Tube Amp for Martin Logan Speakers

Zd542, I was tempted to take a shot at your question but decided not to. Ralph (Atmasphere), Al (Almarg) and several other members are EEs and will do a much better job responding.

Having said that, I will just put out there this simplistic explanation. An autoformer raises the speaker impedance presented to the amp, or perhaps it might be better said, that the device makes the amp think the speaker impedance is higher than it actually is. There's a white paper on the Autoformer web site. Just do a search.

I'll anticipate you next question which is why can't one simply insert a resister in series with the speaker. I think the answer is that it will draw off a lot of watts as heat and will make the output impedance of the amp look higher, thereby screwing up the DF. It may also affect bandwidth ... and there I stop.

Looking forward to reading a better explanation from the EEs.

Ralph, ... if a speaker has a rock and roll impedance curve that varies from 4 ohms to 20 ohms as a function of FR, would using a ZERO to double the impedance, say ranging from 8 ohms to 40 ohms, cause a Voltage Paradigm amp (i.e., a SS amp or low output impedance tube amp) to choke driving such a speaker?

Ralph, George and Mapman .... I have no dog in this fight, but I have read the several posts and would like to suggest that much of what everyone has written has merit. Let me suggest to my fellow hobbyists that you aren't really disagreeing with each other, but you are talking past each other and just getting frustrated.

I think Ralph would agree, and has said as much in other threads, that a high'ish output impedance amp might not mesh well with a low'ish sensitivity speaker that has a roller coaster impedance curve. Indeed, one could expect sonic colorations as a result of the interaction between the two components behaving in according with various electrical principles, e.g., Ohm's Law.

That may be why Ralph says that an autoformer might aleviate some aspects of the "non-optimal" (whatever that means) impedance match in such cases. George, I gather your point is that sticking an autoformer into the circuit is just a "patch" for an underlying issue that maybe shouldn't be an issue in the first place.

If that is a fair restatement of your view, I too happen to be biased (pun intended) against sticking gizmos into the circuit based on a few anecdotal experiences of my own. In the two or three cases I put "artifacts" into the signal path, I had dismal outcomes. I am not saying one would have a similar result with autoformers. Just that I am dubious based on past experiences. And even if they work, I am not sure I would want to shoe-horn an amp and speaker together if the two components weren't a good fit in the first place. But that's just me.

If we can all agree that I fairly restated the dilemna, let me suggest that perhaps one way to view OTL amps is kinda like SETs. That is there are audiophiles who swear by SET amps. But, ... they recognized as a threshold matter that a SET amp will not be a good match with a low sensitivity speaker that has a roller coaster impedance curve. That's why some have suggested matching a speaker like a Daedalus with this type of amp. I suspect that an OTL would perform beautifully being matched with a Daedalus speaker.

Hey ... to a lesser extent, I've got amp/speaker issues too. My ARC Ref 150 amp is driving Paradigm Signature 8 v3 speakers. This is **not** a match made in heaven. The S8s have wild impedance and phase angle curves. Paradigm advises that the S8s should be driven by a high-power/high current SS amp. Oooopss. I blundered. But not that much. I've found some sonic solitude by using the 4 ohm taps (as of late), which seem to have smoothed out the rocky impedance road and improved bass response. Having a 1040 joule power supply "don't" [sic] hurt either.

In short, if everyone takes a minute to re-read the other folks' posts, I think you'll wind-up agreeing with much of what has been written.

IMHO.

Btw, I'm rocking with the Best of Pat Benatar CD. Having a ball!! :) :)

Ralph, as I have come to better appreciate, there are very few if any perfect solutions to audio technical issues. It seems that EE/designer folks like yourself make judgments and compromises to achieve a certain balanced product that has market value within its market niche.

For that reason, I think that it is imprudent to think in absolutes. Having said that, I believe that if one is looking to invest significant sums into their rig, they can avoid obvious pitfalls and clear mismatches, for example, matching a low powered SET amp or a high'ish output impedance tube amp with a low sensitivity rock n' roll impedance curve speaker, like my Paradigm Signature 8s.

(It's only through dumb luck that my Ref 150 can do a decent job driving these beasts. Should have picked another speaker, or kept the speakers and opted for a top drawer SS amp like a Pass or Ayre.)

