A push-pull amp i can live with

[analog_sa]

A few months ago i realised that the PX25 SETs, magical as they are, severely limit my choice of music. Great with jazz trios, chamber music and girls with guitars/pianos/nothing but socks but not quite convincing with anything demanding balls.

Push-pull seemed a possible solution but i wasn't too optimistic. In the past i've had nothing but bad luck with all the classic circuits and never really liked my own QuadII, Leak St20 and Dynaco St70. I also used to have a professionally built Williamson (Partridge iron) and a Klangfilm. Both have been tweaked ad infinum with generally low success.

It was obviously time to try something very different. I heard a SET using Sovtek 2A3 and really liked the sound, ok maybe i had a bit too much wine that night, but for the asking price the Sovteks seemed really good. So i tried them in a fairly standard paraphase-splitter amp and didn't like them one bit. Out went the paraphase and in came a Lundahl LL1660/18mA. I wasn't expecting too much as reportedly it doesn't maintain perfect balance at high frequencies and rings a lot with the secondary unloaded. My ears, otoh, liked it a lot better than the paraphase. It was driven by a 5687 with a 7044 in front. Much too much gain. The 5687 alone just barely sufficed but the sound was much clearer.
Another trip to the parts bin suggested a 6S45PE in place of the 5687.

Now this sounded really good. Deep, solid bass; excellent dynamics and not a hint of any 'tubiness'. A separate PS for the driver stage brought even better subjective dynamics and soundstaging and replacing the cathode bypass with an 'ultrapath cap' cured the slight shrillness of the 6S45P.

It is still 'work in progress' but it sounds really promising. It can play any kind of music and manages an amazing grunt in my Sonus Fabers although i only measured 12W at output. The PS certainly needs some more experimentation - Mercury or damper rectifiers, possibly CCS+VR tubes ala Lynn Olson. Maybe even a PP driver stage.

PS lookinfg at the circuit i realise the grid stopper on the 6S45 is missing. PS is shared between channels.
[IMG][/IMG]

[fdegrove]

Hi,

Nice work...if you can g....

I might give that bias scheme for the DHTs a try in my own 2A3 PP.
Next one a monoblock?
Or at the very least dedicated L/R PS for that extra tight, ultra-stable super-duper stereo image focus....

1M gridleak on the input?
No gremlin stoppers on the output valvies either?
Hmmm....

Cheers,

[jlsem]

...and in came a Lundahl LL1660/18mA. I wasn't expecting too much as reportedly it doesn't maintain perfect balance at high frequencies and rings a lot with the secondary unloaded.

I am curious as to why the Lundahl doesn't maintain balance. I was always under the assumption that their method of using balanced winding halves with core-type construction eliminated that problem, which seems more common with conventional single-bobbin transformers.

Otherwise, that's a good-looking schematic. Did you consider using a phase-splitting input transformer and a pair of 6C45's to drive the output stage through the LL1660?

John

[Kent Smith]

Interesting thread.... I recently converted my little SET amp to Push-pull using a similar circuit (but Hammond iron, and indirectly heated triodes). Tighter bass, more power, and great overall sound.... I'm happy! I suppose using transformers and no coupling caps must help.

[analog_sa]

Next one a monoblock?

Not likely. Unless it gets too heavy for one chasis.

1M gridleak on the input?

he, he. a mistake; it's 100k.

No gremlin stoppers on the output valvies either?

Stoppers after an interstage don't seem common. Could it be the secondaries dc resistance is good enough for that?

I am curious as to why the Lundahl doesn't maintain balance

Not sure why but i distinctly remember measurements above 20kHz showing a lot of imbalance. I think Lundahl introduced the 1660S in order to correct this problem.

Did you consider using a phase-splitting input transformer and a pair of 6C45's to drive the output stage through the LL1660?

Yes, this has always been my intention. I have a 1544 for the input but a pair of 6S45 is unfortunately a non-option - they are just impossible to match. Seeing that i now have enough gain i'd be a bit reluctant to introduce extra stages just for the symmetry

[serengetiplains]

Analog, have you experimented with Broskie's Aikido circuit? Your driver looks to me to be passing power supply noise to your output grids. I realise the PP topology cancels some of that noise, but presumably not all. Just a thought.

How is your impression of class A push-pull some years later now?

[JoshK]

I have heard the 6C45pi to be hard to match, but couldn't you circumnavigate that issue with a CCS in the tails?