Output Transformer Primary Questions

[FlaCharlie]

When drawing load lines, you basically have two lines, one Class-A line, and one steeper Class-B line (and than mirrored obviously for a full sine wave).
I think the easiest explanation of this, can be found on Merlin Blencowe's website; http://www.valvewizard.co.uk/pp.html
So your Class-A power is up till the point the other graph line goes up into Class-B mode, the "kink" in the graph

The maximum output power is just the (maximum) RMS voltage times RMS current P = V * I (or P = E * I, depending where you walk around on this planet)
The RMS voltage is HT-Vmin / 1.41.
HT is the total anode-cathode voltage, Vmin is the voltage at the Vg = 0V line.
The RMS current is; Ipeak (at Vg=0V) / 1.41

So P = Vrms * Irms = ((HT-Vmin) / 1.41) * (Ipeak / 1.41) = (HT-Vmin) * Ipeak / 2.
With the same formula you can calculate the Class-A power.
So before the "kink" in the graph, were now Vmin is the transition point in the graph

A (major) downside of running in Class-A is that things will draw a constant current all the time.
Which means that things get HOT very quickly.
So that is something to keep in mind when choosing which tubes you want to use, don't go for small tubes like EL90/6AQ5, but go for a big sturdy 6V6 for example.

Practically speaking; to run your design more into Class-A mode, just lower the HT voltage somewhat, get an higher primary impedance, and use an higher bias current.

Thanks . . . these are the same formulas I used previously.

My confusion about the power produced using @PakProtector Douglas' operating points and 2k OT was just based on a miscalculation. So that answers my earlier questions #1 and #2.

I assume he suggested those because he wanted to operate in Class A even though the power output will be lower and more heat will be generated.

So to re-state my Question #3, now in two parts:

Is there a sonic advantage to operating a PP amp in Class A vs. Class B?

Are there measurable, technical, advantages when operating a PP amp in Class A vs. Class B?

 

[kward]

Late to thread, but here's my input.

Sonic advantage of PP class A vs AB1? I think very little sonic advantage. I've built Class A and AB1 PP amps that were otherwise identical except for the output stage biasing (and of course plate voltage used). Both used 6L6 tubes, and I can't tell a difference sonically between them. The AB1 amp has more power of course, but within the power band that the class A amp produced, the two sounded nearly identical. YMMV.

Technical advantages PP A vs PP AB1:

• PP Class A
  • Idles output tubes hot to maximize available power, tubes wear out faster
  • Less power output
  • Constant current draw, less output stage distortion
  • Less demanding drive requirements can mean simpler circuit
  • Typically cheaper to build
  • Often paired with triode operation and no global feedback for sweet laid back sound

• PP Class AB1
  • Idles the tubes much cooler, tubes last longer
  • More power output
  • Non constant current draw--higher output stage distortion (not an issue in my mind if a bit of global feedback is used)
  • Typically more expensive to build
  • Often paired with pentode or UL operation with global feedback for more controlled powerful sound

What you didn't ask (or maybe you did and I didn't see it) is advantages of triode vs pentode wired output stage. Here I believe there are advantages sonically to triode operation, depending on the style of music you listen to. Don't use your triode amp for a pool party--leave that job to the large wattage tube pentode or UL amps. But for home listening at not loud levels, the triode amp really draws you in.

I recently built a 6AV5 triode connected push pull class AB1 amp. I absolutely love the sound if this amp. I used a little global feedback (about 10 dB) in that design. As disclaimer I prefer the "East coast" or "British" laid back sound that you cozy up to, vs the "west cost" sound that is more visceral and chest thumping.

 

[kward]

Is there a predictable relationship between screen voltage and load impedance? Such as lower screen voltage if you want to use higher impedance or vice versa?

The relationship is inversely proportional. The lower the screen voltage the higher you want the primary impedance to be of the output transformers to maximize power output available at that particular screen voltage selected.

In other words, you can set the screen voltage to best match the primary impedance of the output transformers you have, to maximize output power that would be obtainable given a set of output transformers with a given primary impedance.

 

[kward]

Question #2: Assuming I've done it correctly, does this mean that the overall maximum power output drops when it transitions into Class B?

No. It's a smooth continuous transition from A to B (in an AB amp). Max power (by definition) of an AB amp is defined by it's class B load line, which is defined by plate voltage, primary impedance, and tube type. At max power the output stage (by definition) is operating on it's class B load line internally.

Note on terminology: an AB amp is always an AB amp regardless of what instantaneous mode it might be operating in internally. In other words the class AB is a definition of architecture, not a definition of instantaneous mode of operation. Another way to say it is an AB amp is not sometimes a class A amp and other times a class B amp depending on instantaneous output power. Rather it is always an AB amp but has two internal modes of operation depending on how much power is being delivered at the given instant.

