Crazy idea - full tilt sound, lower volume.

[jeff]

I was thinking about speaker attenuation today and got an idea. You hear people saying they want the cranked sound but at lower levels. Could this be done by phase cancelation.
Picture you have a parallel PP amp 4 tubes: 2 push, 2 pull.
Now picture instead of sending the in phase signal from the phase inverter to the 2 push tubes and the out of phase signal to the 2 pull tubes you had switch that swaps  the inputs of 1 push tube and 1 pull tube. So when the phase inverter is giving full signal(and assuming everything is matched) all 4 tubes are full tilt, but the amp would not make any sound because the in phase signal is driving 1 push tube and 1 pull tube cancelling each other out, and the out of phase signal is also driving 1 push tube and 1 pull tube canceling each other out.

Now instead of sending full signal to all the power tubes setup a voltage divider so that the in phase signal is sent directly to 1 push tube and through the voltage divider(pot) only a percentage is sent to 1 pull tube. Same with the out of phase signal from the phase inverter. Directaly to 1 pull tube and only a percentage to 1 push tube.

Now the push side would have one output tube full tilt and one output tube at, let's say, 90 percent out of phase with it in parallel.
Also the pull side would have one output tube full tilt and one output tube at 90 percent out of phase with it in parallel.

Would this result in phase cancelation of 90 percent of the power and basically sound like a fully cranked amp at 10 percent the power? Full crank sound because all four tubes are working hard, but lower volume because they are working hard against each other and phase canceling?
I dunno crazy idea huh?

[simonallaway]

Wouldn't it have to be perfect cancellation across the entire spectrum? (well, the audible specturm). I am wondering if imperfections would mean you'd just get a weird filter, instead of the theoretical attenuation.

How about physical attenuation? An actual brake on the speaker. Or another speaker close behind the real one, out of phase.

[jjasilli]

I think a version of this idea is the "single ended mod" by KOC.  With 2 PP tubes, an extra vol pot (after the PI) decreases signal to one power tube.

[tubenit]

I think a version of this idea is the "single ended mod" by KOC.

IF this is what I think it is ?  ................ it was used on KOC's  "Soma 18".  And IF that is correct, I read consistently unfavorable reviews regarding the tone of using the single ended mod dialing down that pot post phase invertor.

Just an FYI.

With respect, Tubenit 

[2deaf]

In the 70's, there was a guy at a music store in my area that would put a "master volume" in 1959 Marshall's that consisted of a single pot. connected between the two outputs of the PI after the blocking cap.'s.  I would remove those MV's (for a fee) because allowing opposite phases to fight it out in order to reduce amplitude results in a sound that is unacceptable to the vast majority of folks (although apparently not as unacceptable as it used to be).  You're going to get the same crappy sound whether you have the post PI signals bucking or the power tubes bucking.

and basically sound like a fully cranked amp at 10 percent the power?

The output tube distortion that people are seeking is the result of many factors, not just the overdriving of an output tube.  NFB from the speaker tap is one of the essential ingredients  of the sound and touch and it will be rendered ineffective with a post PI MV or with incorrectly phased output tubes.  Other factors will also suffer the same fate.

Or another speaker close behind the real one, out of phase.

In order for out-of-phase speakers to cancel, they have to be mounted in the same plane and very close to each other.  They only cancel lower frequencies and produce a funny sound sort of reminiscent of the above mentioned arrangements.  One of those guys with a modified Marshall also had one of the speakers in his cabinet out of phase.  I suspect the same tech did that.

[jeff]

In the 70's, there was a guy at a music store in my area that would put a "master volume" in 1959 Marshall's that consisted of a single pot. connected between the two outputs of the PI after the blocking cap.'s.  I would remove those MV's (for a fee) because allowing opposite phases to fight it out in order to reduce amplitude results in a sound that is unacceptable to the vast majority of folks (although apparently not as unacceptable as it used to be).  You're going to get the same crappy sound whether you have the post PI signals bucking or the power tubes bucking.

I don't believe you are doing the same thing, here's why.
Let's assume that MV is turned all the way down. What is happening? The positive side of the phase inverter is cancelling out the nevatige side and no signal is present on the grids.
But on the other hand if you were to swap the grid of one push tube with one pull tube now each grid gets the full signal from the PI.
In the first example you are using phase cancelation to lower the signal to the power tubes so they aren't being pushed at all.
In the second example you are driving all the power tubes full blast and using phase cancelation to reduce the power to the OT.
The first example is purely PI distorion, the second would be purely power tube distortion. It may very well sound crappy but it would not be the same crappy sound as the post PI MV.

