|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
60.229.242.73
In Reply to: Re: Build ideas for Allen Wright PP-1C Push Pull Power Amp posted by Burney on April 15, 2007 at 23:56:12:
[RANT] The 12AX7 is a VERY poor choice for ANY purpose WHATSOEVER in a Hi-Fi amp. [/RANT]
There, that feels better! :)In this particular instance, the input stage comprises two cascodes, connected as a long-tail pair splitter. Cascodes work best with high gm triodes and the 12AX7 does not meet that criterion. Not many tubes will do a good job here, for that matter, because you also need a tube that will operate well at quite a low plate-cathode voltage. The 6DJ8 (family) is one of the few that are suitable (not a coincidence - it was designed for RF cascode operation).
A pentode might do very well (I'd love to try it!), since a cascode can be considered a way of making a pentode out of two triodes. The cascode doesn't suffer from partition noise like a pentode does, but it poses the problem of having to share B+ between top and bottom tubes and usually doesn't have quite as much gain as a pentode.
One problem with using pentodes could be conflict between the current the pentodes want to draw, under influence of their screen grids, and the current for which the CCS is set up. A way around this could be to experiment first with a shared cathode resistor, find out what cathode current you get, then set up a CCS drawing the same current to replace the resistor.
Obviously a pentode solution makes some things easier, like not having to share B+ between two tubes in series, and not having to worry about heater-cathode voltage limits for the upper tube of a cascode. Allen's design and choice of tubes solves both of these problems comprehensively.
A big cathode bypass cap is something Allen found to work well. Personally, I'd prefer to avoid it by using fixed bias, as you suggest.
Follow Ups:
Doesn't Allen use the Russian 6H30 variants for a gruntier driver?And doesn't he use a CSS in the tails of the output pair to avoid those huge bypass caps? And some form of fixed bias? Combination bias maybe?
When copying someone's design, it's best to start at their design before wandering off elsewhere otherwise how will you know how close you are to the excellence of the original? i.e. where the designer's started himself. Allen has gone off elsewhere since I think :-)
When I built this amp I found that NOS 6922's were much better than current production. IIRC, that advice came from Allen, too. I use Phillips and they sound great. I was easily able to drive 300B's.
"Doesn't Allen use the Russian 6H30 variants for a gruntier driver?"
He might do now, I don't know."And doesn't he use a CSS in the tails of the output pair to avoid those huge bypass caps? And some form of fixed bias? Combination bias maybe?"
I believe he did introduce a CCS in the OP stage but I've never seen that version. I was thinking of the original PP-1C.
"When copying someone's design, it's best to start at their design before wandering off elsewhere otherwise how will you know how close you are to the excellence of the original? i.e. where the designer's started himself."
Hey Ray, I wan't trying to squash yer ideas at all. Getting carried away is what this forum is all about.
I know that. :) What do you think of using pentodes, though? Maybe something like 6AU6. As with cascodes, they would have the disadvantage of high OP impedance, meaning that they might find it hard to drive low input impedance tubes like 300Bs or beam tetrodes with fixed bias. Should be OK for driving pentodes, though, especially with cathode bias.
On second thoughts, I doubt if a pentode LTP splitter would necessarily benefit from a CCS in the tail (despite the fact that one of Gary Pimm's variations on his 47 PP amp used it with a pair of 6AU6s). The problem, as I see it, is that both plate and screen currents go through the cathode, so having a constant total cathode current does not guarantee that the sum of plate currents will always be the same, because the screens could also vary.I suppose this problem might be overcome if the screens were fed from independedn matched CCSs. This should take the screen currents out of the equation. I was going to check if Gary Pimm did this but, of course, his site is no longer available, so I can't. It does seem a heck of a lot of trouble to go to, though, just to use pentodes in an LTP!
This design started out back in the late 70's with efforts to make the old Quad IIs sound better. many many things were tried and one big sonic step forward was to replace the EF86 pentodes with cascode ECC88/6922's with no other changes to the Quad.Since then I've stayed with the cascode as it's fantastically linear and fast!
I'm OK with pentodes at small signal levels loaded down heavily but don't know much about them as drivers though I can see what you're getting at. I tend to be favouring low(er) Z circuits in my designs. Never really got into cascodes as they didn't seem to be to my liking when I tried them. Maybe it was the valves you have to use with that config.Gary Pimm's 47PP amp is something like you mention but he was using some type of feedback IIRC. Sad he took his site down. I liked visiting it as his ideas were sound and well thought out.
Thanks so much.
Another issue that has me worry is type of input. I dont have balanced preamp. So if I short the (-) IN to ground then the bottom cascode looks like grounded grid. What happens then to the balanced stage? Is the gain is reduced for this stage?
Yep, it sure is. You get about half the gain from an LTP splitter than you would get from a differential stage with balanced input.What guarantees good signal balance is the CCS in the tail. It forces the total current of the two halves to be the same at all times. So, if the current through the first half goes down by, say, 0.5 mA then the current in the second half must go up by 0.5 mA. If the plate load resistors are equal then the voltage signals at the plates must also be equal (but 180 degrees out of phase). Note that the CCS doesn't guarantee good DC balance, just good AC balance. For that reason, it's quite common to see two equal resistors between the cathodes and the tail - usually a few hundred ohms - to even out the quiescent currents.
Thanks all so much for the guidance.
Can I have a clear answer if it is ok short the (-) In to ground and putting a 100K log pot in the (+) input is a great idea? Can someone please clarfy?
Is 50k to high?
This post is made possible by the generous support of people like you and our sponsors: