Sansui AU70

[mr2racer]

Hey Everyone, hope your Sunday is going as nicely as mine!

I have a couple of questions about the Sansui AU70 output transformer. I've noticed they connect the cathodes to the opposite ends of the output transformer secondary. I'm assuming they are using the resistance of the windings to act as a cathode bias resistor? While at the same time using the connection for feedback? But somehow having DC in the secondary seems like a bad idea? Also, the 410 volt B+ seems really high even if you had good vintage 7189's.

So the real question is this, if I went with a standard cathode resistor and a more standard feedback arrangement would the resulting amp suffer sonically? All other things being equal of course.

Kevin

 

[JonSnell Electronic]

Half of the circuit is missing here. As I recall from memory. there is -ve DC bias on the grids from a supply line and the cathodes use the windings in the cathodes as positive feedback and current mirror. The greater the load on the transformer the more +ve feedback to improve the loudspeaker imperfections.

 

[Eli Duttman]

Look at the whole schematic. A bias supply is present. The O/P "iron" secondary is grounded via a tap, not an end. It appears that Sansui was using those secondaries for local cathode negative feedback, not bias. It's most unlikely that the engineers did not account for flux balance in the core.

Leave the circuit topology be, except for rewiring the O/P tube sockets to 7189 specifications. Stock of 7189A tubes is NOT available, at anything resembling a rational price. Properly matched pairs of 7189As are wishful thinking.

Install Russian 6П14П-EB (6p14p-ev), AKA EL84M, O/P tubes and you'll be fine. The Russian "firebottles" are genuine 7189 equivalents, which are VERY tough and good sounding too boot. To maximize O/P tube service life, set the "idle" current a tad lower than Sansui's doc. suggests.

 

[mr2racer]

Thanks guys, I see the bias connection now. If you look at the partial schematic in my jpeg there are two connections; C3 and C4 for individual bias. The schematic that Eli posted shows the outputs biased as a matched pair.