Responses to "How an Amp's Output Section Really Works"

[jjasilli]

(NOTE: Topic split to maintain continuity of the source thread. I'll answer some questions here, but a lot of them will probably be answered as the follow-on posts appear. --HBP)


Per the tube charts (e.g., http://www.mif.pg.gda.pl/homepages/frank/sheets/135/6/6V6GTA.pdf) at the approx. value of your example for one tube the plate dissipation is about 11 watts (315V X 35mA), while output is 5.5W.  Thus for 2 tubes, each would dissipate 11 watts, while output would be 5.5 + 5.5 = 11W (minus circuit losses).  This seems to contradict the first sentence of the last post. 


Of interest, the OT should be rated for the watts pulled through it and dissipated by the 2 power tubes = 22W, not the output power of 11W.  Hence the 15W tranny is underrated, except that a Hammond can probably handle 50% more than its rated current, which just about works-out. 


But somehow, with two tubes in Class A1 into an 8000 Ohm load, we get about 20W diss & 14W power output.  http://www.mif.pg.gda.pl/homepages/frank/sheets/049/6/6V6GTA.pdf

[SILVERGUN]

You must have known I was about to order this book...
http://www.ampbooks.com/home/books/power-amps/

I'm gonna have to print this out and read it all 5 or 6 times before I can take the poll....but my pre- answer is YES -- everything you take the time to type helps further my understanding in one way or another.

THANKS for taking the time!

[Stankfut]

Thanks for this information! Like Silvergun, I need to read this a few times, but that's because I'm a little dense, and I do have questions, but not about the material, it is making me see things a little differently. I do like your style of presentation, it doesn't seem to be as hard to get my head around the math.....and I actually see how you used the load lines (always a mystery to me ). I would love to see more "articles" like this!!!!!

[HotBluePlates]

Per the tube charts ...

Of interest, the OT should be rated for ...

But somehow, with two tubes in Class A1 into ...

I'll have to ask you for patience; we're not even done with determining the conditions for a single complete Class A output stage. My intent is to cover Class AB as well, and show why the numbers get all confused (but don't have to).

You must have known I was about to order this book...

And you still should. You'll probably get the answer faster.

I'm doing posts here & there as I can, when this stuff would be a book chapter. I already have a full-time job plus have sole-custody of a 3-yr old, so you can imagine my time is a little sparse. 

... I actually see how you used the load lines (always a mystery to me ). ...

Thanks! We have even presented an actual load line yet, but we'll get there. (Preview: You get less power output and/or more distortion with "non-ideal devices.")

[Jack_Hester]

You must have known I was about to order this book...
http://www.ampbooks.com/home/books/power-amps/

I'm gonna have to print this out and read it all 5 or 6 times before I can take the poll....but my pre- answer is YES -- everything you take the time to type helps further my understanding in one way or another.

THANKS for taking the time!

May be a better price here:

http://www.amazon.com/Richard-Kuehnel/e/B002M91X94/ref=dp_byline_cont_book_1

Jack

[HotBluePlates]

Thanks to Sluckey! I didn't think of locking the topic until I was done with the section at hand...

[printer2]

Of interest, the OT should be rated for the watts pulled through it and dissipated by the 2 power tubes = 22W, not the output power of 11W.  Hence the 15W tranny is underrated, except that a Hammond can probably handle 50% more than its rated current, which just about works-out. 


But somehow, with two tubes in Class A1 into an 8000 Ohm load, we get about 20W diss & 14W power output.  http://www.mif.pg.gda.pl/homepages/frank/sheets/049/6/6V6GTA.pdf

The 15W ratting is fine for the amplifier, no need for 22W. We care about how much power can be transferred from the primary to the secondary at the lowest frequency that we want to reproduce. In a P-P amp the current through each leg at idle is theoretically the same and since they are in opposite directions the net result is a core that is not magnetized. When our signal unbalances the current in the legs we get a transfer of power from the primary to the secondary. So we need a transformer rated for that power level, not the amount of power dissipated by the tubes.