OK guess it was a crazy idea.
Thanks for talking it out with me
Jeff
 

[sluckey]

It's definitely an odd idea. Probably no one has tried it. Why don't you do it an report your findings? Could be the next thang! You'll be famous!

[jeff]

I don't have an amp right now with 4 power tubes or else I'd try it. Seems simple enough. Just a thought experiment and if it works for someone out there, great.
 
If this does work is there anyway to protect this idea from being pateneted? So no one company can have the exclusive rights to use this idea. I know there's one company out there with some BS patents that aren't really original or inovative, they just got to the patent office first and now no one can use that idea. If this works and sounds good I like any person or company that wanted to use it be able to do so.

[sluckey]

If this does work is there anyway to protect this idea from being pateneted?

Too late. You done let the cat outta the bag!   

[jeff]

I want the cat outta the bag! I just don't want anyone else to put the cat in their exclusive bag.
I don't want to own the pantent for this I just don't want any one company to own the patent.
If this works it should be public domain.

[jeff]

Here' the schematic. UNTESTED SCHEMATIC

The 250K ganged pot is wired backward so on 10 the outer tube grid are grounded giving no cancelation but turning off two power tubes. As you turn up(down) it intruduces more cancelation. The pot could be just a set voltage divider.

 EDIT: SCHEMATIC REMOVED DUE TO ERROR (thanks 2deaf)
X

[2deaf]

The junction of the dual-ganged pot.'s goes to the negative bias voltage instead of to ground, right?

[jeff]

Oh shoot! YES you're correct. That wouldn't work as drawn. Good catch.
  The pot would need to go to bias, not ground.
thanks

Ok maybe not such a good idea to have a pot with DC on the grid of a power tube. That could be very noisy when using it. Hmmm....

[Merlin]

I want the cat outta the bag! I just don't want anyone else to put the cat in their exclusive bag.

This thread already nulls any such patent.

With two output tubes cancelling each other out, the load impedance seen by each one will be reduced. The closer you get to complete cancellation, the lower the impedance (theoretically zero if you had perfect cancellation, but not that low in reality). This might potentially lead to redplating. It also means the load impedance -which is an important part of the tone- is not preserved using this method.

Ok maybe not such a good idea to have a pot with DC on the grid of a power tube. That could be very noisy when using it. Hmmm....

Virtually no current flows in the grid leak, so you may still be OK with that.

[2deaf]

Ok maybe not such a good idea to have a pot with DC on the grid of a power tube. That could be very noisy when using it. Hmmm..

I've put dual-ganged post PI MV's hooked up to the bias into other people's amps for decades and they work really good.  This is completely different than having unwanted DC on a pre-amp pot.

Virtually no current flows in the grid leak, so you may still be OK with that.

Once you overdrive an output tube to a certain point, a whole bunch of current flows through the grid.  That still doesn't matter to the post PI MV.

[jjasilli]

With two output tubes cancelling each other out, the load impedance seen by each one will be reduced.

Yes, that's the "beauty", at least in theory, of KOC's single-ended mod.  By reducing signal to one power tube, the plate-to-plate impedance is preserved.  As one-half of the signal is attenuated, phase cancellation becomes imperfect, and eventually zero; but still, no SE OT is needed.  I haven't actually heard one of these, but I trust Tubenit's ear.   


Partial phase cancellation works great in other applications, such as Lindy Fralin Unbucker pickups.


Still reducing the amp's output by 1/2 will not have a significant difference in perceived volume, which is logarithmic.

[kagliostro]

Here is the experience an italian guy had building a PP/SE amp following KOC indications


http://www.mediafire.com/view/zjicp230r2m2tto/SOMA84.pdf


Here a sample of the amp


http://www.mediafire.com/download/4zoa7441kkrsr4j/PP-SE_amp_sample.mp3


K

[2deaf]

Here is the experience an italian guy had building a PP/SE amp following KOC indications

This amp just goes from 2-tube PP to 1-tube SE.  There is no phase cancellation involved like Jeff is talking about.

[kagliostro]

Quote from JJasilli

I think a version of this idea is the "single ended mod" by KOC.