So, I share your view. We are adults and should be polite, albeit a little tongue-in-cheek humor from time-to-time is ok by me. :)

By the way, I thought the primary focus of your comment about using ZEROs was with respect to matching one of your OTL amps with ESLs like Martin Logans which have extremely low impedance specs at high frequencies. Quite honestly, if I loved ML ESLs, I would have some agita matching that transducer with pretty much any amp, SS or tube, unless the amp was rock solid stable and wouldn't distort driving juice into a .5 ohm load.

Btw, btw, after having read your posts again, I changed the output taps on my Ref 150 from 8 ohms to 4 ohms ... again. Although gain was clearly reduced because I was using the 4 ohm taps (about 2.5 db less than the 8 ohm taps), I think it quite possible that the amp was producing cleaner power (i.e., less distortion).

Bass is clearly more extended and tighter. No surprise there since the output impedance off the 4 ohm taps is less than the 8 ohm taps (therefore higher DF). Plus, speaker impedance in the "power zone" (say 60 Hz to 700 Hz) is for the most part 4 to 5 ohms), ergo a better impedance match for the amp where it is being called upon to deliver most of its juice. Imaging was less forward (probably because midrange/treble was less emphasized as a result of higher impedance), so I turned up the gain and imaging came back. In short the rig plays louder, but not as harsh (i.e., maybe less distortion).

Regards to all,

Bruce

P.S. Back to audio business, I just bought a JVC hi-rez redbook CD of Tango music from Acoustic Sounds. I'll report back with comments. I'm trying to keep track of good hi-rez CD label. So far, +1 for MoFi.

Unsound, you state above "that [if] one can get greater sound pressure into a 4 Ohm load with a ss amp, then we can use that extrapolation to suggest that as frequency response is measured in dBs, one can extend bass response and therefore extend frequency response." With respect, IMO, your statement may be a non sequitor.

Here's my thought process. A speaker's SPL is a function of its sensitivity and the amount of watts the amp is driving into the load. While it is true that a 4 ohm load will draw twice as much current as an 8 ohm load, I fail to see why that necessarily means the speaker's SPL is necessarily twice as high at 4 ohms.

Consider my speaker, the Paradigm 8s. These beasts have a impedance curve that looks like a roller coaster, ranging from a low of 4 ohms in the bass/low midrange FR range and a high of 20+ ohms at the midrange/tweeter x-over point of 2K Hz. Yet ... the S8s have a very flat FR.

How can that be, I rhetorically ask? The reason is that these speakers were designed to be driven by a low impedance/high current SS amp. The fact that impedance rises to 20+ ohms at 2K Hz, thereby resulting in the speakers drawing LESS current and power from the amp, only means that the drivers are very sensitive at that frequency. Otherwise, the speaker's measured FR would be grossly skewed.

Consider Maggies. I recall reading that the 1.7s have a flat impedance function of 4 ohms or thereabouts, yet are not sensitive. As a consequence, one needs a beast of a SS amp to drive these pups to decent SPLs. In fact, I also recall reading a review of my Ref 150 that mentions the reviewer hooked up my Ref 150 amp to Maggie 1.7s and found the SPL just adequate.

As far as extended FR is concerned, I surmise that has more to do with the speaker design than it's impedance spec as a function of FR. I suspect what may have a more perceived impact on low end bass response is the amp's DF and the speaker's impedance spec. As Ralph, Al and others have mentioned elsewhere, DF is a factor equal to the quotient of a speaker's impedance at a particular frequency, divided by the amp's output impedance. The output impedance of most SS amps is so low that the amp's DF is not really a relevant stat.

But, if the amp's output impedance is "low'ish," say .5 ohms, and the speaker's impedance is "only" 4 ohms, ironically the DF will only be 8. By contrast, if the speaker's impedance is 8 ohms, then the DF jumps to 16. I surmise that the doubling of the DF may (??) have the effect if making bass sound tighter and perhaps more extended sounding.

I think Ralph was probably on target when he said that if a speaker presented a flat impedance load as a function of FR, say 8 ohms or higher, and zero phase angle over the speaker's full FR bandwidth, I think pretty much any SS or tube amp would be a very happy camper.