But somehow, with two tubes in Class A1 into an 8000 Ohm load, we get about 20W diss & 14W power output.  http://www.mif.pg.gda.pl/homepages/frank/sheets/049/6/6V6GTA.pdf

The power outputs are given at a specific distortion. In SE we have a lower output at the given distortion, in P-P since the distortion cancels we can push the tubes a little harder.

[HotBluePlates]

Per the tube charts (e.g., 6V6GTA) at the approx. value of your example for one tube the plate dissipation is about 11 watts (315V X 35mA), while output is 5.5W.  Thus for 2 tubes, each would dissipate 11 watts, while output would be 5.5 + 5.5 = 11W (minus circuit losses).  This seems to contradict the first sentence of the last post.

The sentences in question were:

Can we operate this output stage [with 15w output] in Class A? Recall that the theoretical maximum efficiency of Class A is 50%, and we have 2x 12w tubes, so it seems the maximum output power to stay in Class A would be 12w. The answer appears to be "No."

To paraphrase, I said, "could we have two 12w tubes, with a combined dissipation of 24w, output 15w of audio?" Your example confirmed that a 6V6 with idle dissipation of about 11w could output half that at 5.5w. At 12% distortion, which is unusually high for a data sheet entry.

But plate voltage, screen voltage, and load impedance are all different from the example, so it really provides no meaningful information as it stands.

It's worth pointing out now that the reason I started from a selected output transformer and fixed primary impedance, rather than arbitrarily drawing loadlines, is that available transformers come in a limited range of values. There are only so many core sizes (and so but a few power levels) and so many primary impedances when the specified speaker loads are attached. So it really makes the most sense to start with the OT specs, and determine what kind of power supply you'll need to fully utilize the OT (or as my 10w Class Example showed, you can do less-than-full utilization if you have a strong reason).

Of interest, the OT should be rated for the watts pulled through it and dissipated by the 2 power tubes = 22W, not the output power of 11W.

Why?

>"the OT should be rated for the watts pulled through it"

Yes, in the form of audio. The d.c. is irrelevant in a push-pull setup, which is what I'm showing. The idle d.c. is important in a single-ended setup, but that is specified as the transformer's allowable unbalanced d.c. (see Hammond's 125-series info). The supply voltage is irrelevant to the OT, unless you're in the several-kV range.

>the OT should be rated ... dissipated by the 2 power tubes

If the idle current is balanced, the OT doesn't care if you pull 1mA or 300mA, so long as it's not so much the wire in the OT melts. When it comes to power, the OT only cares about the a.c. voltage and current, which is the audio power transferred to the speaker. The transformer's power rating is related to the size of its core, and the rating is about having a core big enough to transfer the power.

If you could have a pair of 6L6GC's idling at 100% (30w each for 60w total) and outputing only 15w of audio total, the transformer need only be rated for 15w.

And many transformers in classic amps had a limited bandwidth for their power rating, or gave full-audio bandwidth at a much-reduced power. That's why modern OT's sometimes are more "hi-fi" sounding, because they have full-audio bandwidth at their full power rating, and give lower lows and higher highs.

But somehow, with two tubes in Class A1 into an 8000 Ohm load, we get about 20W diss & 14W power output.  6V6GTA

You're missing what they're doing. That's a pair of 14w tubes (to use the rating on that sheet) outputting 14w of audio, or 50% efficiency. And they had to work at it if the tubes stay in class A. They chose to use less idle current (35mA per tube) and higher supply voltage (285v) with the same load to land at 14w. There's obviously some distortion when total output stage current rises by 31% from idle to full output. So the engineer who came up with that condition worked the same load, but into a point in the tube's curves that were bent to get maximum power output while minimizing THD (probably by maximizing 2nd harmonic, which would then be cancelled in the push-pull OT).

The thread will eventually get to how real tubes depart from the assumed ideal devices, but you gotta crawl before you can run.

Right now I'm presenting a basic design approach, using straightforward math, and assuming ideal devices (with the exception of required minimum plate voltage; we can get into tube limitations later). I invite you to do the same, presenting your start-to-finish output stage design while not copying a published circuit or data sheet condition.

[Ed_Chambley]

HBPThe comment made about Marshall using the same OT for the use of EL34/KT66 and 6L6 is new to me.  Also the generalization of the 4ohm when 6L6 is used.  I have never heard of any of this and it is new.