I haven't seen the schematic that Jeff posted and deleted and I was referring to what JJasilli told

K

[HotBluePlates]

Here is the experience an italian guy had building a PP/SE amp following KOC indications

This amp just goes from 2-tube PP to 1-tube SE.  There is no phase cancellation involved like Jeff is talking about.

You're right, but it accomplishes what Jeff was trying to do in a way that work in actual practice.

The output stage requires equal push-pull signals to develop full output power across the transformer primary. By turning down the drive to one side from full to zero, power output is reduced from full power to some much-lower output. Load impedance is maintained throughout, though arguably too-low for the SE mode to develop all the power it might be capable of, but given that's the lowest-power setting this isn't a problem.

So this amp reduces output power by reducing the total signal current through the OT primary. It just doesn't use phase cancellation, though the next effect is similar.

[2deaf]

it accomplishes what Jeff was trying to do in a way that work in actual practice.

Jeff is trying to get quad output tube overdrive without a visit from his local police department, but his idea may involve a misunderstanding about how PP tube amps work.

Imagine an alternate reality with a class B quad PP tube amp.  Now remove two tubes from one side and name the tubes on the other side "A" and "B".  Hook "A" up to one side of the PI and "B" up to the other side and apply a sine wave to the PI.  Now when the grid of "A" gets the positive half of the PI signal, it puts out half a sine wave.  When A gets the negative half, it does nothing.  "B" does the same thing except 180 degrees later, so it puts out half a sine wave while "A" is doing nothing.  This is an alternate reality so that the idle plate of "A" doesn't affect the active plate of "B".  Now you're getting twice as many positive halves of sine wave out as you put in.  Put a 1kHz sine wave in and get a 2kHz ugly signal out.

Now sleep off whatever put you into an alternate reality and see if this is what happens in normal reality with a class AB tube amp.

[jeff]

OK so here's a schematic of the idea with the switch in normal position. Basic PP, nothing special.

EDIT: I drew the connections made by the switch and omitted the connections broken by the switch for ease of viewing in this and the next 2 schematics. Full scematic with all connections is at the end. Sorry if this caused confusion but I wanted to show my line of thought step by step and by showing only the connections being made by the switch one position at a time I thought it would be less jumbled and easier to read.

In an attempt to make it less confusing I made it more confusing. sorry.EDIT

[jeff]

Now let's say the other side of the switch is wired like this:
Wouldn't you be driving all four power tubes full blast, but getting total phase cancelation?

[jeff]

By just looking at tubes A and C for a second, each tube is fully cranked, but because they are in phase, and attached to opposite sides of the OT wouldn't they cancel at the OT? Whatever A is doing C is also doing, total cancelation.

Same would go for tubes B and D except they'd be driven by the other side of the PI.

[jeff]

But we don't want total phase cancelation, no point in that, so replace the 250K resistors with a ganged 250K pot and wire it so tubes A ant D get 100% the PI signal and tubes B and C get a variable percentage less than A and D.

So if we set the pot output to 90%:
 A is at 100% being canceled by C at 90% = 10% push
 B is at 100% being canceled by D at 90% = 10% pull

Since full up on the pot is total cancelation wire the pot backwards.
So with the switch in this position wouldn't the output would be varible from 50% to 0%?

[jeff]

Here's the full schematic UNTESTED SCHEMATIC!
You'd momentarily lose bias as the switch is flipped so maybe a make before break switch is needed.

[HotBluePlates]

You're doing a lot of switching work to accomplish what a cross-line master volume does with a single pot, and completely variable from full-up to full-off.

I don't have a lot of experience with cross-line master volumes, but I recall Tubenit saying he didn't care for the sound.

[kagliostro]

The way Jeff draw his schematic, the resistance of the pot that is in series with the signal vary also for the bias


I don't know what will happen about that


K

[2deaf]

cross-line master volume does with a single pot

What is a cross-line master volume?

[HotBluePlates]

The way Jeff draw his schematic, the resistance of the pot that is in series with the signal vary also for the bias

There's no direct current through those pots, so the bias won't change.

What is a cross-line master volume?

A pot from the push drive line to the pull drive line. The pot is wired as a rheostat, and at minimum resistance the two signals are shorted together; output tubes get no drive signal at that setting.

There is usually a cap on one or both sides of the pot to block d.c.