But as far as a speaker's FR being flat, extended or whatever, I think that has more to do with the speaker's (i) design objectives and build, (ii) impedance variations, (iii) phase angle stats, and (iv) SPL sensitivity. Of course, also critical is whether the designer intended that the speaker be driven by a SS or tube amp.

Really sorry for the long post. I hope it's cogent and the reasoning linear.

I'm sure I mixed and matched concepts, but hopefully our tech members will correct my meandering musings.

Regards,

Bruce

The title of this thread is "Tube Amp for Martin Logan Speakers." The lst couple of posts give me considerable pause. Let's go back to square one basics.

Seems to me the first question that comes to mind is what did the ESL designers have in mind? That is did they design these speakers to be driven by a low output impedance amp (e.g., SS) or a high impedance tube amp. Regardless of whether the amp in question can drive these beasts without shutting down, with or without ZEROs, the impedance plots are so wacko that matching the wrong type of amp may likely cause considerable sonic colorations.

So, does anyone definitively **know** the answer to my question? What did the designers intend?

Folks, some may find the Stereophile article about "EPDR" quite interesting at this link:

http://www.stereophile.com/reference/707heavy/index.html

I think one important take-a-way is that speaker impedance ratings, by themselves, do not take into account the full picture of speaker's load difficulty. Other factors include a speaker's phase angle characteristics. Take a look at the Final 600i ESL. Talk about an amp ball-breaker.

But perhaps even more relevant to this thread and Ralph's comments is that most amps have a real world "SOA" re power dissipation that is a lot less than one might expect. As defined in the article, the acronym SOA stands for Safe Operating Area. Please note that an amp's SOA power dissipation contracts considerably as impedance decreases and negative phase angles increase in magnitude.

I'll let our techy members comment further, but suffice to say that low impedance loads can cause an amp to operate at or outside its SOA, thereby causing distortion. I surmise that if a particular brand amp's SOA is small when driving a low impedance load, the answer is to buy a more powerful amp that has a less restrictive SOA or try ZEROs.

Final two points. First, the article speaks about SS amps. Nonetheless, I surmise that similar principles apply to tube amps. I understand the SS amps hard clip while tube amps generally soft clip. But I think common to both types of amp is that distortion dramtically increases once the amp exceeds its SOA. The second point is that when Ralph uses the term "happy amp" he no doubt is referring to an amp that is operating within its SOA.

My heads hurts because I have no idea what I just wrote.

Goodnight.

Well stated Mapman! :)

I think the EPDR link I provided supports Ralph's comments that low impedance speakers may have an even lower EPDR rating than what is nominally reported or measured. Even more important is that speakers presenting low EPDR factors significantly contract an amp's so-called SOA.

I am not saying that 4 ohm speakers are not good performers. Just that a nominal 4 ohm load in the bass region may really squeeze the amp and possibly result in a referral to the SPCA. At a minimum, a less than optimal amp/speaker match may, as the article explains, result in distortion. So, I happen to think Ralph's comments make sense for the reasons set forth in the attached article.

Bruce

P.S. The SPCA is the Society for the Prevention of Cruelty to Amplifers.

George, not having ever used the ZEROS, I can't speak to their pros or cons. Having said that, as I mentioned in one of my posts above, I tried a couple of gizmo tweaks in the signal path in various places and was disappointed. So, I suppose based on those experiences, I would be biased (pun) against inserting **any** artifact in the signal path .... unless my back was up against the wall.

For that reason, I would shy away from using ESL speakers with wacko impedance and phase angle curves, even if ZEROS would enable my amp to drive them. At one point, I was wondering whether picking up a pair of QUAD 2805s would be worthwhile. After doing some reading about their impedance curve, I decided against it.

Hey ... there's a lot of amp ball-busting speakers out there. The impedance and phase angle specs of some "fav" dynamic speakers are killers. So I happen to concur with the parts of your last post that counsels caution when matching amps and speakers.

As I also said above, it is only through dumb luck that my ARC Ref 150 seems (??) able to do a good job with my Paradigm S8s. Talk about amp ball-busters. I think the reason the combo works is because (i) I use a self powered sub woofer to pick up some of the low end load, (ii) the S8s are spec'ed at 92 db sensitivity, (iii) the Ref 150 is rated at 150 wpc, and (iv) the Ref 150 can presumably deliver serious current because it has a beefed up capacitive power supply of 1040 joules.

