Plate vs OT distortion

[JoeTele]

I was studying the frequency response of my recent hifi (3-3) build, and I noticed some distortion in the bass frequencies, 300hz and below when the amp is at it's maximum power rating. Audibly speaking, I'm not that concerned with it, and initially figured it was due to the budget, general purpose transformer I used. But I started probing out of curiosity and found that the distortion is on power tube plate as well as the output. Can distortion that starts at the OT be seen at the plate under any circumstances? I'll admit that to save a shipping bill, I used a 22uf bypass cap around the cathode resistor (single ended amp) instead of the 25uf cap that was called for, and wonder if that could be a factor. I'm most interested in the plate/transformer question though.

Thanks as always!

Joe

[Phil_S]

Sorry, the discussion of distortion is over my head. The difference between 22u and 25u for the bypass cap is insignificant. The tolerance rating is probably +/-20%.

[NickC]

Distortion is what makes tube hifi gear sound so good.

[R.G.]

Check and see whether the distortion goes down with increasing frequency. If it does, it's mostly OT distortion. If it doesn't, it is likely to be not the OT.

For tubes, the audio band is almost entirely "DC", while transformer defects get worse at very low and very high frequencies.

[Stevem]

What is your plate voltage at idle and what does it drop off to at 300 HZ when output stage clipping shows up?
Also what is your RMS power and peak power?

[JoeTele]

Yes, it's frequency dependent. By 500 or so hertz it's clean as a whistle. I'll check the voltages. RMS power is 3 watts. At 1khz you can push it a little past that before visible clipping on the scope but it's advertised as a 3-watt amp so I'll judge it at that power. What's interesting is that even with the NFB bass boost set to flat/"off" there is a natural spike in the low frequencies when you back off from 1khz/3 watts.

Bottom line, though, is that distortion introduced by the OT can in fact be seen on the plate also, at least in some circumstances?

Joe

[pdf64]

Is the budget, general purpose OT suitable for single ended applications?

[JoeTele]

Yep. It's a Hammond 125DSE single ended series rated for 100hz to 15,000HZ at 10w. I'm running it at 3 tops, so it does a bit better.

[trobbins]

Have you disconnected the feedback, and used a resistive load for your assessment? That would simplify the assessment.

Does the OT have a maximum/nominal DC bias current spec, and what is your idle cathode DC current?

How are you measuring distortion level?

[JoeTele]

I've got an 8-ohm 1% 120 watt power resistor I use for measurements. Max DC bias for the OT is 70 ma and I'm running around 53ma. Could disconnecting the negative feedback introduce distortion?

Joe

[trobbins]

Could disconnecting the negative feedback introduce distortion?
Joe

It would add a layer of complexity to any assessment.

One path you could take is to present a spectrum at a few signal frequencies at a few power levels (under any rated/clipping level), and then maybe increase the idle DC bias (although that may impact first stage) and repeat another set of test results.

[strelok]

Bottom line, though, is that distortion introduced by the OT can in fact be seen on the plate also, at least in some circumstances?

Joe

Disclaimer: 5am post

I'm going to throw out a resounding yes to this. I refer to it as back propagating of distortion. Its quite similar to what happens when the grid of a tube is pushed very close to or greater than zero volts. The input impedance with stage suddenly drops and the previous stage can no longer drive such a load of that low of impedance and starts to distort.

However with the OT its slightly different. Inductors have a ever decreasing impedance as frequency goes down. So as the frequency decreases so does the input impedance to the OT, this loads down the output stage and since its no longer presented with as high of a load as it wants to see the signal on the plate will start to decrease in amplitude and eventually start to distort as you sweep the frequency lower. This is a VERY simplified explaination of what's happening but should give you a general idea of why you see it on the plate as well as the output.

As to the NFB I can almost guarantee that its reducing the distortion. At least in the sense of what you'd see for a single frequency on the scope.

Here's and easy way to test it. While the amp is off desolder the NFB from the speaker jack. Clip an insulated test lead (shorter the better) from the desoldered wire to the lug it was connected to. Fire up the amp into a dummy load and run the signal generator to the input. Pick a frequency, any frequency, 1khz will do. Clip a scope across the dummy load. Turn the input singal up enough to where you start to see visible distortion on the scope, then back it off just enough to see it clean up. Now disconnect the test lead. You will most likely see the signal level increase and distort.

Just be careful with the test leads while the amp is on, and bear in mind that the longer wire might cause it to oscillate.

[pdf64]

... Fire up the amp into a dummy load and run the signal generator to the input. Pick a frequency, any frequency, 1khz will do. Clip a scope across the dummy load. Turn the input singal up enough to where you start to see visible distortion on the scope, then back it off just enough to see it clean up. Now disconnect the test lead. You will most likely see the signal level increase and distort...

I think that would most likely happen purely due to the natural increase in system gain resulting from open loop operation?

Regarding the OP, I think that as the plate of a power pentode is a high impedance source, distortion generated in the OT may appear in some degree at both the primary and secondary.
It's a different situation to feeding the primary from a low impedance source.

[strelok]

I think that would most likely happen purely due to the natural increase in system gain resulting from open loop operation?

That's correct, however if you increase the input level with the NFB installed to make up for the lost gain; you should see *some* increase in clean headroom before clipping. In theory anyway. The biggest difference however is in the non-clipping region of operation where the NFB is helping to compensate for some of the non-linear response of the tube. However you probably won't see it on a scope. You'd need a distortion or spectrum analyzer to most likely. If you look at it on a spectrum analyzer you'll see the lower ordered harmonics decrease with an increase in NFB however the higher ordered once will go up ever so slightly. Which will make the amp sound brighter (harsh on the extreme end of things) due to the way our ears perceive harmonic content.

[JoeTele]

Thanks for the great info! Knowing that the presence of the distortion on the plate doesn't disqualify the "economical" OT from being the culprit, I think I'll avoid messing too much with the NFB, as doing so could mess with the stability of the amp. However, I may take the bass boost circuitry out of the NFB just to see if that impacts the voltage spike that happens in the low frequencies when I drive it to it's clean limit at 1khz and then roll it back to 80hz or so, and maybe see if I can flatten it out by tweaking the value of the cathode bypass cap of the EL84.

Thanks again!

Joe