[2deaf]

I'm pretty sure that is what I was talking about at the start of this thread.  No cap.'s were used because the pot. was in the realm of negative bias voltage.  The idea was to allow the PI to overdrive and then send whatever level you wanted of this clipped signal to the output tubes.  The two main problems are that PI's sound very harsh when overdriven and that phase cancellation with asymmetrical, uneven clipped signals yields what I called a "crackling" sound.  People who had those in their amps agreed with tubenit.

If you're hellbent on having a MV like that, it helps to use a 91K and a 100K for your PI plate resistors instead of 82K and 100K.

[2deaf]

they are in phase, and attached to opposite sides of the OT wouldn't they cancel at the OT?

No.  Probably knock it down an order of magnitude or so, but wouldn't cancel.  If it would totally cancel, you wouldn't need any filter cap.s in your bias circuit.

Whatever A is doing C is also doing, total cancelation.

You're thinking Class B.  Class AB doesn't do that.

So with the switch in this position wouldn't the output would be varible from 50% to 0%?

No.

[2deaf]

I forgot to ask what you are going to do about the frequency doubling problem.

[HotBluePlates]

I'm pretty sure that is what I was talking about at the start of this thread. 

Exactly. You just asked what I meant about "cross-line master volume." Here's the cross-line in a Matchless Clubman.

You're right about the caps not being needed; dunno what I was thinking, as the master is after the coupling caps.

[kagliostro]

At the moment I'm tired, so I'm not sure, but to me seems that this can be on the topic

http://www.ampbooks.com/home/amp-technology/amp-patent-7173488/

Franco

[jeff]

Let me ask a hypothectical question.
What would happen if you took a CT OT and soldered the two end taps together then applied an AC voltage across the CT and the two ends. Would you have and voltage on the secondary, or would the voltages phase cancel?
Now let's say you have one power tube. If you attached both ends to the plate and the CT to the B+ would you get anything from the secondary?
Now let's say you have two power tubes. Biased the same, inputs in parallel, both tubes in phase, both tubes outputting the exact same wave in phase. If you connected one end of the OT to one tubes plate and the other end of the OT to the other tubes plate and connected the CT to B+, would you get any output or would the power tube phase cancel? (no PI circuit involved at all)


Now take a look at the third schematic again. Shouldn't tube A and tube C phase cancel?
(The picture only shows tubes A and C, I drew tubes B and D faintly with a pencil just to focus on what A and C are doing but they didn't show up on the scan. B and D are still in the circuit and driven from the other side of the PI but just can't be seen in the picture. I just wanted to highlight what A and C were doing)

In the second schematic, grid A and grid C have the same input, and grid B and grid D have the same(out of phase) input instead of normally tube A and B having the same input and tubes C and D having the same out of phase input. So shouldn't tube A and C phase cancel, and tube B and D phase cancel at any input level?

I may have be confuzing  because how I drew each schematic. I omitted the connections that were disconnected by the switch, only showing the connections made by the switch in each position, so each schematic wouldn't look so jumbled and we could focus on what's happening in each position step by step.

 look at the final schematic to see all the connections.


 

[jeff]

Please read my previous post 1st.
I definatly dig what you're saying about the cross line MV. But the whole reason I started thinking about this is because this works differently.
Let's think about each set to 0:
With the cross line MV it's not really applying the same signal to the push tubes as you are to the pull tubes and driving the power tubes hard. What's happening is the pos half of the PI is canceling out the negative half of the PI before it gets to the grids. *NO* signal is getting to the power tubes(assuming perfectly balanced PI)

With this crazy idea set up you're applying all the signal from the PI to the power tubes grids. But you're applying the pos side of the PI to one push tube and one pull tube, and the negative side of the PI to one push tube and one pull tube. Power tubes are at full tilt but no signal on the OTs secondary.

So if both amps are set to 0 neither amp will make any sound(theoretically). But the power tubes of the cross signal MV amp won't be doing anything, whereas with this crazy idea the power tubes will be fully cranked. Power tube distortion vs. PI distortion.

Maybe the reason the CSMV sounds so bad by all accounts is because it's making the wrong tube work hard(PI not power). That's basically why I came up with this idea, keep the power tubes cranked, but the output low.

I don't know how well this would work or sound on 1 but it seems different than the cross signal MV to me in so much as which tubes are doing the distorting and seem worth asking, and learning about.