Btw, I currently use my amp's 4 ohm output taps. Output impedance is probably on the order of .5 ohms or so. If correct, the "actual" damping factor is probably 8 or 10'ish in the S8's bass region (60 Hz to 500+ Hz) because the S8's spec'ed impedance is about 4 or 5 ohms in that frequency corridor. That is clearly better than if my amp had an output impedance of 4 ohms, in which case the DF would be about ONE (1) .... not so good.

Finally, I think there is more we agree about than not. Hopefully, we'll all read each other's comments more carefully to avoid talking past each other.

Cheers,

Bruce

Thanks George. Very interesting plots for the Quads and MLs.

Looking at the Quads, I see the bass impedance spike is literally off the chart. Heck, it's almost an open circuit. I don't know, .... schlepping around a 75 amp to a dealer ... auugh. I ain't gettin' the warm and fuzzies.

The MLs have one of the flatest phase angle plots I have ever seen. Even still, I would be concerned that my amp might be challenged driving the MLs because its impedance is incredibly low in the treble region, almost a short.

So here, if I blundered into a combo involving the MLs and the Ref 150 and my back was up against the wall, ZEROs might help. So ... if impedance was doubled, the bass region impedance wouldn't be outrageously high. But jeez, impedance in the treble region is about 1 ohm. Even if doubled, we're still talking about 2 ohms. That's pretty low.

I suppose the only way to know for sure is to try it out and see what happens. But why would I want to drop a ton of bucks on such an uncertain proposition, even with ZEROs.

I gotta tell ya George, ... unless I read a lot of testimonials on A'gon declaring the Ref 150 and one of these ESLs is a match made in heaven, I would be very skeptical. Even if there were a number of approbations, I'd still call ARC and the ESL manufacturers to get a bead from the tech folks.

It kinda raises a different question -- what ARE folks using to drive this ESL speakers? A hydro-electric plant??

Thanks,

Bruce

Well that's kind of an important question Mapman because given the wacko impedance plots of these ESLs, hooking them up to a high'ish impedance tube amp could really color their sonic presentation if indeed these speakers were designed to be driven by an amp whose output impedance is measureable in decimals points.

And while in such cases ZEROs may "trick" a tube amp into thinking it's not driving into a short, the sonic presentation will still be skewed because the ZEROs will not flatten or "normalize" the ESLs impedance plot.

Ralph and others, with respect, do you know for a **fact** that MLs or Quads were **designed** to be driven by a Constant Voltage Paradigm (SS) amp or a Power Paradigm (traditional high'ish output impedance tube) amp?

Seems to me that just because they work "ok" with one type amp or another, with or without ZEROs, doesn't answer the threshold of whether the sonic presentation of these ESL speakers will be colored if driven by the one type of amp or another.

Or ... are you saying the relationship between an amp's output impedance and an ESL's varying impedance plot is simply not a relevant consideration to the ESL's sonic output because of some unique or different electrical characteristic peculiar to ESLs??

Thanks

George, again with respect, all your last post tells me is that the Levinson amp is a Constant Voltage Paradigm amp which doubles watts as ohms halves. Your post also states that the Levinson can deliver power into a near short, .5 ohms.

But that doesn't answer my question. I'll restate it -- did the designers of ML and QUAD ESLs design these speakers with the expectation that they would be driven by a SS - Voltage Paradign or tube Power Paradigm amp?? Put aside the difficulty of the load and the amp's ability to drive the beast.

Just simply what did the designers have in mind -- low output impedance (say less than .1 ohms) SS or "high'ish" output impedance (say 3 or 4 ohms or more) tube?? Once we get this answewr, then we can start to respond to the OP's Q.

Thanks for the Stereophile URL link Unsound. My tongue-in-cheek quip was dead-on: the MLs should be driven by a hydroelectric plant - actually 2; one for the ESL panels and the other for the woofer. :)

Tom, if you read my posts, I framed my question by referring to those tube amps having a "high'ish" output impedance (say 3 or 4, or more ohms).

