5E3 X-Y display of output

[loudthud]

This experiment was setup as follows. The output of a vintage 5E3 was connected to a 0.1 ohm current source resistor and a two way footswitch so the current and voltage of the load could be displayed on an oscilloscope. The footswitch selects the 5E3s speaker or an 8 ohm dummy load.

The Horizontal is connected to monitor the voltage at 10V per division and the vertical measures the voltage across the 0.1 ohm current sense resistor for a deflection factor of 1 amp per division. The first part of the video clip is the 8 ohm load resistor, the second part is with the speaker connected. You can see that the reactance of the speaker allows the amp to supply more voltage to the speaker than the resistor. Voltage into the resistor is about +/- 16V at roughly 2 amps. That's 32W peak or 16W rms. With the speaker some voltage spikes approach +/- 50V but the current is limited to about +/- 2amps peak.

Well, the video wouldn't upload so the stills will have to do.

[Chuck H]

That makes some sense since the actual speaker impedance can be VERY high at certain frequencies. The only thing the amp knows to do with a high impedance is to supply more volts even if it means less current. You can hear the effect as well if you've ever played through a resistive attenuator.

[J M Fahey]

Impressive

*Very* useful experiment which puts a lot of myths in due perspective.
In this case the ability of tube amps to effortlessly drive very difficult real world loads (such as guitar speakers) compared to SS ones.
Pity I canīt *double* thank, because this deserves it.
Try to re-upload the video.
If not possible, post it in You Tube and link to it.
We need a , say, 40Hz to 6KHz sweep showing the phase and amplitude across, say, a single 12" speaker in an open backed cabinet.
Thanks again.

[Steve Conner]

Great stuff!

Can you try it again with the "flyback diodes" that are typically installed from plates to ground on new production tube amps?

[Chuck H]

We need a , say, 40Hz to 6KHz sweep showing the phase and amplitude across, say, a single 12" speaker in an open backed cabinet.
Thanks again.

Not to put any pressure on... But +1. I'd love to see a sweep.

[Chuck H]

Great stuff!

Can you try it again with the "flyback diodes" that are typically installed from plates to ground on new production tube amps?

Another +1

Sorry LT... Nothing like being the guy who gets stuck holding the door

[Chuck H]

OOOh, also... What about the same test on amp with NFB??? And one with switchable secondaries so you could test different primary impedances??? Ah. I just moved and I don't have my new bench space set up yet. And I'm still getting to other things on my "short list".

[g1]

the ability of tube amps to effortlessly drive very difficult real world loads (such as guitar speakers) compared to SS ones.

I was thinking it would be nice to see this done with a SS amp for comparison. However, I don't think it would be possible to say that any differences were completely due to tubes vs. SS, there would be too many variables due to all the other differences in the amps.

[J M Fahey]

True.
One *big* difference is that the typical reflected load in a tube amp is placed on the very high impedance plate of a pentode, which to boot has gain in the order of 10X; while the speaker load in an SS one is placed in the unity gain, very low impedance emitter.
That alone accounts for *a lot* of difference; compared to this the *method* used to pass a controlled current of electrons through the load becomes relatively unimportant.
In fact, OTL tube designs, with an equivalent architecture, start sounding similar to SS ones.

[Steve Conner]

Well, current feedback has been used to fix that since the Polytone amps. And the really old transformer driven ones actually ran the output transistors as gain stages, and had naturally high output impedance. (They look like a modern totem pole output, but the crucial difference is that the base drive windings are floating.)

But what they can't do is swing the output beyond the supply rails for large signals. That's analogous to the plates swinging below ground in a tube output stage, hence why I'd like Loudthud to try the flyback diodes.

If he feels that it sounds better without the diodes, well, I think I can build a transistor output stage that swings outside the rails too.

[J M Fahey]

It can be done.
I already have.
Perceived effect is slightly cleaner attack, somewhat more definite sound.
Nothing to write home about, but noticeable.
As a reference: exactly the same that you lose when you add flyback diodes to a tube amp.
SS amps have them built-in.

EDIT: and SS current feedback does not provide *that* sound, because it works when itīs not needed and fails miserably when it is.
Murphyīs law at its best.

[loudthud]

I wish I had the time and energy to perform all the requests. By next week I should have a real video camera to make some more shots if I can figure out how to use it LOL. My current camera only makes 5 second video files. I did do the experiment on an solid state power amp I built recently. It's a class A design, only about 6 watts. Fairly similar to the 5E3 plots but you can see the current limit more clearly. Sort of a tilted box look. I don't have any chip amps to test at this time. I do have a 5F6A re-issue that might be next.

