1.5W top boost-style amp

[oxbow_lake]

They're jumpered on the sockets.

[bepone]

pls oscilograms on anode1 and anode 2 (output trafo connected normally with speaker or resistor) when signal on the output grids (g1) is 2 Vpp

[bepone]

suspecting shorted secondary somewhere (maybe on output jack) which is transferring short circuit to anodes

[oxbow_lake]

Ok. The first image is the signal at each grid. The second image is the anode voltage (AC coupled to get rid of the 280VDC) at each plate. Looks pretty nasty.

The purple text is hard to read in the photo, but the voltage scale is 1V/div for both the grids image and 5V/div for both channels in the anode image.

FWIW the traces look the same if I swap out for a different 6BN11.

[chumchai]

Your amp circuit is similar to this Nakid Micro Amp but with an unconventional power supply.

https://robrobinette.com/How_the_Saldan ... 80_Pentode

Personally, I would not use a dual pentode in one bulb. These are not matched and when one goes bad then both need be be tossed. 6BN11 is a dual 6EW6 RF pentode. To fix your issues I would try the following:
Install grid stopper resistors 1.5k-2.2k to the output tubes. Oscillograms show some oscillation on one side.
Install single cathode resistors which will help balance the tubes.
Screen grids need to be bypassed with capacitors to GND, so get rid of R12, R18 screen grid resistors, not needed, as you already have a common screen R35 resistor, bypassed (correctly) by C21 capacitor. Screen grid voltage is kinda low, could be increased for power increase.
If you have an audio generator, I would feed 1kHz signal directly to the PA tube grids to eliminate any pre-amp noise (remove PI tube) and go from there.
As a side note, a 6AB8 tube would make a nice Micro Amplifier. Just saying.

[oxbow_lake]

Hello and thanks for the advice! My output circuit is indeed similar to Rob's EF80 push-pull design, because I largely tried to steal it. I have the screen voltage low to avoid exceeding the maximum screen dissipation from the datasheet, but I think I should still be getting 1.5W or more out of the tubes.

You're right that the dual pentode is not necessarily a sensible approach. However, even if it's not sensible, it should work, shouldn't it? I implemented your changes - removed the screen grids, installed 2k grid stoppers, and separated the cathode resistors. Both cathodes now draw 14.5mA exactly. The oscillation out of the noisy pentode is significantly reduced. However, the output volume is unchanged.

Really appreciate the help!

[chumchai]

Since you have access to an oscilloscope, use an 8R load resistor and measure the signal there. Calculate the RMS output power, then you know for sure.
In the meantime, jump temporarily (short) the D1 Zener diode and see if that makes any difference.

[chumchai]

14.5 mA each tube? That is class A bias, 3.9W. Suggest to change to class AB bias (50-70% class A) that would be about 7.25 - 10.15 mA each tube, to keep it within limits.

[oxbow_lake]

I increased Rk to 150R for each pentode. Idle current in the cathodes is now 10.5mA, which, given that 2-3 of that is screen current puts the idle plate current around 9mA, which is less heavily class A biased at least, and puts quiescent plate dissipation at ~2.5W, which seems safe. Output volume is similar. Scoping the peak-to-peak output voltage at maximum signal without clipping over an 8R dummy resistor yields 300mV, so peak output power of 10mW, or RMS 3 mW??

[martin manning]

The operating point looks fine, so the output tube seems to be good. You can't get signal through the OT... Any chance it is damaged? I still wonder why you can't measure something close to the expected impedance ratio.

[chumchai]

Forrest Cook built a single ended mini amp with a similar pentode and got only 100mW out of it. I don't have any experience with these tiny TV pentodes.
http://www.solorb.com/elect/musiccirc/6U8AmpV3/

Point of concern is your dirty & clipping input signal. Suggest removing all pre-amp tubes and concentrating on the PA stage.
Short the PA pentode inputs to GND( Depth pot wiper to ground) and scope the output load. Must have a clean a strait line.
Anything other than that, must come from your power supply, PA oscillations, ground loop hum or oscillation.

I am running out of ideas.

[oxbow_lake]

Well, 100mW would be no fun.

I think I have some useful results here. First, an image of the baseline noise - no preamp tubes installed, grids grounded. The yellow trace is at the grid, and the purple trace is at the primary. So, a little noisy, but maybe not horrible. As you will see in the next oscillograms, the grid clipping has gone away since installing grid stoppers.

I disconnected the OT and connected 15k resistors from each pentode plate to the B+, and measured the plate voltage with an test signal. Basically trying to make sure that each pentode works just as a class A, and removing the OT from the picture. Now I get the voltage swings I expect. The middle image is looking at one plate (yellow trace) with a 10V test signal being fed directly to the grid (just prior to the grid stopper). This looks kind of reasonable! At least, it looks like voltage swing is happening and power is being dissipated in the load resistor. The last image is with the phase inverter tube installed, feeding the test signal into that, and then measuring the plate voltage again. This time I increased the input signal to the point where the pentode output started to clip, and here we see full voltage swing and plate dissipation of almost 4 watts!

All of which is making me think that the design is maybe okay, and the OT is in fact damaged. Thoughts? Is it time to pull the trigger and buy a new OT?

[wpaulvogel]

If it was me, I’d simply the power supply to the power tube. The B+ and B+1. Tie the red CT to the output of the rectifier moving the reservoir caps to the other side of the 820R resistor. Tie the screens after the 820R with filtering cap and ditch the Zener diode. Try it with the output transformer and forget about the load line figuring right now. You’ll be safe with 22.5k P-P load. I believe with all the built in resistance and low screen voltage your idle conditions might match your load line prediction but dynamic performance is sagging your voltages so far that the amp is run out of steam. My whole point here is similar to the Marshall amps that swapped 6550’s for the original EL34’s. 6550’s require a completely different load resistance, screen supply and bias supply IF you follow the data sheet and anode characteristics chart but in reality they need a little more negative voltage to the control grid to operate within dissipation limits and the excessive screen voltage as per the data vs in the amp has no real disadvantages. I think the main problem is that you have constricted the power section to the point it can’t breathe.

[martin manning]

I increased Rk to 150R for each pentode. Idle current in the cathodes is now 10.5mA, which, given that 2-3 of that is screen current puts the idle plate current around 9mA, which is less heavily class A biased at least, and puts quiescent plate dissipation at ~2.5W, which seems safe.

All of which is making me think that the design is maybe okay, and the OT is in fact damaged. Thoughts? Is it time to pull the trigger and buy a new OT?

I can confirm all of the above re the load line and operating point with the OT connected, so it appears that the output pentodes are responding to voltages per the data sheet. Load line power with a 22k5 load is 2.9W. All that seems to point to a defective OT.

[chumchai]

I could manage to get 2.25W out of this amp in LTspice simulation, with some circuit tweaking.
Input signal to phase inverter 1V p-p.

Pentode grid sine clipping at 1.8V:

1.5Wsim_grids.png

AC voltage swing at the plates:

1.5Wsim2.png

Output sine on the 8R load resistor:

1.5Wsim.png

RMS Power calculation: V^2 / 2Rload - 6^2 /16 = 2.25W rms
So, it can be done, theoretically. As I see it the output pentodes are problematic, with a transconductance of 13 mA/V or 13000 umhos, 200k plate resistance, 2 Volts p-p drives them fully open. They would be great as 1st stage voltage amplifiers.