Constant Power

[Duke LeJeune]

Just for kicks, I did a simulation of a subwoofer designed to have flat response when driven from a low output impedance. It is a Dayton RSS390HF-4 in a 5.5 cubic foot box.

That's not a woofer that I would ever use with a power paradigm amplifier.

Instead, let's look at the Faital 12FH520. This is a high efficiency 12" prosound woofer with a very good motor, smooth response to nearly 3 kHz on-axis, and inadequate bass response when driven by a voltage paradigm amp. Except for that inadequate bass response, it's a good candidate for a high performance two-way with a horn or waveguide for the top end. My modelling software assumes an amplifier with negligible output impedance, and suggested .8 cubic feet tuned to 62 Hz for a F3 of 75 Hz as being a "maximally flat" alignment. Hardly inspiring bass performance.

Here's a link to this woofer's spec sheet:

http://www.faitalpro.com/products/files/12FH520/8/12FH520_datasheet_8.pdf

I don't have the program you have, so will have to hope that you don't mind running that simulation again for this woofer. Try a six cubic foot box tuned to 35 Hz, and run it with an output impedance of 0 to 6 ohms in one ohm increments. At about 4 ohms (the output impedance of the Atma-Sphere M-60), you should see a nice flat bottom end that extends to the lower 30's.

So the Faital 12FH520, when driven by a solid state amp, is at best a midwoofer that would need help from a subwoofer, but it becomes a serious fullrange woofer when driven by one of Ralph's amps.

Do you see how approaching this woofer from the power paradigm gives us radically different results than when we approached it from the voltage paradigm? The difference came to more than one octave of bass extension. Sure we had to make the box bigger and tune it lower to take advantage of the power paradigm amp, but we still got to enjoy its dynamics and smoothness without having to buy a subwoofer.

 

[Duke LeJeune]

If one "Power Paradigm" amp's source impedance is 20 ohms and another is 1 ohm, what can you as a designer rely on for consistency?

I doubt that I could design a single speaker system to work well with that wide a range of output impedances, so I have a smaller range than that in mind when I design a speaker. I'm confident that I can do 0 to 7 ohms, if you let me pick the woofer and box size and use adjustable-length ports and (if necessary) port plugs.

Having a wide range of load tolerance is not an idiosyncrasy in my book. Eliminating unpredictable and uncontrollable variables from a system is a good thing.

It depends on whether your top priority is sound quality or virtually carefree system matching. If the latter, then solid state amps with large amounts of global negative feedback offer all you could ever ask for.

RETRACTION: It has been pointed out to me that the above paragraph is prejudicial, and I concede that point, as it assumes facts not in evidence to the satisfaction of Mr. Dressler and others in this thread.

 

[mep]

Any woofer with an FS of 50 Hz is not a woofer I would try and use in a subwoofer application. There is a reason the frequency response of this woofer is listed at starting at 50 Hz. I have built many transmission line subwoofers over the years, and the first thing I look for is what the FS of the woofer is. Even if you could somehow get this speaker to put out flat power into 30 Hz, aren't you going to have a big resonant hump at 50Hz?

 

[Duke LeJeune]

Any woofer with an FS of 50 Hz is not a woofer I would try and use in a subwoofer application.

I never said it was a subwoofer.

There is a reason the frequency response of this woofer is listed at starting at 50 Hz. I have built many transmission line subwoofers over the years, and the first thing I look for is what the FS of the woofer is. Even if you could somehow get this speaker to put out flat power into 30 Hz, aren't you going to have a big resonant hump at 50Hz?

Let me ask you this: If it is indeed possible to get this woofer to go flat into the low 30's without a resonant bump, would that change your opinion of the potential usefulness of high output impedance amplifiers?

If you have a box modelling program, let me know and I'll tell you how to run the simulation for an amplifier with a 4 ohm output impedance. That goes for anyone else who might be interested.

 

[andy_c]

I don't have the program you have, so will have to hope that you don't mind running that simulation again for this woofer. Try a six cubic foot box tuned to 35 Hz, and run it with an output impedance of 0 to 6 ohms in one ohm increments. At about 4 ohms (the output impedance of the Atma-Sphere M-60), you should see a nice flat bottom end that extends to the lower 30's.

Okay, here is the simulation as requested. Bottom curve is 0 Ohms, top is 6 Ohms. I've assumed the box Q (QL) is 7.

I don't see why one would design a speaker for substandard performance with the vast majority of available amplifiers though, just so it worked well with one unusual one.

Do you sell these amps by chance?