 

[GordonW]

What is the difference in architecture of a PP class A vs. PP class AB? Do you consider twisting a pot slightly to change the wiper position a different architecture?

It takes more than that, to go from an optimal A to an optimal AB operating mode.

Usually, a different primary impedance, and a different B+ voltage, is needed for each type of operation.

Just turning the bias pot setting takes the amp toward or further away from its optimal operating parameters. It doesn't change whether it's more suited for A or AB.

Regards,
Gordon.

 

[Tom Bavis]

Well, by twisting a pot, you can make a class A amp into a poorly optimized class AB amp, or vice versa. Assuming the same power tubes, class AB would use a higher supply voltage and lower transformer impedance than class A, and more voltage swing from the driver. Class A can be cathode biased, class AB must be fixed bias. There's more to it than idle current.

 

[GordonW]

The definition of each class has no "optimization" requirement. Have you ever built the optimum amp for its class?

Whenever possible, in fact.

There's a distinct range of optimal values for each output tube, for Class A, and another different range of optimal values for Class AB1. It's shown right on the tube data sheets, and is verifiable by plotting the plate curves.

Regards,
Gordon.

 

[GordonW]

T'aint right.

GOOD Class AB1 MUST be fixed bias.

A cathode bias AB amp is basically a distortion generator, if it's going very far into cutoff on the idle tube on each side of the cycle, due to the change in effective transformer loading, and by going near the onset of blocking distortion. It may be made listenable by feedback- but it doesn't inherently work well.

Regards,
Gordon.

 

[GordonW]

Sounds a bit snooty. Lotta Dynaco and Maggie fans would disagree. Probably a lot of other DIY builds, too.

None of those go appreciably far into cutoff. Especially Magnavox. They are usually pretty low B+, and high primary impedance.

And- do note- the cathode-bias Dynaco amps need a COMMON cathode resistor, to stabilize the cathode voltage- or, they won't make full power. Look at the thread Dave Gillespie posted on this very subject! In effect, the common cathode resistor acts as a "quasi-fixed" bias- reducing blocking distortion and bias shift.

The Magnavox amps also have a common cathode resistor- which performs the same function. Though, with the "light" loading of the Magnavox transformers (high primary impedance), it's less of a factor than in the Dynaco amps.


Regards,
Gordon.

 

[kward]

What is the difference in architecture of a PP class A vs. PP class AB?

AB combines A and B to achieve a mode of operation with more efficiency than A but with lower distortion than B. It’s done by biasing both tubes tubes in the push pull pair so they both conduct when the signal is close to zero, but at max input signal positive only one tube is conducting, and at max signal negative only the other tube is conducting.

You don’t have those conditions in a class A stage or a class B stage. Hence it is it’s own class of operation.

 

[kward]

If there’s a class A envelope as you say then there is a space outside of that envelope that is not the class A envelope. You call the space inside the class A envelope as class A, I call it the A portion of the AB mode. You call the space outside the class A envelope as class AB. I call it the B portion of the AB mode.

I label that entire thing as the definition of what class AB is, and that may be where we differ in semantics.

You say potayto I say potahto Anyway, clearly I think we are describing the same thing.

 

[PakProtector]

The AB covers it sufficiently. It means both conduct at idle and one at full power. The two conducting is still not A; that is reserved for the operation when max power is achieved *AND* both are still conducting. As youse have said, A is one thing, B is another and AB is a third that describes conditions that are not either A or B. If it is AB, the whole thing is AB. AB has no A or B...
cheers,
Douglas

 

[FlaCharlie]

The discussion of theory is informative but I'd like to focus, for a moment at least, on the practical aspects of the project I'm considering.

There's a distinct range of optimal values for each output tube, for Class A, and another different range of optimal values for Class AB1. It's shown right on the tube data sheets, and is verifiable by plotting the plate curves.

So, back to the data sheet in my first post. It suggested the following operating points:

Footnote 1: Class A (SE Pentode) with 150v Plate, 150v G2, -7v Grid using a 6K load yields 1.5W

Footnote 2: Class A (SE Pentode) with 240v Plate, 160v G2, -10.2v Grid using a 7K load yields 2.6W

But there is no mention of PP operating points at all. So I'm trying to figure out a good load in PP based on the data available for SE and the curves I posted earlier.

I'm not too concerned about whether I'm running in Class A or AB. I have lots of PP amps and I assume most all (maybe all) are running in AB.