By contrast, some amps like my ARC Ref 150, have a "low'ish" output impedance, possibly because they use negative feedback. Specifically, the output impedances off my amp's 4 and 8 ohms taps is about .5'ish and .6'ish ohms, respectively. As a consequence, the Ref 150's output voltage regulation is pretty tight: about +/- .4 db and +/- .8 db off the 4 and 8 ohm taps, respectively. So in effect, tube amps that have a low'ish output impedance function somewhat like a low impedance SS amp -- constant voltage source.

I mention all of this because you say in your last post that "ML approves of using tube amps with their speakers." Well ... that may be. But then there are tube amps, ... and then there are tube amps.

Regards,

Bruce

Quick postscript: my posts are not just academic. The OP is asking about matching a tube amp with ML ESLs. My comments are trying to focus attention on whether the ML ESLs in question should be driven by a low impedance amp that acts as a constant voltage source, i.e., a typical SS amp or a low impedance tube amp.

As I said above, "there are tube amps, ... and then there are tube amps." My somewhat tongue-in-cheek point is that if the ML ESLs were designed to be driven by a low impedance amp, then the OP should be cognizant about the output impedance of the tube amp he has in mind. The consequence of using a "high'ish" output impedance tube amp like an Atmasphere OTL is that the sonic colorations may result in augmented bass and shelved treble. Of course ... even a low impedance tube amp may need some serious current capabilities, even if it performs SS-like. Consider the EPDR article and its reference to SOA.

By the way, as I also mentioned, while ZEROs may raise the apparent across the board impedance load presented to the amp (SS or tube), I am dubious that the device will smooth out the sonic presenation of the MLs if the wrong type of amp is used. Perhaps Ralph or one of the other tech members can speak to how much SPL variation will result if the "wrong" type of amp is used. Perhaps, for discussion purposes, we should assume the MLs were designed to be driven by a SS "constant voltage paradigm" amp and the "wrong" type of amp is a Power Paradigm amp.

I surmise that if the tech members respond, we will all gain a better understanding of whether one should drive MLs with a Power Paradigm tube amp if they were voiced to be driven by a SS/Constant Voltage Paradigm amp.

Thanks Al .... that's exactly what I surmised the directionality of the impact might be. Btw, if ZEROs are used to double the ML's apparent impedance plot, would that change your calculations?? And your comment about whether any amp, SS or tube, can drive the MLs within its respective SOA (see EPDR article) is a whole "nuther" kettle of fish.

Btw, did the ML's phase angle plots factor into your calculation? If so, directionally, how so (if phase angle is negative or positve)? I intuit that harsh phase angles can constrict the SOA of an amp (SS or tube). I am not clear of the impact on sonic coloration, assuming the amp is operating within its respective SOA.

Incidentally, to my sorry old ears, I find a sonic coloration of even 1.6 db to be noticeable. I base that assertion on my amp/speaker combo.

Based on John Atkison's review of my amp, he reported that the output impedance off the 8 ohm taps is about 1'ish ohms, resulting in output voltage regulation variation of +/- .8 db (ergo a theoretical spread of 1.6 db). By contrast, those figures are roughly halved off the 4 ohm taps.

In practice, I find that my speakers sound less forward and have deeper, more extended bass when driven off the 4 ohm taps. That is consistent with the impedance plot of my S8s, which have roughly a 20 ohm peak at 2K Hz (x-over point) as compared to a 4 ohm saddle in the "power range" of 60 to 500 Hz).

Thanks again Al. You manage to lift the veil of confusion for non-techs folks like me.

Regards,

Bruce

That's right George. My ARC Ref 150 might (???) be able to handle the current demands because it has a beefed up power supply -- 1040 joules. And, as I mentioned above, the output impedance off the 4 ohm taps is roughly .5 or so ohms. As a result, my amp performs somewhat like a solid state amp.

Hopefully, Al will come back with a response to my follow up questions: (1) whether ZEROs will change his calculations; and (2) how does the phase angle spec at a particular frequency affect his calcs.

That folks is why I persisted with my comments. I can't do the math, but after 2 or 3 years of this back and forth with Al, Ralph and some of our other tech members, I better appreciate the challenge of matching amps and speakers. Also, when I see that a speaker's impedance plot is literally off the chart, e.g., the ML ESLs and some of the B&Ws, I raise an eyebrow.