Thanks for all the help

[2deaf]

you took a CT OT and soldered the two end taps together then applied an AC voltage across the CT and the two ends.

Ahh, now that should cancel, at least to the extent that both sides are symmetrical.

Now let's say you have one power tube. If you attached both ends to the plate and the CT to the B+ would you get anything from the secondary?

Same as above.

Now let's say you have two power tubes.

Now you have a very different situation.  In the first two examples, you have the same signal source.  In the third example, you have two different signal sources.  They will not behave the same.

this works differently.

I've understood what you meant the whole time, its just that I'm reasonably certain that it will not work.  I use phase cancellation extensively for gain reduction, signal reduction, active tone controls, frequency response, high fidelity, and in distortion-producing circuits.  I've worked on output tube overdrive for decades (of course the only thing I achieved was to upgrade my working hypothesis to a theory).  I don't know exactly what you would get with your setup, but I would expect a real ugly signal that probably has a notch or two in it.

[I would say that last N.O. touchdown just iced the cake.]

And there is still that nagging problem with frequency doubling.

If someone had told me in 1974 what I am telling you today, I wouldn't have believed it anymore than you believe me.  I think you should build it.  I could be wrong.

 

[PRR]

I don't see a frequency doubling issue, as long at it all runs class A.

The instant any tube hits cutoff, I think all hell breaks loose.

Building a 30 Watt amp plus more knobs seems a hard way to go to get 4 Watts.

But WTH. Build it and try it.

[2deaf]

The instant any tube hits cutoff, I think all hell breaks loose.

That's where I think the frequency doubles.

[jeff]

Now let's say you have two power tubes.

Now you have a very different situation.  In the first two examples, you have the same signal source.  In the third example, you have two different signal sources.  They will not behave the same.
 

I don't think you read my whole question. Read what comes next. I wasn't saying two tubes PP. Just two identacal power tubes identically biased with inputs in parallel(identical signal to each tube), but attached ot opposite ends of a CT OT. Why wouldn't that phase cancel?

[2deaf]

I don't think you read my whole question

Is this different than CMV Half.jpg?

[jeff]

No, but is there any reaon that wouldn't phase cancel?

[jeff]

If you took a 4 tube PP amp and disconnected the connections between the two outer tubes grid to their appropreat sides of the PI and reversed them would you get total phase cancelation?

I guess I'm having trouble seeing why not.

[2deaf]

I don't see a frequency doubling issue, as long at it all runs class A.

I guess if the combined current from the two tubes on one side is always the same, frequency is a moot point.  If the rise in current in one tube trumps the decrease in current in the other tube on the same side, then there will be an overall increase in current and this event will happen twice in every cycle instead of once.

[jeff]

OK I think I finally get it why this won't work.
 
At first I was only seeing that if you removed tubes B and D, tubes A and C would totally cancel. And without tubes A and C tubes B and D would cancel. But get it now. With all 4 tubes in something very different happens.
 
Thank you guys very much for talking this out with me. I like this site because people here don't say that's a bad idea it won't work, but take the time to explain and help me understand why it won't work and doing so I learn.
I learned alot from you guys over the years and am still learning, thank you all for your patience with my crazy ideas, you guys are the best.
 
Thank you all
  Jeff

[2deaf]

How do you insert schematics into your post?

[sluckey]

I played with my Marshall 18W today. Disconnected the PI drive to one of the EL84s and drove both EL84s from the same side of the PI. Almost total cancellation. I had to dime the volume to get any sound. It was a nasty distorted sound.

So the phase cancellation idea does work as I expected. Common mode rejection is a characteristic of P/P amps. I also believe there will be significant cancellation with a parallel P/P amp but my only amp with that output is a TRRI. Too much trouble for me. I'm unconvinced that it's a worthwhile idea though. The whole idea just seems too inefficient.

[2deaf]

I have more numbers and 'scope pictures, but I'm missing something about how to attach them.  My son got the schematic above to attach, but he doesn't know how and can't seem to do it again.

[PRR]

2deaf, are those notes from actual experiment?

I have a simulation running. So far I have confirmed that simulations suck, and four tubes is too many to keep straight in head and on drawing. However it does cancel (of course), *and* I do not see any odd action in class B except the expected crossover S-curve with 50% drive to the cancellation pair and fairly high drive levels.