Are you saying Marshall used the same OT for KT66 and EL34, or that others do it?  If I read the KT66 data correctly, using the same OT for this and EL34's doesn't seem like something Marshall would do.  I know many people swap the EL34 and KT66, but I never would have thought they did so without a replacement OT.


My question is if Marshall did indeed use the same wind ratio for both tubes, what would be the purpose?  Or am I simply reading the data wrong for the KT66?  Generally, in class AB wouldn't you be looking for a reflected impedance of 8K for max Pout of 30w, or is that why the JTM45 is considered to be 36watts as the design conditions of the 4ohm reflected with around 430 plate?


This is where I get confused and why I am asking.  When designing for Pout can some or all tubes be used under different conditions to achieve characteristics desired without damage to tubes or the OT? 


I use the Marshall 2 tube power section as an example as it is the one I am most familiar with through experimentation.  Also the KT66 tube I really like in many different conditions.


Also, is this tube designed for different characteristics or are all tubes like this, just the data sheet does not give examples?  I know you have just begun and I will be patient as you go through this , but this is something which is really confusing.


If a KT66 is producing a Pout of 30w with all data being such to produce this wattage at 8K why cant you make a speaker load change to reduce Pout and not overload any components?  Is it not the load (speaker in this case) that created the reflected impedance?  If so, then wouldn't it be s myth that connecting to an improper load will toast your OT or tubes?


Get ready for a lot of questions from ED.  I hope this is why you are doing this.  BTW, having a hard copy of the RDH4 is much more beneficial to me.


http://www.svetlana.com/pdf/genalex/kt66-genalex.pdf

[jjasilli]

Thanks again for your article & responses.  Designing from the starting point of readily available OT's is a great approach.


I'm still having trouble with the post in the main article, and your response in this thread.   Agreed that it is impossible to specifically get 15W Class A output out of 2X tubes with a max plate diss of 12W ea.  Because output is limited to 50% of plate diss the most we can get is 6W output for ea tube X 2 = 12W.  15 is literally impossible.  But 12 is within 20% of 15.  So we can get within a reasonable tolerance of 15. 


Though the voltages, impedances, etc. of the GE & RCA tube charts do not exactly match your examples, they are all well within 10% of your stated values.  This is close enough to be considered identical; whereby it seems that they should not be dismissed as inapplicable.


The RCA chart outputs 14W when it should be limited to 12W.  Perhaps the explanation is that the RCA example is cheating by not fully operating in pure Class A???

[HotBluePlates]

The RCA chart outputs 14W when it should be limited to 12W.  Perhaps the explanation is that the RCA example is cheating by not fully operating in pure Class A???

But they're using "14w" as the plate dissipation rating. I believe they are using some tricks & experimental results to get 50% efficiency from those tubes, while staying in Class A.

Specifically, distortion. I'm only at the stage in the thread of calculating with "theoretical, ideal devices". These have zero distortion. Real tubes constrict what the "ideal device" predicted, if you are aiming for a particular minimum distortion target. But beam power tubes (like the 6V6) have low even order distortion, and push-pull cancels even harmonics. I think the data sheet condition is trying to maximize output power by setting conditions for very high 2nd harmonic, then counting on push-pull operation to cancel that. The net result is higher apparent efficiency.

Why do I think that? An ideal distortionless Class A amp idles at a certain current, a single tube peaks at double that current while the other tube is moving to zero current. So average maximum power current should be largely unchanged from total idle current. But the condition you cited idles the tubes at under 10w, and has a pretty hefty current change from idle to max output power. I think they are still running Class A (tubes never shut off) but that the current change indicates (mostly-even) distortion. The relatively low distortion % is probably due to cancellation.

All that to say it's a numbers game.

[kagliostro]

It will be nice if the tread How an Amp's Output Section Really Works


will moved to the Archives of favorite topics

Franco

[HotBluePlates]

HBPThe comment made about Marshall using the same OT for the use of EL34/KT66 and 6L6 is new to me.  Also the generalization of the 4ohm when 6L6 is used.  I have never heard of any of this and it is new.

