Plate vs OT distortion

[strelok]

When you say voltage spike are you referring to a sudden increase in amplitude when you dip the frequency down to 80hz or so? If that's the case you're probably just hitting a resonant peak of the OT. Most OT's have a frequency response curve that looks like a band-pass filter with a hump just before the stop bands that's known as resonance. See this image below:

[img:612:359]http://audio.thedelimagazine.com/wp-con ... 12x359.jpg[/img]

Except that a band-pass filter is mirrored and would have a similar roll off on the low frequency end. Sorry was the best image I could find.

In any case, while I'm definitely not opposed to removing the bass control from the circuit, don't be surprised if the issue you're describing doesn't go away once you do.

This among other reasons is a difference between Hi-fi and guitar amp OT's. Most Hifi OT's are rated full bandwidth (20-20khz) while guitar is generally about 150-10khz. The logic on the latter being that the lowest note on the guitar is about 75hz or so (can't remember the exact frequency) which puts it at around the 3db down point on the response curve, so you don't really need a lot below that point. On the other end anything with a strong response above 5-10khz for guitar will make it sound rather harsh and brittle.

When it comes to heavy overdrive that's part of what makes the OT in a guitar amp such a wonderful thing. It smooths out and removes a lot of the higher ordered harsh sounding harmonics and leaves you with the good stuff. The speaker does this even further, with most guitar speakers being rated for not much more than 5khz. If you've ever heard an amp recorded into a dummy load and direct into the computer without any speaker emulation you'll know what I'm talking about. Its sounds absolutely terrible.

[pdf64]

I don't think that an OT in a regular tube power stage will tend to resonate at the low end in the same way that may occur at the high end.
For it to happen, I think there may need to be a cap in series with the OT to create such a resonant circuit.

[strelok]

I certainly could be wrong. I've seen it happen on a few amps on the bench but maybe it was something specific to the circuit and/or due to those very specific OT's, for some reason I thought it was a general characteristic.

Now I'm gonna have to throw some amps on the scope and see what's up. Last thing we need is someone spreading misconceptions based on a couple casual observations.

[pdf64]

Thanks, I'll do likewise
Low end resonance is usually most likely due to speaker mechanical resonance, so to remove that effect, a resistive load should be used.

I think that the high frequency self resonance may be due to the parasitic capacitive and inductive characteristics of an OT.

[JoeTele]

FYI the voltage spike I was referring to was indeed an amplitude increase as I went from 1khz down. I can't remember exactly where it started but it was well above 80hz, perhaps a couple to a few hundred hz, and was somewhat gradual ("spike" was the wrong word) but steady down to 80 and below. I'll check again. I started with an EH EL84 for dependability and, not expecting to "correct" anything, bought a Sovtek EL84M after reading that a lot of people like them in hi fi, and was blown away by the difference. Among other things, the bass seems a bit crisper. So I'm going the throw the unit back on the scope (as soon as I can stop listening to it) and see if I'm just wishful hearing or if I can "see" any of the difference.

[strelok]

Let us know what you find if you get a chance to put it back on there. That being said, from the sounds of how happy you are with it, I'd leave it alone and not worry much about what the bench tells you outside of curiosity/analysis.
Decades have been spent trying to build amplifiers that have the perfect performance on the bench and here we are discussing something that by all rights should have been obsolete 50 years ago. Yet here we are discussing it. So take what you read on the bench with a grain of salt before you try to modify or fix certain things about the frequency response.


In regards to OT resonance discussion, it occurred to me that the power supply capacitance (in conjunction with the OT inductance) could be responsible for the low end "hump" in the frequency response curve. I suppose an easy way to test this would be test different mains filtering values and see how it affects it. Though I'm not sure that would totally rule out other factors.

[teemuk]

Yes, insufficient "decoupling" - largely in the power supply circuit - can sometimes lead to resonance as signal creeps back to earlier stages through supply rails as positive feedback. This is usually an issue only when no decoupling at all is employed in critical supply nodes.

In other words, if you already employ a filter capacitor then you most likely decouple that particular node more than sufficiently (e.g. the source impedance of the gain stage is much higher than that of the power supply so capacitance to decouple is more than ample). Modifying the power supply to feature somewhat higher filtering capacitance (e.g. from 100uF to 220uF, or multiply by then in case of SS circuits) can cure issues like excessive hum or sag but I don't think it resolves any occuring decoupling issues as it already does it fair share in that.

I would be looking around things like no supply rail decoupling between PI and preamp, or no supply rail decoupling between three or more series gain stages in the preamp, particularly those that operate "in-phase". If these seem to be dealt with then I would suspect that cause for low end resonance is elsewhere than in insufficient supply decoupling.

[JoeTele]

Ok I finally got back to the bench. I started taking measurements, writing things down, etc., one thing lead to another, and I realize I may need a lesson signal measurement fundamentals. If I set my signal generator to 1000hz, 21mv rms, and adjust my amp's volume so that the dummy load is seeing 1.5 watts, there is a steady +8mv in = +.5 watts out for the next couple of watts, which is all I measured. If I leave the generator's amplitude alone, but boost the frequency, the amplitude will decrease on the generator as a result. However, if I boost it to where the signal from the generator decreases by the 8mv mentioned above (in this case to about 2650hz), the output has only decreased by .06-.1 watts rather than .5 watts. Is this "correct"? At first, I was worried that my measurements were being impacted by the generator's amplitude changing with it's frequency. But when I tried manually maintaining a constant amplitude while boosting the frequency, my output went through the roof as the frequency increased. Could someone explain?

Thanks!

Joe

[JoeTele]

Never mind. Think I was thinking too much.

Joe

[JoeTele]

OK, so I've gone back over this carefully and learned the following. Part, though not all, of my issue was a low reading meter. I found a good deal on one on ebay with an Indiana Standards Lab certification sticker good until April of next year, so hopefully I've resolved that. I did the same thing as before, started at 1000hz/4.9 volts (3watts w/8ohm load) and worked my way down to 100 with two tube configurations. The original, an EH EL84 and Shuguang EF86; and the Sovtek EL84M paired with and NOS France-made Amperex EF86 which I purchased after hearing how much the Sovtek did for the amp. Both increased the voltage on the way down, but more or less steadily. The EH landed at about 5.1 volts and the Sovtek about 5.3. Both had a slight kink in the wave. However, when each was then adjusted down to 4.9 volts at the 100hz (the bottom end of the OT rating), the Sovtek/Amperex cleaned up much more. Pics of the traces under these conditions are attached. So under "lab" conditions, the Sovtek does show cleaner bass at the 3 watt output. However, given the inherent boost at those frequencies, the final product is the same in the real world, sort of. Now that I've done measured power/loudness tests, I would be surprised my typical listening power is even at a watt. So whether the o-scope confirms my suspicion that the bass is better/tighter, I don't know, but I'm sticking to the story anyway! Perhaps it's the extra voltage at the low frequencies? Oh well, + 1 for not allowing the listening pleasure to be undermined by what's on the scope. And BTW, that Amperex also added a little.

Joe