UL Connection results in higher distortion

[TheGimp]

I’ve been working with the single end amp using a 6N1P as input and 6P1P as output (image 1). Input stage has an LED with 100ohm resistor in series (for feedback) for bias. Anode is current source driven. The current source is adjustable for tuning distortion. I’ve been changing feedback in the output stage between UL, GNFB and a combination of the two.

GNFB elicits the response I expect in that it reduces output amplitude, and when the output is re-adjusted to 0.500V I have lower distortion than I had measured without the GNFB.

However, I’ve found that when I hook up the UL tap to the screen, the amplitude drops as expected, but when I re-adjust the output to 0.500Vrms the THD (image 3) has gone up above the level seen without UL connections (image 2)!

I can remove the GNFB and get similar results in that UL decreases amplitude (so it is negative feedback) but it increases distortion over what is seen without it at the same signal level.

The transformer is a Edcor XSE10-8-10K. I checked the signal amplitude on the transformer primary and the screen tap signal amplitude is about 40% of the plate connection signal amplitude, so the transformer taps are wired correctly and it is hooked up correctly. I also tried a second transformer with the same results.

Test frequency is 1KHz.

IM seems to go down as seen in the sidelobes of the 1KHz test signal, but overall THD goes up.

[DF96]

It looks like UL gives slightly higher 2nd-order and slightly lower 3rd-order. This would probably sound a little better.

I think the reason for this is that the UL connection is reducing the gain slightly more than the open-loop 2nd-order distortion, so the overall 2nd goes up slightly. However, some of the 3rd-order distortion with global feedback is coming from recycling the 2nd-order so dropping the raw 2nd reduces the 3rd, in effect, twice over.

[Sch3mat1c]

It's a maybe-not-so-well-known fact that UL and triode modes both have higher distortion than pentode mode with an equal amount of NFB.

The reason: the screen is a nonlinear input with low gain. It also trashes power output because maximum plate current is controlled directly by screen voltage.

If anything, connect the screen to the *opposite* side of the OPT, providing a limited amount of positive feedback. This will greatly enhance the peak current and gain of the tube, allowing more power output for a given supply voltage, while reducing overall distortion because gain is higher. (Note that the screen taps must be fairly low to avoid hysteresis, and the OPT must be constructed well to avoid phase shifts.)

Tim

[DF96]

It is true that the screen is a non-linear input, but it has the advantage that its non-linearity to some extent is the same shape as the control grid's non-linearity.

If you introduce positive feedback in the output stage you have to be careful not to make a power oscillator. If you manage to avoid this, then you may find that the 2nd-order has gone down, the 3rd-order has come up and you will have a rather loud amplifier - which some may prefer. I think you will have traded quality for quantity. Positive feedback within a negative feedback loop can sometimes work, but I would have thought it needs much more loop gain in order to adequately reduce all the recycled higher-order distortion. Some folk will then say that it sounds like a solid-state amp.

[tubelab.com]

If anything, connect the screen to the *opposite* side of the OPT, providing a limited amount of positive feedback......you will have a rather loud amplifier - which some may prefer

I also found a little drawback to this connection. Play it too loud for too long and the screen grid in your output tubes will MELT leading to fireworks and fried parts. Some tubes can deal with this connection, and some can not. The EL84 can not. Don't ask me how I know this!

[Sch3mat1c]

Yes, UL requires tubes that tolerate high screen voltages; inverse UL, even moreso. 6L6 is a good candidate; sweep tubes are not!

Tim

[kenpeter]

What you think of inverse UL up top, Schaded cathode below?

[TheGimp]

Is this particular to Single ended amps vs PP amps?

I was under the impression that UL would reduce both the pentode output impedance and THD. In the PP version it did reduce distortion.

I'm going to go back and take the measurements over again tonight.

I plan to try Schade as well, but wonder how effective it will be feeding back to the plate of the driver with it being driven from a current source. I'm limited as to how much feedback I can go with before I upset the current source.

[SY]

Originally Posted by TheGimp

In the PP version it did reduce distortion.

Remember, in push-pull, you can increase 2nd to decrease odd orders since the second will cancel out.

[20to20]

Is it possible you really have a mismatched transformer to your tube? 10K hanging from a 5K tube? Isn't the 6P1P = to the 6bq5?