I didn't say "Marshall used 1 transformer with many tube types." I said people get hung up about whether Marshall used 4kΩ or 6.6kΩ for KT66's; and that you could use EL34 or 6L6 or KT66 with either a 4kΩ or 6.6kΩ OT. The tube type is immaterial, as long as it doesn't overheat in use.

Don't know how you overlooked the "kilo" in "4kΩ".

Here's a Marshall OT, a Fender OT and a Vox OT. They're all close-enough to 4kΩ. The amps they're used in have EL34's, 6L6's and EL84's. What I want people to see is that you're not "matching the OT to the tube" but that the OT primary impedance and power supply dictate output power, within the limitations imposed by practical tubes.

[jjasilli]

Thanks for your responses!  Re OT rating: I now also realize that because the tube plates have resistance (or impedance) in excess of the OT, then for any given current flowing through that circuit, the OT will dissipate less watts than the plates.


Re the RCA chart: at 0 signal voltage:  285V X .070 Amps = 20W / 2 = 10W plate diss per tube.  At max signal voltage:  285V X .092 Amps = 26W / 2 = 13W plate diss per tube.  So yes they're cheating by 0.1% over max.   

[Ed_Chambley]

HBPThe comment made about Marshall using the same OT for the use of EL34/KT66 and 6L6 is new to me.  Also the generalization of the 4ohm when 6L6 is used.  I have never heard of any of this and it is new.

I didn't say "Marshall used 1 transformer with many tube types." I said people get hung up about whether Mashall used 4kΩ or 6.6kΩ for KT66's; and that you could use RL34 or 6L6 or KT66 with either a 4kΩ or 6.6kΩ OT. The tube type is immaterial, as long as it doesn't overheat in use.

Don't know how you overlooked the "kilo" in "4kΩ".

Here's a Marshall OT, a Fender OT and a Vox OT. They're all close-enough to 4kΩ. The amps they're used in have EL34's, 6L6's and EL84's. What I want people to see is that you're not "matching the OT to the tube" but that the OT primary impedance and power supply dictate output power, within the limitations imposed by practical tubes.

I get what you are saying.  The speaker load is messing me up.  I did not miss the Kilo.  In the datasheet I posted under 2 tubes Pentode connection, the 425v.  The sheet states 8Kohm.  This is what I was referring to.


I asked if you were saying Marshall used one OT and not that you said it.  The reason is the KT66 tube is often used in a much lower reflected impedance.  I have seen it in a 4kohm, pentode at 430or so volts.  I was asking why it would be used in this manner and why wouldn't it cause problems.  I guess all the other stuff I said convoluted this question.


I will just wait till you cover this and maybe I will get it.


The references to Classictone OT's do match what you were saying.  They have a OT for the AC15 which is 6kohm which is different than what people usually do to select an OT.  The normal choice is closer to 8K unless it is an older amp where I find 10k to be more traditional, however the 1959 AC15 did have a 6k4ohm which is really close to 6k.


I am missing something as it is not clicking in my brain.  I am genuine in my questions as I somewhat lost.  I am getting most of it.

[HotBluePlates]

Re OT rating: I now also realize that because the tube plates have resistance (or impedance) in excess of the OT, then for any given current flowing through that circuit, the OT will dissipate less watts than the plates.

Measure an OT's d.c. resistance. That number times current will give you the power dissipated by the OT.

The load for the tube is an impedance, which is several-kilohms for a.c. only. Same as why the choke in your power supply cuts ripple but doesn't get hot. The difference between the OT and choke is that the choke's impedance is a result of its inductance; the OT's impedance is reduced to the value quoted on the specs by the action of the transformer.

A transformer load is an entirely different thing than a load resistor in a preamp. If you're following the thread, set aside all notions other than what I'm presenting, and you'll arrive at the right place. But since I presented the OT as a resistance, when you calculate "power dissipated by the OT" you should understand that value represents power transferred to the speaker.

Because impedance is composed of resistance (the "real" part) and reactance (the "imaginary" part), and most of the OT's impedance is a result of its reactance (and transformer action), voltage and current are out of phase and there is very little "real power" (heat) dissipated. NEETS breaks some of that down (if I recall correctly).

