Hoping to get some advice/proofing on output stage design and layout.

[SoulFetish]

I'm hoping to get a critique of this design to see what improvements I can make, or catch any errors I may have made.
Nothing crazy. It's a LPT buffered by 12AU7 cathode followers, which DC coupled to the output tube grids. But, ever since I powered this thing on, I've been running into noise issues. The worst of it is the parasitic oscillation I'm getting in the output stage. The preamp seems to be dead quiet, as far as I can tell. I've been reading some old texts on DC coupled amplifiers and I came across a paragraph in RDH4 which seems to nail the symptoms of the distortion and parasitics I'm experiencing. I can barely drive the amp when the notes just fizz and buzz out in the speaker. Both with and without the feedback engaged. Two methods that L. Smith mention is to connect capacitors from grid to earth or Plate to earth. I would reason that I should connect them from plate to earth so as to not place to much of a capacitive load on cathode follower drivers. As always, your analysis and insight is greatly appreciated.

The design is pretty simple, I've attached a schematic of the output stage, power supply, and a layout (which, hopefully is fairly easy to follow).








edit: bias voltages added to schematic

[Enzo]

Before you start throwing solutions at it - caps to someplace - you ought to identify the problem.

What bias is on the power tubes? ie what is the cathode voltage and what is on the grids? What is on the cathodes of the drivers?

[SoulFetish]

Before you start throwing solutions at it - caps to someplace - you ought to identify the problem.

What bias is on the power tubes? ie what is the cathode voltage and what is on the grids? What is on the cathodes of the drivers?

Ah, great question! I should have included that in the schematic. The driver cathodes at the driving junction sits at -4V as do the output tube grids, while the EL84 cathodes at +7.2V (they're biased at 30mA-ish, but easily adjusted).

[SoulFetish]

I've updated the schematic with voltages at the grids and cathodes of the output tubes.

[catalin gramada]

Hi.
From you layout, your feedback return should be tied to inverter stage cap. Normaly I use shielded wire for this but tight twisted pair works aso. The order is: common OT to output jack and further from this jack to feedback return directly from the stage was aplied.
Also ,passing the FB over three stages can generate some oscilations troubles (normal you last interstage DC coupling should help to avoid that).You will need a scope to determine if is the case.
I think you exagerate with 1K screen resistors. Is nothing wrong but my first choice will be as small I can use it. (220 ohm should be enough).
And a personal choice: 12ax7 instead 12au7 into buffer stage (You can use 470k,or even less here as grid leaks instead 1M in this stage without any problems for stability consideration).

Cheers

[SoulFetish]

I, not having much of an option, I decided to try and fire up my Tek 556 and actually see if it worked.
Results:
Total failure - The scope powered on, lighting up the jewel indicator and fan, but nothing at all on the screen. Oh, and a good spark from inside, which prompted me to pull the power. I really had my fingers crossed that this thing would work so I could get to seeing what's going on. So I'm anxiously waiting on Craigslist for a decent scope throughout my entire driving radius.

Hi.
From you layout, your feedback return should be tied to inverter stage cap. Normaly I use shielded wire for this but tight twisted pair works aso.
I think you exagerate with 1K screen resistors. Is nothing wrong but my first choice will be as small I can use it. (220 ohm should be enough).
And a personal choice: 12ax7 instead 12au7 into buffer stage (You can use 470k,or even less here as grid leaks instead 1M in this stage without any problems for stability consideration).

Cheers

Among some other changes, I switched up to shielded wire on my feedback signal run.
As far as the screen resistors, this amp lives in overdrive and I like the compression and dynamics of modulating the screen voltage at full drive.
You like the 12AX7 as a driver? I'd be curious to hear it. The original driver was designed to be operated at +150V/-150V, but my secondaries ended up being low by about 50V. Interestingly, judging by the curves, I really like the look of a 12AV7 under this operation. It looks to be operating very linearly, with maximum headroom. This is an easy swap too, because it works well under that same load as the 12AU7 and would take an easy quick adjustment to it's bias.
Check out the load lines of a 12AX7, 12AU7, and 12AV7.