 

[Phelonious Ponk]

Tim,

Not at all. You can transfer electrical signals using voltage, current or power. For effective voltage transfer impedance of the source should be much smaller than that of the receiving unit, for current transfer the opposite, and for power transfer you should have similar impedance in both.

All these terms we are debating come primarily from the rules associated with these ratios. For example , people using antena's always think about the power paradigm.

Cordell main divergence with Ralph is on semantics and also mainly because he does not acknowledge the model of speaker that Ralph suggests, as he only considers the speakers that sound better (are voiced, as he says) for a constant voltage input.

I'm not trying to be argumentative, micro, but if "power paradigm" is a common way of describing the transfer of electricity, why would it not show up in a Google search? Why would a company as small as Atmasphere show up as the first relevant reference on the first page? Is this, perhaps, an uncommon way of referring to a common electrical process? And when you say you can transfer electrical signals using "power" what, exactly, does that mean? When you transfer electrical signals using voltage or current you use "power." What kind of power are you talking about?

Tim

 

[Duke LeJeune]

Okay, here is the simulation as requested. Bottom curve is 0 Ohms, top is 6 Ohms. I've assumed the box Q (QL) is 7.

I don't see why one would design a speaker for substandard performance with the vast majority of available amplifiers though, just so it worked well with one unusual one.

Thank you VERY MUCH for taking the time to run that simulation. As you can see, the curve for 4 ohms output impedance looks pretty good; the 5 ohm curve arguably even better. (Actually it's unlikely that any of these would be my target curve for a speaker, because the anticipated room effects have yet to be taken into account. My point was to illustrate a realistic scenario in which a high output impedance amp delivers better performance than a low output impedance amp.)

The only valid reason to "design a speaker for substandard performance with the vast majority of available amplifiers though, just so it worked well with one unusual one" would be if that one unusual amp had desirable qualities that you wouldn't find in that vast majority.

Just for the record, that's not what I do; I design my speakers to work with a wider variety of amps than most, and that variety includes Ralph's OTL amps. In practice, it is rare to find an amplifier with a higher output impedance than his S-30 (7.1 ohms), so that's my target at one end of the spread.

Do you sell these amps by chance?

Yes, I disclosed that in my first post back on page 2:

"I am a dealer for several amplifier lines, both tube and solid state, including Atma-Sphere. So I design my home audio speakers to work well with a wide variety of amps by keeping the impedance curve fairly high and substantially smooth..."

 

[andy_c]

The only valid reason to "design a speaker for substandard performance with the vast majority of available amplifiers though, just so it worked well with one unusual one" would be if that one unusual amp had desirable qualities that you wouldn't find in that vast majority.

Just for the record, that's not what I do; I design my speakers to work with a wider variety of amps than most, and that variety includes Ralph's OTL amps. In practice, it is rare to find an amplifier with a higher output impedance than his S-30 (7.1 ohms), so that's my target at one end of the spread.

It's clear from the graph above that you've optimized the performance for amps having an output impedance of 4-5 Ohms (higher than most SETs even) at the expense of degraded performance with amps having a low output impedance (the vast majority of available amps). You've done this by choosing a driver whose QTS is far too low for optimum operation in a vented box when driven from an amp with high damping factor.

 

[amirm]

Good discussion guys .

 

[FrantzM]

Methinks that so far, Duke is making the case for an active speaker. One in which drivers and amplifiers are matched ... Still the reference to the term "power Paradigm" seems to me forced. Active speakers , I repeat makes the point of closely matching drivers, crossovers and amps. Active speakers. I still don't see what that changes in crossover behavior in the vast majority of cases...
Googling "Power Paradigm" doesn't come up with many results, I am afraid ... Very arcane notion ... Still don't get it ... Especially when said amplifiers that are designed under this paradigm don't seem to have constant power under all loads .... Maybe updated measurements would give us more to ponder and understand.

 

[Duke LeJeune]

It's clear from the graph above that you've optimized the performance for amps having an output impedance of 4-5 Ohms (higher than most SETs even) at the expense of degraded performance with amps having a low output impedance (the vast majority of available amps). You've done this by choosing a driver whose QTS is far too low for optimum operation in a vented box when driven from an amp with high damping factor.

Sorry but I'm not sure what your point is. Yes I chose a driver that illustrated my point, my point being that OTL amps and suitable speakers can deliver good bass. Thanks for helping me by running the simulation.

Obviously I am not suggesting that either that speaker or that amp is a universal solution.

But just look at the synergy! Correct me if I'm wrong, but I'm guessing that combination modelled far better than you would have expected based on your understanding of either high output impedance amps or low Qts woofers.