So ... this goes back to what Ralph (Atmasphere) said all along. Optimally, if a speaker manufacturer could design a speaker with a flat impedance plot, say between 8 and 16 ohms, with a zero phase angle across the frequency spectrum, the SPCA would go out of business.

And once again, the SPCA stands for the Society for the Prevention of Cruelty to Amplifiers. :)

Cheers,

Bruce

Al, perhaps the EPDR article I referred to above might put a little gloss on what you said. The underlying thesis of the EPDR article is that a speaker's "nominal" impedance is further reduced if it is coupled with a negative phase angle at the particular frequency in question. So, if the "effective" impedance, or EPDR, is even lower, I can only assume that the "db delta" between a SS amp and a tube amp, even with a "low'ish" output impedance, would likely be greater. And I believe you surmise the same in your last post.

I think the take-a-way for us is that we may not be able to tell whether a particular amp/speaker combo will be a good match by looking at specs alone. But we should be able to identify (or at least be wary of) a combo that may be a bad match. Of course, if someone likes sonic coloration -- hey, all the power to sonic coloration.

Yeah ... I agree with the comments above, either the manufacturers and/or the dealers need to do a better job of helping customers understand this stuff before pushing product.

Thanks again Al.

Tom, you quote the ML manual as advising that "an ideally stable amplifier will typically be able to deliver nearly twice its rated wattage into 4 Ohms and should again increase into 2 Ohms."

My take-a-way is that ML recommends using a rock-solid stable SS/Voltage Paradigm amp, which as discussed above has a very low output impedance. Stated differently, the ML ESLs in question were likely voiced to be driven by a SS amp.

Hence, based on what Al explained, using a tube amp which has a "high'ish" output impedance will likely result in sonic coloration to some degree. As Al also said, some may not find this to be a problem and that's ok. :)

So ... after all that has been written, I think the best answer to the OP's Q is that if he/she really wants to use a tube amp to drive his/her ML ESLs, consider a low output impedance BEAST that can deliver serious current.

Frankly, I would be biased (pun) in favor of using a high quality/high power SS amp e.g., a Pass, Ayre, Bryston, inter alia.

Regards,

Bruce

Capdek, glad to read that your Manley amp does the job. Just curious, do you know what the output impedance of the amp is? Does the amp have different taps? I ask about taps because you mention it's ultra-linear.

Thanks George. I looked at the review and picked up the following concluding statement that "[t]he Manley's output impedance [1.5 ohms], while not unusually high for a tube amp, is still high compared with typical solid-state designs, and may therefore have some effect on the frequency response of the speaker-amp combination.—Thomas J. Norton"

The reason I asked the Q is because Capdek said that the Manley does a good job with his ML ESLs. I'm sure Capdek has good reason to be happy with his Manley/ML combo. But I suspect that the treble may be a tad shelved and the bass a bit augmented -- just based on what Thomas Norton said above.

Obviously, unless one measured the ML's frequency response in a controlled environment, it's difficult to know what's really going on with Capdek's rig. So, I'll leave it there. If Capdek is happy -- then that's all that matters.

This has been an interesting thread. I think all of us have leaned a lot.

Al and Ralph, I read the article that George kindly provided. I also re-read the posts at the link Al provided. I think Al correctly states that there is a bit of friction caused by the use of terminology.

I was looking for, but didn't find the post where I coined the terms "synthetic" or "apparent" constant voltage source where negative feedback (NF) is used. I think that using such terminology is useful insofar as it helps one to better appreciate whether an amp becomes a true constant voltage source if NF is used or just acts like a constant voltage source to some extent.

Let me explain. My take-away from George's article is that if an amp has an output impedance that is greater than Zero, some part of its power output is being consumed internally, resulting in a voltage drop at the output taps when driving an external load.

The author illustrated this effect with some simple examples using an external speaker load of 8 ohms and two hypothetical amps -- one having an output impedance of: (a) .1 ohms; and (2) the other having an output impedance of 2 ohms. The math then follows illustrating how the amp's output voltage and corresponding power output changes under each scenario.