I'm thinking of building this in a Magnavox 8802 chassis with the stock PT. The Sams shows a B+ of 255v but I expect a bit more due to modern wall voltage.

Although I have a few options for OTs, using the stock OTs would be the simplest.

Despite being listed in the Sams as 6k, they actually measure ~7.75k into 4 ohms or 15.5k into 8 ohms. I would like to be able to use 8 ohm speakers.

Earlier, I gave the example of a highly regarded PP amp that was designed by Gary Pimm years ago using the 47. The 47's data sheet, like the 4P1L / 12P17L, suggests a 7k load for a single tube and it also lacks PP info. Pimm suggested the use of a 14k load, which is 2x the SE load in the data.

Douglas acknowledged this approach:

The usual thing is to double the SE load for PP. This works as long as you STAY CLASS A. If you go into an AB design, it gets more interesting...

So, given the expected B+ of ~255v and the use of a 15.5k OT, what operating points would ensure that the amp "stays in Class A"?

The load lines I've drawn all seem to be AB and I have some operating points that would seem to work with the expected B+ and loads of 7k to 8k.

 

[GordonW]

The discussion of theory is informative but I'd like to focus, for a moment at least, on the practical aspects of the project I'm considering.


So, back to the data sheet in my first post. It suggested the following operating points:

Footnote 1: Class A (SE Pentode) with 150v Plate, 150v G2, -7v Grid using a 6K load yields 1.5W

Footnote 2: Class A (SE Pentode) with 240v Plate, 160v G2, -10.2v Grid using a 7K load yields 2.6W

But there is no mention of PP operating points at all. So I'm trying to figure out a good load in PP based on the data available for SE and the curves I posted earlier.

I'm not too concerned about whether I'm running in Class A or AB. I have lots of PP amps and I assume most all (maybe all) are running in AB.

I'm thinking of building this in a Magnavox 8802 chassis with the stock PT. The Sams shows a B+ of 255v but I expect a bit more due to modern wall voltage.

Although I have a few options for OTs, using the stock OTs would be the simplest.

Despite being listed in the Sams as 6k, they actually measure ~7.75k into 4 ohms or 15.5k into 8 ohms. I would like to be able to use 8 ohm speakers.

Earlier, I gave the example of a highly regarded PP amp that was designed by Gary Pimm years ago using the 47. The 47's data sheet, like the 4P1L / 12P17L, suggests a 7k load for a single tube and it also lacks PP info. Pimm suggested the use of a 14k load, which is 2x the SE load in the data.

Douglas acknowledged this approach:



So, given the expected B+ of ~255v and the use of a 15.5k OT, what operating points would ensure that the amp "stays in Class A"?

The load lines I've drawn all seem to be AB and I have some operating points that would seem to work with the expected B+ and loads of 7k to 8k.

It sounds like you need more B+ voltage. That will allow the effective use of higher primary impedance. You will just need to make sure you can provide the extra B+ from the power supply (if it had a tube rectifier, going to a soild-state rectifier can help).

If it looks like it is going to be AB- just make it fixed bias. It will use more of the B+ that way, anyway (no voltage lost to a cathode bias resistor).

Regards,
Gordon.

 

[FlaCharlie]

It sounds like you need more B+ voltage. That will allow the effective use of higher primary impedance. You will just need to make sure you can provide the extra B+ from the power supply (if it had a tube rectifier, going to a soild-state rectifier can help).

If it looks like it is going to be AB- just make it fixed bias. It will use more of the B+ that way, anyway (no voltage lost to a cathode bias resistor).

Any idea about approximately how much more B+ is needed to enable the use of OTs that are 13k to 15.5k? I'm wondering if going to SS rectifier will be enough. It's got a 6CA4 rectifier stock.

The AB operating points I've come up with include the use a 9v Lithium battery to provide a non-adjustable fixed bias. The downside of that option is that I'll either have to use 4 ohm speakers with the stock OTs or replace the OTs, which requires drilling the steel chassis. That won't be easy since I don't have a drill press, just a hand drill.

 

[FlaCharlie]

If your PS can deliver the current needed and your OPT primary can handle the current calculated for 95%-100% of your tube's maximum plate dissipation, then you set the bias voltage to provide that current calculated via the plate voltage you'll have. What is the max. plate W dissipation for your tubes?

Maximum plate dissipation is 7.5w, although some of the guys who run these in triode PSE are known to push them a bit harder.

The stock amp runs 6V6s and, according to the Sams, each pair is running at ~42mA. Who knows how much current the OTs can actually handle. I'll probably be running a bit more, I suspect. The guys running them in triode claim they sound best with 25-30mA per tube.