(How about you, mep? See any signs of that "big resonant hump at 50Hz"??)

I could expand on my point and show that an OTL amp + optimized speaker will, for a given box size and efficiency, give you deeper and smoother bass extension than you'd get from the same size box and efficiency optimized for a solid state amp. In other words, we can bend Hoffman's Iron Law to our advantage if we do it right by choosing an amplifier that delivers increased power into a duly optimized speaker's bass impedance peaks (like in my example above). Better performance at the expense of near-universal compatibility might not interest you, but I bet it would interest some people.

And just for the record, the raison d'etre for OTL and SET amps isn't getting deeper bass out of high efficiency speakers. That's a by-product. The real reason is to diminish the types of distortion that matter, even if that results in much higher amounts of a type of distortion that doesn't.

* * * *

For those who take issue with the terminology Ralph used in his paper, if we are going to differentiate between amplifiers whose power (wattage) varies inversely with the speaker load (but tracks the voltage closely) vs those whose power (wattage) doesn't change very much with the speaker load (and therefore clearly does not track the voltage), what terms would you suggest? Whether you like Ralph's terminology or not, he has clearly identified and delineated two major classifications of amplifier based on their behavior, and likewise identified two major classifications of loudspeakers based on what type of amplifier they work best with. SO, regardless of what type of amp you like, or what type of speaker you like, you definitely want a good pairing instead of a bad one, right? Ralph's paper gives people the knowledge to avoid making an obviously bad pairing by telling them why it would be a bad pairing. How many audiophiles know that Lowthers sound good with SET amps and lousy with chip amps, but can't begin to tell you why?

 

[DaveyF]

One question for mep, do you believe that you have more experience and knowledge of amp design than Ralph Karsten? Because, IF you don't, what is the point of this thread?

 

[Duke LeJeune]

Methinks that so far, Duke is making the case for an active speaker. One in which drivers and amplifiers are matched.

Well I was hoping to make a case for matching speaker systems with appropriate amplifiers; going active would be matching individual drivers with dedicated amplifiers. Nothing wrong with that if it serves your needs.

I still don't see what that changes in crossover behavior in the vast majority of cases...

I'm guessing that you don't see how a power paradigm amp changes the crossover's behavior vs a voltage paradigm amp. If that's not it, skip on to the next post, because this is gonna be really boring!

Okay imagine you have an 8-ohm nominal speaker whose impedance curve has a 16 ohm peak in the 3 kHz crossover region. That peak is an artifact of the crossover. And at a level where an approximately constant-voltage (or "voltage paradigm") amplifier is delivering 1 watt into the speaker's nominal 8 ohm impedance, it is delivering only 1/2 watt into that 16-ohm peak in the crossover region. Assuming the designer intends for you to use a voltage paradigm amp, he voices the speaker to sound correctly balanced with that 1/2 watt into the 16-ohm peak.

Now what happens if we try to drive this speaker with one of Ralph's "power paradigm" amps (it's closer to constant power than it is to constant voltage or constant current)? The amp delivers approximately the same amount of power into that 16 ohm peak as into the rest of the speaker's nominal 8 ohm load, which is twice as much wattage as the designer intended, and the result is a 3 dB peak in the crossover region! Now that's not the end of the world as far as peaks go, but it is right smack in a region where the ear is especially sensitive. So the speaker will sound forward, maybe even a little edgy. We hear that, it wasn't there with the solid state amp, and so we mistakenly blame Ralph's amp for the problem, when the real culprit was poor amplifier-speaker matching (which we would have known about if we'd read Ralph's paper).

Now what if the designer designed a speaker for use with power paradigm amps, but it was measured on a voltage paradigm amp? Well, its response might look awful because the amplifier-speaker matching is all wrong. Here's an example of exactly that situation:

http://www.soundstagemagazine.com/measurements/silverline_panatella_ii/

First thing you notice is a huge hole in the frequency response in the 3 kHz crossover region. Okay, now scroll down to the impedance curve. And there we see a huge (30 ohm) peak centered on the 3 kHz crossover region! If we drive this speaker with a power paradigm amp, that big hole in the frequency response will fill in by 6 dB! Not enough to completely fill it in, but I think the designer wanted the speaker to be a bit laid back in the lower treble region anyway (makes for a very relaxing and forgiving presentation). So in this case, a power paradigm amp significantly improves the speaker's frequency response.