I surmise that the terminology confusion comes into play because of what NF does ... and does NOT do. That is, if an amp has an output impedance of say 2 ohms, then as Ralph says, that is its output impedance ... period. But if NF is used, the amp compensates for the voltage drop by increasing output voltage to comepensate, thereby causing the amp to perform like, but not actually be, a constant voltage source amp.

As such, with output voltage adjusted to offset the voltage drop at the output taps resulting from the interaction of the amp's own internal impedance and the speaker's impedance, the amp functions as though it was a constant voltage source amp, or like one to a greater extent.

But there is always a BUTT. This picks up on what Al, Ralph and the author of the EPDR article have been saying over and over again. Using NF cannot create energy. That is ... an amp using NF will be able to generate power only to the extent it is capable of doing so based upon its internal architecture. Or, as the EPDR article said, within its SOA (safe operatwing area). If the amp operates outside its SOA, it will cut back its power output, clip, become unstable, produce distortion, etc.

I said this at the beginning of this thread and I'll repeat it again. I think we are all agreeing with each other much more than not. Butt, I think we are talking past each other.

IMO, given that **most** (not all) speaker manufacturers seem to be making speakers with roller coaster impedance and phase angle plots, erring on the side of am amp having lots of SOA headroom and low output impedance is a safe play. That generally equates to a SS beast that can deliver lots of current/power.

OTOH, as I said here and in other threads, to illustrate my understand, my ARC tube amp functions within a pretty wide operating corridor like a SS amp, but to do so, requires NF, and a massive power supply. Or using my own terminology, my amp performs like a "synthetic" SS amp, to some extent within a large "SOA."

Hope this helps ... butt it probably doesn't.

Thanks Al. My apologies for the numerous typos in my last post. I was typing with my laptop literally on my lap, drinking coffee with one hand and typing with the other hand. Truly a multi-handed effort.

That aside, I wanted to say that this has been one of the best threads to get posted in a long time. So kudos to all of the contributors.

I believe that if and when I ever change out my speakers or my amp, the terrific comments posted in this thread will make selecting a good amp/speaker combo a less haphhazard (sp) exercise. At the very least, I'll be able to ask better questions ... and know what to expect in terms of getting better answers.

Thanks again to all, including of course Ralph, Al, George, Mapman, Unsound and Judy.

Cheers,

Bruce

Jan, please re-read my non-tech take 2 or 3 posts north of this post. What you describe is exactly the effect of using NF -- the amp compensates for the output voltage drop based on the amount of NF applied to increase the output voltage to compensate for the interaction of the amp's internal impedance load and the external load of the speakers.

The effect is more precisely described as the amp is behaving as though it has low output impedance. So ... the NF circuit does not really change the amp's internal impedance. Instead, it acts like a servo to cause the amp to compensate for the Ohm's Law effects.

Of course, as Al, Ralph and others have said, the amp can only do this if it operating within its power delivery limits -- i.e., SOA (safe operating area). Also, there are rarely any free rides. Hence, NF has its own problems, e.g., increased odd ordered harmonic distortion ... and its detractors, e.g., Ralph. IMO, amp designers make trade-offs in order to design and build amps that perform at a certain level at a certain price point.

Probably still not saying this correctly, but it goes to show how the terminology can get in the way even though we are all really saying the same thing.

Unsound ... I found that same point confusing too, especially in light of ML's recommendation to use an amp that increases power as speaker impedance drops. That sounds like a Voltage Paradigm/SS amp. But our other "enthusiastic" techie members wore me out. Glad you picked up on that point too.

Going out on a limb here by getting technical than I should. But I recall reading posts from some our tech members that described the "common" ESL impedance function to be like capacitor. That is high impedance at low frequencies and low impedance (sometimes very low) at high frequencies. Indeed, some of the posts said that only amps that are stable driving such loads need apply.

I think another important question to think about is whether the ESLs were voiced to be driven by a SS or tube amp. As Ralph Karsten (Atmasphere) has explained in his white paper, most SS amps operate under the Voltage Paradigm, i.e., these amps have a very low output impedance and maintain constant output voltage. Under this model, power (i.e., watts) delivered to the speaker will inversely vary with the speaker's impedance, i.e., less watts at high impedances; more watts at low impedances.

