Tube vs solid state power amp volumes and speaker load

[printer2]

The key thing is that the current doesn't need to be there, as, over the frequency range of interest, speaker impedance is predominately greater than nominal. The high output impedance of a tube amp will probably allow better power output, into load impedance values greater than nominal, than the minimal output impedance of a regular solid state amp.

True to a point. The reason we have a presence control, the amp and speaker combination I am working on right now sounds muffled with NFB but it helps out in the bass end. I am rebuilding it right now and will see if I need tone shaping in the NFB loop.

Yes, but that's into a resistive load of the rated nominal value.
How about if you double it, eg as at treble resonance; which gives more then, tube or solid state? How about x10 nominal, eg at bass resonance?

I don't use a resistive load when I am tweaking a circuit and so far I have stayed under 20W on my amps so the sound level is manageable. I have a good set of ear muffs that I put on when cranking the amps. So my comment on the amps still putting out square waves when they hit the rail was with a speaker attached. I'll check out the speaker resonance on the one I am working on when I get it all put together.

The thing is that tubes are lossy, so a tube amp HT power supply needs to have more capability than a similarly rated solid state amp.

Things would be evened out if the solid state amp was configured to have a similar (full load) output impedance as a regular tube amp, but then its power supply would need to have higher voltage rails and it would be less efficient.

But that is what I am saying.

[Mick Bailey]

Comparisons between SS and tube relating to output/speaker impedance become a little more even when we consider current or mixed-mode feedback. I've had pretty impressive results when taking a 'vanilla' voltage feedback power amp and converting to mixed-mode. The playing dynamics and frequency response are transformed.

[ric]

Comparisons between SS and tube relating to output/speaker impedance become a little more even when we consider current or mixed-mode feedback. I've had pretty impressive results when taking a 'vanilla' voltage feedback power amp and converting to mixed-mode. The playing dynamics and frequency response are transformed.

Would anybody care to lay out the basics of "mixed mode feedback"? Read most of this thread and just went back n skimmed it all n did a search...

I'm guessing voltage fb is what is common- signal from ot secondary + is returned through a resistor and sometimes a cap as well to a previous stage...

Mixed mode?

Thanks- r

[Malcolm Irving]

What causes the "shoulder spikes?"

[ATTACH=CONFIG]38924[/ATTACH]

I think that while the signal is passing through zero, both tubes are cut-off, due to a big bias shift. (In other words we have class 'C' operation when heavily overdriving the output tubes.) The leakage inductance of the output transformer reacts to the sudden attempt to switch off, and then change the direction of, its output current by creating a 'fly-back voltage' - which is the shoulder spike.

EDIT: Is it the leakage inductance that matters here or is it the magnetising inductance (or both)?

[loudthud]

I think that while the signal is passing through zero, both tubes are cut-off, due to a big bias shift. (In other words we have class 'C' operation when heavily overdriving the output tubes.) The leakage inductance of the output transformer reacts to the sudden attempt to switch off, and then change the direction of, its output current by creating a 'fly-back voltage' - which is the shoulder spike.

Leakage inductance won't do it, just connect a resistive load. It's the speaker inductance. I've made a solid state amp without an output transformer that will create the spikes that can reach 100's of Volts.

@Printer2, if you compare the clipping Voltage between what a tube amp will do with a speaker vs a dummy load, the speaker will allow about 20% more peak Voltage. That's about 1dB.

I this thread: http://music-electronics-forum.com/t35493/ I posted a video clip of a solid state amp that produces HV crossover spikes.

[J M Fahey]

What causes the "shoulder spikes?"

[ATTACH=CONFIG]38924[/ATTACH]

Very simplified answer:

1) inductors try to keep current constant and accept only slow changes.
One known example is power supply chokes: they try to keep current constant, so they pass DC (not changing) easily and ripple (changing at 100/120Hz) with difficulty, ironing it out.

How does it do it?
If you want to change current abruptly, by changing applied voltage abruptly, inductor will fight back creating a pulse, as strong as is needed and as strong as it can *opposing* said voltage change.

Speakers and transformers are inductive so they will behave as such.

2) one abrupt change is turning a switch ON-OFF , that one is clear, remember power ON-OFF click/thump caused when powering amo ON-OFF .

What's not so clear for many is that a transistor or tube can also be considered switches, if driven well beyond linear zone, and can also create spikes.

All we need is to get into the square wave zone, a.k.a heavy clipping.

There, at every abrupt change, be it an abrupt waveform edge but also any abrupt change anywhere else, specially when one tube/transistor passes current handling duty to the other one in push-pull amplifiers.

One spikey change is when signal reaches saturation, it brutally smashes against rails or ground, that's why we have a leading edge peak ; usually no abrupt change if one tube passes current handling to the other in Class AB BUT if we are working in the Class C area (brutally overdriven) then we do not have a smooth exchange, see that there is a dead zone between both semicycles, we actually have two squarewaves separated by deadspace:

a) the positive one .
Notice it does not fully use its 50% time allowance, but it starts a little late and stops a little early.

b) deadspace between both.

That it's deadspace does not mean "nothing happens" but first semicycle turned OFF abruptly causes a spike, then second semicycle starting, stops whatever is happening during deadspace ... again a spike.