So, if we're going to design a crossover, and we recognize that we're probably going to have an impedance peak in the crossover region (most speakers do), we should take into account how these two different amplifier types are going to interact with that impedance peak. If we're building an 85 dB 6" two-way stand-mount, we probably want to voice our speaker so that a voltage paradigm amp interacting with the impedance curve gives the frequency response we want. If we're building a 98 dB 15" + horn monster, we probably want to voice our speaker so that a power paradigm amp interacting with the impedance curve gives the frequency response we want.

If you don't know what kind of amp your speaker designer had in mind, ask him.

 

[Phelonious Ponk]

Sorry but I'm not sure what your point is. Yes I chose a driver that illustrated my point, my point being that OTL amps and suitable speakers can deliver good bass. Thanks for helping me by running the simulation.

Obviously I am not suggesting that either that speaker or that amp is a universal solution.

But just look at the synergy! Correct me if I'm wrong, but I'm guessing that combination modelled far better than you would have expected based on your understanding of either high output impedance amps or low Qts woofers.

(How about you, mep? See any signs of that "big resonant hump at 50Hz"??)

I could expand on my point and show that an OTL amp + optimized speaker will, for a given box size and efficiency, give you deeper and smoother bass extension than you'd get from the same size box and efficiency optimized for a solid state amp. In other words, we can bend Hoffman's Iron Law to our advantage if we do it right by choosing an amplifier that delivers increased power into a duly optimized speaker's bass impedance peaks (like in my example above). Better performance at the expense of near-universal compatibility might not interest you, but I bet it would interest some people.

And just for the record, the raison d'etre for OTL and SET amps isn't getting deeper bass out of high efficiency speakers. That's a by-product. The real reason is to diminish the types of distortion that matter, even if that results in much higher amounts of a type of distortion that doesn't.

* * * *

For those who take issue with the terminology Ralph used in his paper, if we are going to differentiate between amplifiers whose power (wattage) varies inversely with the speaker load (but tracks the voltage closely) vs those whose power (wattage) doesn't change very much with the speaker load (and therefore clearly does not track the voltage), what terms would you suggest? Whether you like Ralph's terminology or not, he has clearly identified and delineated two major classifications of amplifier based on their behavior, and likewise identified two major classifications of loudspeakers based on what type of amplifier they work best with. SO, regardless of what type of amp you like, or what type of speaker you like, you definitely want a good pairing instead of a bad one, right? Ralph's paper gives people the knowledge to avoid making an obviously bad pairing by telling them why it would be a bad pairing. How many audiophiles know that Lowthers sound good with SET amps and lousy with chip amps, but can't begin to tell you why?

Bold 1: Perhaps you should, because so far all you've done is provide evidence that given enough care in matching amp and speakers, one of OTL's classic weaknesses might be overcome. And by smoother, I assume you mean more linear?

Bold 2: This "real" reason for the use of OTL amps, I'm sure you've noticed, is almost universally accepted by the devotees of OTL amps, but it doesn't get any traction outside the club. While you're making your case, if you really have an interest in doing that, you may want to show some statistical evidence of which distortions matter to human hearing, at what levels, and some examples of OTL amps audibly reducing the distortions that do (once you've established that), relative to SS amps of equal quality. Otherwise, you're just another OTL fan praising that thing you like the sound of. This is an audiophile board; we don't need to be convinced that some people enjoy THD. I do need to be convinced that the presence of a bunch of it audibly reduces distortions that have been demonstrated to be more audible and objectionable to more listeners, not just anecdotally reported as such by fans. Don't get me wrong, this is a OTL discussion with a chart in it; I appreciate the progress, but we're still barely lifted above he said/he said.

I don't really have any issue with Ralph's terminology, but I think it is useful to know when "engineering" terminology is really not - not universally defined, understood or used -- but is, instead, the invention of the author and adds no credibility, substance, or even a point of reference beyond the author's position.

Tim

 

[microstrip]

I'm not trying to be argumentative, micro, but if "power paradigm" is a common way of describing the transfer of electricity, why would it not show up in a Google search? Why would a company as small as Atmasphere show up as the first relevant reference on the first page? Is this, perhaps, an uncommon way of referring to a common electrical process? And when you say you can transfer electrical signals using "power" what, exactly, does that mean? When you transfer electrical signals using voltage or current you use "power." What kind of power are you talking about?

Tim

Tim,

I will answer the way I usually answer to lazy youngsters - just because it does not show in the internet, it does not mean it does not exist. You have to go the library and dig in old texts about high impedance speakers and tubes. The older tube OTL description I have seen used an 800 ohm speaker. I was fortunate because my department library had the old US Navy tube courses and manuals, and still as a student I have spent many afternoons going through them. Unfortunately it was decades ago, and meanwhile I forgot most of it.