The scrambled egg look is a power cord. Individual notes look more like a curly elipse. Guitar was a 70's SG. Volume and tone were set about 12:00. First tube is a 12AX7.

It's interesting that some of the small Marshall amps like the 5205 have no current limiting and use 10 amp darlington output transistors.

[Steve Conner]

JM, do you mean that the current feedback only works when the amp isn't clipping? Which is completely true, as soon as the output hits the rail it becomes low-impedance again. But a tube output stage never hits the rail. The plate can sail merrily past and go negative, driven by stored inductive energy in the voice coil and kinetic energy in the woofer cone, and maybe a bit in the OT's magnetising inductance too. I think this is why we see the big lobes to left and right in Loudthud's oscillogram.

So, I think with the flyback diodes these lobes would be squared off and the speaker cone braked somewhat.

JM, did you find that your "outside the rails" SS output stage was any louder than the same output stage restricted to the rails?

[J M Fahey]

JM, do you mean that the current feedback only works when the amp isn't clipping?

Exactly.
In a simplified example, "a Twin" is relatively low impedance when clean, because feedback is working, and becomes *very* high impedance when clipping.
Current feedback SS Power amps do *exactly* the opposite, as you noticed.
Iīm opening a can of worms here, but thatīs why "a Twin" is sweet when clean, but bites when overdriven; while the typical current feedback SS amp used *everywhere* is harsh when clean and muddy overdriven.
The good part is that if you know what you are trying to accomplish, this and many other things can be emulated.
My pet theory is that anything can be accomplished with judicious use of Op Amps.

Which is completely true, as soon as the output hits the rail it becomes low-impedance again.

But a tube output stage never hits the rail. The plate can sail merrily past and go negative, driven by stored inductive energy in the voice coil and kinetic energy in the woofer cone, and maybe a bit in the OT's magnetising inductance too. I think this is why we see the big lobes to left and right in Loudthud's oscillogram.

Well, more or less so.
The way I see it, the plate does hit the rail (ground) when saturated, but when reverse biased (because the inductive current in the speaker continues pulling it "down", into negative voltages) the tube becomes instantly disconnected, open circuit. so that negative peak continues freely.
The exact complementary waveform appears on the opposite plate, because of the see saw action of the transformer, thatīs why you find an unexplained plate voltage higher than B+.
The true explanation lies in looking at what happens on the negative (or saturating ) side.
We are too used to "thinking positive" (I mean, electrically).
Now that I stated so, you can imagine that a diode matrix can be designed to accomplish that in SS.

So, I think with the flyback diodes these lobes would be squared off and the speaker cone braked somewhat.

True.

JM, did you find that your "outside the rails" SS output stage was any louder than the same output stage restricted to the rails?

Not "louder" but "sharper" and, paradoxically "cleaner", less muddy, which is the curse of the overdriving SS stage.
To be fair, I should not say SS because itīs unfair.
You can do anything SS with the proper architecture
I should say "BIG Op Amp" output stage
I say not significantly louder because, after all, all we add is a small spike at the leading edge of every squarewave generated by overdriving.
It was typically about 5 to 8V peak beyond the rail voltage, (equivalent to, say, 50 to 80 "extra" volts in a tube stage) but it was also typically, say, 10% to 20% the width of the squarewave.
You work with SMPS so you have seen it may times: itīs the leading edge peak overshoot which is typically killed with snubbers.
The difference is not in actual volume but in perceived note definition.

I listen all day long to "Belly up for Blues" Internet radio, which I *strongly* recommend to everybody.
Itīs an endless source of *killer* guitar sounds , and most of the time I "see the waveforms in my head" because I matched waveform to sound for ages.
The myth "transistors clip squarewave , tubes clip rounded" doesnīt last a second there.

[Steve Conner]

Well, more note definition and less mud sounds like a good deal for the price of 2 diodes, I'll go ahead and try it! Hopefully it'll help to "cut through" in a band context.

I hope this doesn't seem too off-topic, but to me Loudthud's scope shots summed up what a tube power amp does that a "big op-amp" doesn't.

I was especially disappointed with the Rocktron Velocity when I saw the schematics. It may be a current feedback topology, but it has a low impedance output, and an op-amp filter that imposes a generic "speaker impedance curve" on the incoming signal, as described in Keir's Marshall patent.

I totally agree, you can build transistor circuits with very rounded clipping, and some classic tube circuits clip very hard. (18 watt Marshall anyone? ) And soft clipping isn't always musically useful, either. Sometimes it needs to be hard and nasty. You get a guitar tone that sounds dreadful on its own but fits perfectly in a band mix.