That's why we have 2, opposite polarity spikes at start and end of deadspace.

c) second semicycle: it also stops abruptly when reaching ground and causes its own leading edge peak.

So in the M18W we have 4 spikes per complete semicycle: 2 front edge peaks which we expect and 2 crossover/deadspace we didn't ... but are clearly seen.

Why other amps don't show it?

Because we need a few conditions for it to happen:

a) tube amp.

Transistors work as clamps when reverse biased "going beyond rails" because BC diode becomes forward biased or they even include reverse biased diodes in their structure (TIP14x or MosFets) or BE diode Zeners (~ 4V) or designer just adds them so as to avoid these peaks which are very destructive to semiconductors, while tubes don't mind.

b) no NFB so there is no correction and since load is fed fron pentode plates, very high impedance source by definition.

c) tubes are heavily overdriven so they work as switches producing abrupt edge squarewaves

d) grid rectification is so strong that power tubes work *well* into Class C

In a nutshell: here you heve 4 abrupt changes: 2 when reaching rails (ground is also a rail), 2 when one tube stops conducting and the other one is not there so you have 4 peaks.
This does not happen in most amps because even if cold biased by grid rectification, we do not reach an intermediate point where *both* are OFF.

Well, hope it makes sense ... looking at the waveforms while reading this makes it clearer.

Notes:
1) but where does the inductor pull from the voltage needed to create the peaks?
How can a pulse be "higher" voltage than what's available at the supply?
Are we creating "free energy"?

Nope, energy needed was *previously* stored inside the inductor, as magnetic flux in its core.

Think of inductors as "special capacitors" with a twist: a capacitor can give us a current pulse (does the word ring a bell ? ) way higher than what a supply can give us; an inductor is a "current capacitor" (instead of regular voltage capacitors) and can supply a voltage pulse even if the supply is not .... er .... supplying it.

A current peak across a load will develop a voltage peak.

[pdf64]

The thing is that tubes are lossy, so a tube amp HT power supply needs to have more capability than a similarly rated solid state amp.

Things would be evened out if the solid state amp was configured to have a similar (full load) output impedance as a regular tube amp, but then its power supply would need to have higher voltage rails and it would be less efficient.

But that is what I am saying.

I think that such a solid state amp would have to have rather higher power supply rail voltages then would normally be required for a regular solid state amp of similar rated power output.

[Malcolm Irving]

Leakage inductance won't do it, just connect a resistive load. It's the speaker inductance.
....

Not sure about that. I think those Marshall 18W curves may have been produced with a resistive load. I've seen very similar curves myself with just a resistive load.

[nickb]

While this test was into an the scope probe as load, it does show the effect of speaker inductance giving a 52V spike. I crudely used a 10V PSU to 'charge' the speaker and then removed it.

[robrob]

Thank you Juan and Malcom for answering the question so thoroughly.

J M, that wasn't a "very simplified answer."

"Flyback voltage," would have been a very simplified answer.

[Mick Bailey]

Would anybody care to lay out the basics of "mixed mode feedback"?

Some previous discussion here;

http://music-electronics-forum.com/t37800/

[printer2]

I think that such a solid state amp would have to have rather higher power supply rail voltages then would normally be required for a regular solid state amp of similar rated power output.

I wonder how much higher. We really do not know, well at least myself, how much extra output it really provides. Probably could mic and do a frequency response plot with a SS (traditional design) and then repeat the test with a tube amp. As far as I can see the initial test does not have to be done at clipping as the tube amp's impedance should not (iffy statement here) be too much different than when overdriven. This does exclude a NFB amp. Once the plot is done one could be done at a similar clipped level. A high power amp would be annoying and rough on speakers, a 10W level should suffice.

Just need s few 10W amps, a tube Class A and P-P AB amp, a chip amp which would represent many of the practice amps out there, a looser SS amp (was thinking of my LiL Tiger), a class SS A amp. Would be good to try a SS amp with an output transformer, I have an old amp with one, no idea what the rest of the circuit is yet. Have to be careful with the NFB around the transformer as you would a tube amp. A bit of work, more of a winter project though. Have a few more ideas of where this can lead, actually been thinking about it for a while.

[printer2]

Leakage inductance won't do it, just connect a resistive load. It's the speaker inductance. I've made a solid state amp without an output transformer that will create the spikes that can reach 100's of Volts.

@Printer2, if you compare the clipping Voltage between what a tube amp will do with a speaker vs a dummy load, the speaker will allow about 20% more peak Voltage. That's about 1dB.

I this thread: http://music-electronics-forum.com/t35493/ I posted a video clip of a solid state amp that produces HV crossover spikes.

I don't think I saw this one, think you had others but maybe my memory is playing tricks on me. Did you do the same test with a tube amp?

[ric]

Some previous discussion here;

http://music-electronics-forum.com/t37800/

Thanks, Mick.

[loudthud]

I don't think I saw this one, think you had others but maybe my memory is playing tricks on me. Did you do the same test with a tube amp?

There are these threads:

Link: http://music-electronics-forum.com/t39829-2/ posts 52, 55, 56.

Link: http://music-electronics-forum.com/t37469-2/ post 37.