Congratulations, you just built a $2K antenna !

[zigzagflux]

Wow, this doesn't seem to be getting any better. Pretty much resolved my 60 Hz hum; was being picked up due to local EMI.

Now, a more difficult problem, RF.

Pictures attached showing schematic and scope traces. With the input shorted, I get the attached input noise as measured at C1-R5 node. This is fundamental 2.5 MHz, with oodles of even and odd harmonics. About 180 mV p-p !!!!!

At the output, I get the attached output noise, cleaned up a little to a 5MHz signal, little bit of odd harmonics. Forgot the level, but its definitely there.

All noise is gone as soon as amp is powered down. Would this be due to power supply, resonance, or what? All rectifiers are HEXFET types with bypass caps.

http://www.just4sheep.com/Input.jpg
http://www.just4sheep.com/Output.jpg
http://www.just4sheep.com/ST70%20BOM.pdf
http://www.just4sheep.com/schematic.pdf

Thanks for any help.

[SY]

If you're seeing the oscillation at the input, it may be grounding-related. Are the bypass caps C9 and C10 local, i.e., returned to the bottom of their respective cathode resistors? If you remove the input tubes, does the oscillation persist? How about if you remove the output tubes?

[zigzagflux]

Pulled the 6922 input tubes (including concertina for safety) and all was silent.

C10 and R17 have their own dedicated star point. C9, R8, R1, and C1 have their own dedicated star point, along with the RCA input shield. These two star points are returned to the power supply common. Output stage has its own star point and bypass cap also. A single common point to chassis is made through a 100 ohm resistor (makes no difference if its shorted).

I will take some time tomorrow to move around some of the wire routing, and maybe take a picture for your viewing pleasure.

Can the 6922's really be that sensitive? I've got some NOS philips low noise, I'll try those tomorrow as well. Currently using new production JJ.

[SY]

6922 are high-gm RF tubes. They will cheerfully oscillate if you give them half a chance. Put a grid stopper on the upper tube, too. Does C3 go to the first star (i.e., C9,R8) ?

[ray_moth]

According to Norman Koren, RF is a more common problem than many people realise. Speaker leads can act as an antenna and the problem can be exacerbated by the shunt capacitor typically used for HF correction in a global NFB loop.

[KimBOlesen]

Hi

Have you tryed without NFB ? High frequency phase shift in a bad transformer will turn into Positive Feed Back....

[zigzagflux]

Originally posted by SY
6922 are high-gm RF tubes. They will cheerfully oscillate if you give them half a chance. Put a grid stopper on the upper tube, too. Does C3 go to the first star (i.e., C9,R8) ?

Are there better tube alternatives, or should I toss the whole cascode idea? Currently using 100 ohm stoppers. Should I increase them all to 1K? And yes, C3 does go to the first star.

Originally posted by KimBOlesen
Have you tryed without NFB ? High frequency phase shift in a bad transformer will turn into Positive Feed Back....

Took out C13 and the noise looked a little different, but was still there.

[KimBOlesen]

Hi

what about R23 ? it is hooked up to 8ohm tap...

[zigzagflux]

Originally posted by KimBOlesen
what about R23 ? it is hooked up to 8ohm tap...

I didn't, and now I'm in the process of rewiring some things, and all the tubes are out. It would surprise me if that made any difference, especially at RF. Might give it a try later on.

Picture of the preamp section is attached. I actually get measureable RF voltage drop across a 3" section of solid core, so there must be some grounding issue, but I paid such close attention. I could maybe save a little extra length in places, but can't see anything obvious. It will also cause me to do some major relocation and part replacement to clean up the star grounding any further, as I am so cramped for space already.

Next time, I go with a separate power supply.

Would this also explain why the amp in general doesn't sound very good on the highs? It is hard to be subjective when you put so much effort into the thing, and what you hear at the end basically sounds washed out. Mids are fine; highs are poor. My Yamaha class D amp produces better highs.

[zigzagflux]

How about this:
Any ideas for a driver stage with three 9 pin sockets? I would rather scrap the cascode and go with something a little more robust against RF. My grounding may not be absolutely perfect, but then again, the original ST70 was much worse, and it didn't have these problems to this extent.

I think the output stage is fine, so I would like to keep that. At this point in my life, with my aged ears and nonsensitive speakers, I will be using PP outputs.

No specific preference of concertina/LTP, and a pentode input doesn't specifically bother me, as long as it does better than this big turd.