My apologies, but I do not have the time or will to write about electrical signal transfer basics in WBF.

And yes, you are correct, the "power paradigm" is just a way of referring to a common and simple electrical process.

 

[Phelonious Ponk]

Tim,

I will answer the way I usually answer to lazy youngsters - just because it does not show in the internet, it does not mean it does not exist. You have to go the library and dig in old texts about high impedance speakers and tubes. The older tube OTL description I have seen used an 800 ohm speaker. I was fortunate because my department library had the old US Navy tube courses and manuals, and still as a student I have spent many afternoons going through them. Unfortunately it was decades ago, and meanwhile I forgot most of it.

My apologies, but I do not have the time or will to write about electrical signal transfer basics in WBF.

And yes, you are correct, the "power paradigm" is just a way of referring to a common and simple electrical process.

You answered my question, Micro, thanks. You even answered it pretty simply and directly, allowing me the "laziness" of not having to ask it two or three different ways to get to the truth, that it is a made-up term for a common process that could much more clearly be described in more universally accepted terms, though they would sound neither proprietary nor as important.

Tim

 

[FrantzM]

Tim,

I will answer the way I usually answer to lazy youngsters - just because it does not show in the internet, it does not mean it does not exist. You have to go the library and dig in old texts about high impedance speakers and tubes. The older tube OTL description I have seen used an 800 ohm speaker. I was fortunate because my department library had the old US Navy tube courses and manuals, and still as a student I have spent many afternoons going through them. Unfortunately it was decades ago, and meanwhile I forgot most of it.

My apologies, but I do not have the time or will to write about electrical signal transfer basics in WBF.

And yes, you are correct, the "power paradigm" is just a way of referring to a common and simple electrical process.

You could have just held yourself to this

Rather than starting the post with the condescending

I will answer the way I usually answer to lazy youngsters

...

I don't think such advance the discussion .. BTW so far a case for matching speakers to amp ... Nothing new here ... No Paradigm Shift

 

[microstrip]

You answered my question, Micro, thanks. You even answered it pretty simply and directly, allowing me the "laziness" of not having to ask it two or three different ways to get to the truth, that it is a made-up term for a common process that could much more clearly be described in more universally accepted terms, though they would sound neither proprietary nor as important.

Tim

The term is very clear for any one having some technical knowledge. And the importance is surely determined by those who are interested in understanding and debating it, such as those who owned electrostatic speakers and find curious why an home made OTL using a few tubes can have much better bass response than a Krell KSA250 when using such speaker.

There are several easily objectionable parts in Ralph explanations, mostly connected with subjective appreciations. But any WBF reader knows this type of situation happens permanently in this forum.

 

[ack]

One question for mep, do you believe that you have more experience and knowledge of amp design than Ralph Karsten? Because, IF you don't, what is the point of this thread?

There are several points being made, directly or indirectly, and many others that are intentionally being eluded... That there are amplifier designs that better match certain speaker designs; that you can design speakers to follow one of the paradigms - fine, we know that. That we have yet no clear indication whether Atmasphere designs are true constant-power designs or not, and at what %THD rating - especially in real-world applications as opposed to test-bench results; it seems the case is being made that they are, but with very high THD, which is allegedly tolerable; and that the designs use no feedback - a good thing - to achieve that. Perhaps we can accept that at face value, but the arguments so far are somewhat weak (to me at least) especially when very high THD is involved. On the other hand, there are still many open questions, the most important of which for me are:

a) No one has yet come forward to name *current-day* SS amplifiers or other tube amplifiers that are not following the power paradigm; instead, vague arguments (in Ralph's paper) have relied on manufacturer power ratings that show higher power output at lower load impedances; even more vague arguments about SS amps people on the forum rave about that have low damping factors. Let's start clearly naming all of these amps.

b) No one has directly discussed amplifier output impedance and how that fits into either the so-called voltage paradigm or the the power paradigm; more importantly, what is it that makes a design follow one paradigm vs/ the other - we would like to see formulas that also involve the load itself. We don't even know what the true output impedance characteristics of Atmasphere's OTLs are.

In my mind, mep and Frantz are correctly calling out the issues they see, and I remain overall unconvinced and unimpressed myself as well.

 

[microstrip]

You could have just held yourself to this

Rather than starting the post with the condescending ...

Frantz,

I love when the most condescending poster in WBF comments my style. Just consider that I am learning from you ...