The PT supposedly supplies 100mA, which includes the current draw from an external preamp and/or tuner. But, based on my recent experience with another build using a different Magnavox chassis and transformers, I suspect it might be able to handle more than that.

So the question remains . . . approximately how much more B+ do I need in order to make use of the stock OTs, which are 15.5k into 8 ohms?

 

[kward]

approximately how much more B+ do I need in order to make use of the stock OTs, which are 15.5k into 8 ohms?

With a (randomly chosen) 300V B+ and 150V screen, and with the 15.6k OPTs, you could build a PP class AB1 amplifier that would deliver somewhere between 6 and 7 Watts output. You would bias the output stage at something like 18 mA per tube. That idles each tube at approximately 5.5 Watts static dissipation.

 

[Kidmoe]

Despite being listed in the Sams as 6k, they actually measure ~7.75k into 4 ohms or 15.5k into 8 ohms. I would like to be able to use 8 ohm speakers.

Not to belabor the thread but are these your own measurements of the stock 8802 output transformers or a reference to someone else measurements? Reason for asking is because if you reference the big 8802 thread that Dave G. did he gets a winding ratio of about 50:1 which equated to a 10K load into 4 ohms. I have verified these measurements to be accurate via 3 8802 chassis that have passed through here.

Any idea about approximately how much more B+ is needed to enable the use of OTs that are 13k to 15.5k? I'm wondering if going to SS rectifier will be enough

Also, referencing Dave G's 8802 thread (as well as my own measurements) he gets 305V to the center tap of each output transformer when each tube is drawing 40ma of current. This is with SS rectification. Even if you use tube types that draw less current, there is only so much B+ that you can squeeze out of the little 8800 series pt.

 

[kward]

So, given the expected B+ of ~255v and the use of a 15.5k OT, what operating points would ensure that the amp "stays in Class A"?

At 15.6k primary impedance, 260V B+, 150V screen, PP class A operation, you would idle each tube right at the limit of 30 mA. The stage would deliver approximately 3 wats output. This output stage configuration is highly NON-optimized for class A operation because of non-even current swing-ability. The plate voltage is too high so that the tubes will be run into cutoff on the negative going input signal long before the tubes hit zero grid volts on the positive going input signal. The stage is really operating as an AB1 stage under those conditions. But it isn't even optimized for that class of operation either...so it's a lose-lose proposition.

To optimize the stage for class A operation (where optimization in this context means max output power the stage can deliver) you would want 180V plate, 150V screen, and bias each tube at 40 mA (this assuming 15.6k OPTs). That will position the bias point so that there is roughly equal current swing-ability on either side of the bias point. In this situation, the stage will deliver right around 5 watts output.

 

[FlaCharlie]

Not to belabor the thread but are these your own measurements of the stock 8802 output transformers or a reference to someone else measurements? Reason for asking is because if you reference the big 8802 thread that Dave G. did he gets a winding ratio of about 50:1 which equated to a 10K load into 4 ohms. I have verified these measurements to be accurate via 3 8802 chassis that have passed through here.

Also, referencing Dave G's 8802 thread (as well as my own measurements) he gets 305V to the center tap of each output transformer when each tube is drawing 40ma of current. This is with SS rectification. Even if you use tube types that draw less current, there is only so much B+ that you can squeeze out of the little 8800 series pt.

Those are my measurements of the OTs on one of my Maggie 8802s. I have a couple of others but haven't measured them.

I fed 10v into the OT secondaries and measured 439v on one primary and 444v on the other. So ratios are 43.9:1 and 44.4:1. With 4 ohm loads they're 7.7k and 7.88k. With 8 ohm loads they're 15.4k and 15.77k.

Hopefully I did the math correctly.

I've seen Dave's thread but I didn't read it closely, or completely, and it's been a while. I wonder if he may have used the second 6.3v heater winding as a bucker. If so, the B+ would be a bit higher without it.

Both the tubes I'm considering have mostly the same characteristics. The difference is that they have different heater voltages and the 4P1L is directly heated and 12P17L is indirectly heated.

If I use the 12P17L, I plan on connecting the two heater windings in series. If there's enough B+ available with tube rectification, I'll use a 12BW4. If not, I'll go with diodes.

If I use the 4P1L they will need separate filament supplies and I can use one 6.3v winding to heat the other tubes. If I need more B+, the second 6.3v winding could be connected as a booster, I suppose. If that's still not enough, I could go with diodes and run both heater windings as boosters. Lots of options.

You're certainly correct, though, that 8802 PT isn't particularly large and I am concerned about its limits. If necessary, I also have a Magnavox 185 chassis that could be used. It has a somewhat larger PT and its OTs measure ~13k into 8 ohms.