Re the RCA chart: at 0 signal voltage:  285V X .070 Amps = 20W / 2 = 10W plate diss per tube.  At max signal voltage:  285V X .092 Amps = 26W / 2 = 13W plate diss per tube.  So yes they're cheating by 0.1% over max.   

You're not following what I'm saying when I say "cheat". The 92mA isn't a peak, or even an RMS value. It's an average value, which is a major hassle to calculate and is best found by inserting an ammeter into the shared cathode connection for the tubes when maximum output power is being dissipated into a load resistor (and/or power meter). And peak currents are always at minimum plate voltage, not at B+ voltage. You'll find the procedure for calculating average current for a push-pull stage on page 579 of RDH4.

It's also more complicated to figure out tube dissipation with an applied signal. You have to use the average current from above to calculate tube dissipation (for all tubes in the output stage) then subtract output power transferred to the load to find the actual heat dissipated by the tube.

Regardless, what did my example show for idle, maximum and minimum current? Max current was 2*Idle, and minimum current was zero. What is the average of those? (2*Idle + 0)/2 = Idle. Current increased by the same amount as it decreased, voltage increased by the same amount as it decreased and distortion was zero.

The data sheet condition has an average current that changes from 70mA to 92mA (92mA/70mA = 31.4% increase in average current). That's important to notice, but the thread is not yet ready to present why that is.

[HotBluePlates]

The speaker load is messing me up.

The speaker is immaterial at this point in the thread. There is no speaker shown in any diagram, only the OT primary impedance that results from attaching the specified load to a secondary tap.

In the datasheet I posted under 2 tubes Pentode connection, the 425v.  The sheet states 8Kohm. 

I asked if you were saying Marshall used one OT and not that you said it.  The reason is the KT66 tube is often used in a much lower reflected impedance.

It also has 5kΩ in the condition right after that one. Voltage is slightly higher but output power is up to 50w from 30w for the 8kΩ load.

You're just gonna have to sit tight to see how this stuff works. If you guys don't mentally jump ahead of what I'm presenting, it'll be a lot easier to make sense of the info. All this stuff builds a foundation; if you don't have the foundation, nothing seems to make sense.

[printer2]

You do realize your answers to the questions are just as useful as your other thread. Maybe when you are finished it this one could be tacked onto the other?

[HotBluePlates]

You do realize your answers to the questions are just as useful as your other thread. Maybe when you are finished it this one could be tacked onto the other?

We'll keep them separate. The answers are only necessary because there's some skipping ahead. 

But I hear you: nothing beats an answer to a direct question.

[John]

I vote for sticky'ing the other thread. Maybe even this one too!

[PRR]

> I was about to order this book...
Guitar Amplifier Power Amps by Richard Kuehnel
http://www.ampbooks.com/home/books/power-amps/

Get Richard's book. No single author can hammer this stuff into everybody's head. Different explanations trigger different "ah-ha" moments.

Also HBP is typing it out sorta for the first time. (He's posted most of it here and there, but just now altogether.) Which is good, and exciting. But Richard's book is several years old, and from his other books I know he goes back and cleans-up a bit for each printing, which is also good.

Also Richard retired from a good-pay job and is tending his roses and writing-up his interests. While HBP has been serving in the military and now busy in a civilian job and life. One author may be more thoughtful and another may get to the point faster. Both approaches have merit.

As for getting Kuehnel from his own house or Amazon.... I bet Richard gets more money from his own sales, but sixty bucks does seem steep. Even if it is much more than $60 of good dope (it is).

[Willabe]

No single author can hammer this stuff into everybody's head. Different explanations trigger different "ah-ha" moments.

I believe this is a very true and important statement. Just the way we are all wired.


                Brad   

[Willabe]

Also HBP is typing it out sorta for the first time. (He's posted most of it here and there, but just now altogether.) Which is good, and exciting.

Yes it is very exciting to see HBP putting it all together in 1 place!


           Brad     

[andyfromdenver]

Thanks Hot Blues Plate for the great information thread

[HotBluePlates]

Guys, I re-scoped the thread to be about Class A amps. It's getting long, and we're not out of theoretical yet (though we're about to make that leap).

I'm thinking a 2nd thread for Class AB makes sense, though it will depend on the reader already knowing all the stuff from the Class A thread.