By contrast, many tube amp have higher output impedances and operate under the Power Paradigm. In such cases, power (watts) will not vary as much with changes in speaker impedances.

The main point is that as long as the amp is operating within its safe operating range, it's more important to know whether the ESL was voiced to be driven by a SS or tube amp.

Btw, I think Ralph has suggested using ZERO autoformers to simulate higher speaker impedance. The theory is that Zeros can help amps manage the very low speaker impedances presented by some ESLs at high frequencies.

Generally yes ... reason ... there's much less energy called for in the high frequencies. If your music source required the amp to push 100 watts of power into the tweeter, not sure what would blow first ... your ears or the tweeter.

Most of the power demands placed on an amp are in the bass/low midrange. Much less power demands in the upper end.

George, that's a pretty clear clue that ML advises that the owner should use a rock stable SS amp. I don't know how low the impedance gets ... but that is a factor to consider when picking an amp.

This issue has come up before .. possibly in this thread. I'm a tube guy, so I'll let the SS guys weight in. I seem to recall that Ralph thinks Zeros can help in these situations. Never used em' so I stop here.

George, you posted "[a]ll you have to do is look at Stereophile speaker simulated load graphs of tube amps especially to see that they cannot give a flat frequency response into those types of loads, especially ones that dip down to 1ohms."

I generally concur, but as I am sure you know, not all tube amps are made the same. For example, take a look at JA's bench measurements of my amp, the ARC Ref 150:

http://www.stereophile.com/content/audio-research-reference-150-power-amplifier-measurements#52FW4Aq5RbfXP6dv.97

As you can see, the amp's voltage output plot is a bit wavy when a simulated load is plugged into the amp's 8 ohm taps. That said, JA mentioned that "[t]he figures for the 8 ohm tap [ranged between] 1 and 1.4 ohms; for the 4 ohm tap, they [ranged between] 0.55 and 0.87 ohm." I surmise that the same simulated load would be less wavy if plugged into the amp's 4 ohm taps.

Now ... an actual "off the bench" report from me. If you get a chance, check some of my posts on the "DEQX Game Changer" thread. I bought a DEQX PreMATE, which effects both time domain alignment adjustments and room equalization correction.

To set up the DEQX, actual mic'd measurement are taken at the listener position. The FR of my speakers was frankly a mess. I surmise most of the FR aberrations were caused by room anomalies, not by my amp's output impedance interactions with the speaker. Btw, I drive my speakers off the 4 ohm taps.

One other point of interest. I sent Al my DEQX data files for review and comment. I think Al might concur with my observations. And right now, my speakers are pretty well adjusted via the DEQX. I am enjoying a very pleasant musical experience.

So, based on the foregoing, I suspect that if my amp/speaker combo was checked in an anechoic chamber, my tube amp/speaker FR plots would measure pretty close to the results obtained if my speakers were driven by a high quality SS amp under similar conditions.

Last point. The reason my amp's output impedance is low'ish and output voltage somewhat constant is because ARC uses a prudent amount of negative feedback. There is also some sort of local negative feedback effect achieved by reason of a circuit configuration between the power tubes and the output trannies. Ralph or Al can better explain how that works.

Kudos to all for the good comments.

Cheers,

Bruce

Unless the amp is being tortured by the load and sensitivity of the speakers:

See this article:

http://www.stereophile.com/reference/707heavy#ovW5G38gcYb8whWE.97

Simple ... it's a function of the transducer system itself. That is what is meant by how the speaker is voiced.

Take a simple example. A speaker system may present an impedance load of 20ohms at the mid/tweeter x-over point. Yet, assuming the drivers are phase coherent at the x-over point, and if the speakers are well designed, the FR should be flat over the x-over point. Similarly, a speaker's impedance function even outside the x-over points may fluctuate. But still, the driver system emits a level FR SPL. It's about how the speaker was voiced.

There are some, but not many, speakers whose impedance function is near linear. Unless the impedance level is either ridiculously low or high, and the phase angle plots are not wacko, such a speaker could very well be both SS and tube (Power Paradigm) friendly. Just throwing this out there, but I think Maggies might represent a somewhat level 4 ohm load and pretty benign phase angles. I'm sure there others.