuming
there is a bypass cap at the cathode resistor. Here it is:

"The explination for how the circuit works was described to me by thorston at the london show last year, I will try to relay my understanding of it. If the B+ rail falls, either under dynamic signal or ripple conditions, then an unwanted AC voltage will form across the OPT. To correct for this an error signal of amplitude reduced by the gain of the output stage and inverted in phase has to be fed into the grid of the output tube. A convinient way of removing the need to invert the signal is to add it into the cathode circuit so it is subtracted from the input. Since the AC conditions of the cathode are determined by the cathode by-pass capacitor, it is easy to inject this signal using a capacitive divider made with the cathode capacitor as the lower leg and an additional "ultra-path" capacitor connected between the cathode and the B+. To get the correct cancelation the AC voltage divider needs to have the correct ratio, the ultra path capacitor should therefore be the gain of the stage minus one times smaller than the cathode by-pass. Also the "ultra-path" capacitor should be of a similar construction type and characteristic to maintain this ratio over wide operating conditions.
If the value is severly wrong then positive feedback on the signal will occur, but since the dynamic impeadence characteristics of the PSU are frequency dependant, the amount of positive feed back will also be frequency dependant, causing a rise in the bass response.
HTH
Chris "

I think that in practice, one has to go about it with trial and
error to determine the best value cap, or caps in parallel if
need be. In the Electraprint circuit, I think they use the
largest value of cap that is practical since no bypass is
used at the cathode resistor. I am guessing here !
The 100 ohm looks like a filter combo when used with the
ultrapath cap.

Let us know how it works out for you.

cheers
Yv

[planet10]

[stigla]

Yves;

Thanks! So, as I understands it; If I would be using a BP cap, it would have been 220uF. Thus, the UP-cap would have been approx. 220uF / (stagegain-1) = 220 / (22-1) = 10,5uF.
But this is WITH BP cap, so it would have to be higher, yes? around 22-33uF should work... but this will be experimental stuff when I get the Amp up and running.

Do you think that the 33uF cap to a 100uF cap at the output is wrong? I read in Tube CAD, and the way I understand it it should be 3 times smaller.
Well, using your method, it goes something like; 100uf / (4,2-1) = 31,25uF. So 33uF ain't all that wrong. The value is probably not very critical, or what? The cap will have a tolerance of around 10% anyways...

Thanks for all the help everyone!!!

I cant wait to get it ready, It will be used on Klipsch RF3-MKII. 98dB /1W.

[stigla]

Hello Yves;

Looking about in the RCA-13 i stumbled over a so called "Class A audio frequency Amplifier" stating to provide 12 W, at page 185, using 2x 2A3 and 1x56 with an IT. Intersting circuit!

Though the Class A aspect can be discussed when the tubes are biased at 40mA and -60V....

Well, to the subject; The IT loaded 56 was the intersting bit, it had implemented Ultrapath! Though using a 5uF BP cap on a 2,2k resistor is strange, and the 2uF Ultrapath cap is well overdimensioned. Using your equations, a 5uF cap in that configuration would need only around 0,39uF...
Even more interesting is that the PSU drop resistor of 5K had NO filter cap between the IT and itself. just; 5k resistor ----> 2uF Ultrapath cap and IT primary...

Why like this? Would a filter cap from the 5k'er to ground destroy the ultrapath?

The total of the UP and the BP cap will only work as a 1,43uF cap when seen from the PSU.
This is not very unnlike the Electraprint circuit, but it had only a smallish 100R resistor. (and no BP cap.)
So, if I would try making myself a Ultrapath in the input, would I have to leave the VR tubes (0c3's) out?

[stigla]

Okay, finally got some parts in, and started breadboardnig. One channel.

First experiences;

hooking up all the parts where easy, then applying power to the amp, using the old light-boulb trick. Everything seemed okay, no smoke and no flashes anywhere. Good.
Removed the lightboulb and plugged in 230V to the power tranny. Waiting for some nervous 35 seconds... Hmmm, B+ of 320,6 volts... the 0c3's glowing purple... but what is this? 49,3 volts on the cathodes of the 6b4gs? errr.... hmmm, must be the higher voltage I gained from swithcing from the GZ-37 to the 6D22S's...

214 volts on the ecc99 B+. Very nice!

hmmm, 6,43 volts on the 6b4g filaments... 6,38 to the ECC99. Just a tad too high...

man, the 6D22S are getting very hot.

I let it run for a good 30 min. Then swithced the 8ohm power resistor to a speaker. Hum level to high... Guess I'll try out changing the 2 paralelled 5,1k caddock's with a 1,5k and a 1k'er in series, with a 22uF cap in the middle later today.

Well, so far I am very pleased, no smoke and just minor problems. Only thing I did'nt expect was all the heat! All the tubes get much hotter than I imagined, must take this into account when I'm to (re) design the chassis.

summary;

- a little higher filament voltage than I would like... what would be the easyest to do here?

- not the exact 45 volts I would want on the cathodes of the output-tubes. 49,3V = 98,4 mA through the tubes. Maybe Ill try out 750ohms on the cathodes giving me perhaps around 60mA and 45 volts...

- very much heat dissipated. not awefully lot, but much higher than I thought.

- A very nice light show in dim light!

- Hummm.....

I hope someone can help me out with the filament voltage-problem.... using series resistors seems like a bad idea... Probably not very exact windings on my power trannys, but I overdimensioned (for current that is...) the heaterwindings on purpose, so I guess thats the reason.


and thanks for all the help so far, would prob. not have buildt this thing if it werent for you guys!!!


Stig