Speaker choice: impedance, sensitivity, and tubes


Looking at most speakers' specifications I cannot help but notice that very few are rated 8 ohms, and most dip well below their nominal specifications at certain frequencies. This makes makes me wonder how audiophiles with tube amplifiers select their speakers. Most tube amps have 8 and 4 ohm taps only, and from what I understand tube amps don't take kindly to dips in impedance. Is there a rule to abide by when selecting a speaker to match well with a tube amp when it comes to impedance?

Same with sensitivity. Perhaps not as important as impedance, but a lot of popular brands out there have disappointingly low sensitivity (85-86 dB). Why is that? I never understood it since the higher sensitivity, the easier it is to drive a speaker without having to have a beast of an amp. Seems all manufacturers should be striving to design their speakers to have higher sensitivity. Is is more expensive to make speakers with higher sensitivity?

Case in point, I own two pairs of speakers, one rated 88 dB and the other 91 dB. The 91 dB pair has an impedance of 8 ohms flat (according to the manufacturer) while the 88 dB pair has a nominal impedance of 6 ohms (not sure about dips). The difference is quite dramatic in terms of volume on my 180 watt tube amps. I often have to crank the volume way up to get sufficient loudness level in my rather small listening room (11x12) with the less efficient pair. That to me is crazy. The speakers are my fall-back pair from my digital past, and knowing how they behave, I'd never purchase them for my all-analog system today.

So as I'm thinking of upgrading my speakers, I'm quite perplexed about finding a speaker that would match well with my tube monoblocks - provide an easy load and good loudness level without strain. I've been considering giving Harbeths a shot, but I'm really turned off by their low sensitivity of 85 dB (30.1) and impedance of 6 ohms. How big an amp would I need if speakers with a sensitivity of 88 dB barely generate sufficient volume with 180 watts per channel?!

Can anyone explain in technical terms how tube amps handle speaker impedance and, secondarily, sensitivity? And perhaps more important from the practical standpoint - how does one select a speaker to match a tube amp?
actusreus
Hi Marek,

I think that this recent thread answers a lot of your questions, and will be well worth reading.
Perhaps not as important as impedance, but a lot of popular brands out there have disappointingly low sensitivity (85-86 dB). Why is that?
One basic reason is that efficiency, deep bass extension, and physical size trade off with one another. If you want to increase efficiency you have to sacrifice deep bass extension and/or make the cabinet larger. Also, many speakers are designed with the expectation that they will be used with solid state amps, which can provide large amounts of power at much lower cost than a tube amp of comparable quality and power capability.

Regarding the lack of volume from your 88 db speakers, based on the observations you have described I suspect that the spec is inaccurate, and significantly overstates their sensitivity. 180 watts into 88 db speakers should easily produce higher than sensible volume levels in an 11 x 12 room.

Also, when you consider new speakers keep in mind that speaker sensitivity is often specified based on an input of 2.83 volts, rather than 1 watt. 2.83 volts corresponds to 1 watt into 8 ohms, so if the speaker's impedance is 8 ohms it makes no difference which way the number is defined. However, if the speaker's impedance is 4 ohms subtract 3 db to convert 2.83 volt sensitivity to 1 watt sensitivity (or more precisely, 1 watt efficiency). If the speaker's impedance is 6 ohms subtract 1.25 db to do that conversion.

Best regards,
-- Al
I own the Rogue Cronus Magnum (90 tube watts) with Harbeth C7es3 and at half volume I can hear the music across the street and 3 doors down with all of my windows closed. Only tried it once because I don't enjoy that much volume. 180 tube watts is more than I've ever read anyone using with Harbeth.
Thanks Al. The linked thread was indeed very interesting and informational. I agree the specs on the speaker must be overstated.

Don,
Which tap do you use for your Harbeths, 4 or 8 ohm?
Some additional information such as specific products might allow others to be more helpful.

I am using Quicksilver V-4 amps (120 wpc) to drive my Proac Response 2.5 (83 - 85 db efficient)in an 11x17 room and have no issue producing volume level louder than most people like.
Actusreus, I'm gonna jump in with my crude understanding of how the various electronic variables of tube vs SS amps interact with "conventional" dynamic cone speakers.

First, read Ralph Karsten's (Atmasphere's) white paper that discusses the so called Voltage vs Power Paradigms. In short, in the case of SS amps, voltage output remains somewhat constant, but current (amps) vary with the load. That's why SS amps can increase wattage when driving low impedance speakers, particulalrly in the low bass region. As impedance drops, power (watts) increase.

Not so for tube amps. In the case involving typical tube amps, current (amps) remains relatively constant regardless of impedance, but voltage varies. As a result, power (watts) doesn't change as much over the frequency range where impedance changes all over the place.

My ARC tube amp power rating is pretty constant regardless of impedance load, which is consistent with the Power Paradigm as described in Ralph's white paper.

Another difference is damping factor. SS amps generally have higher DFs than tube amps because SS amps generally have much lower output impedance factors than tube amps. I think DF is calculated as: speaker impedance/amp output impedance. Ralph and Al have written that much ado has been written about DF, and the bottom line is that it is a misunderstood and over-rated stat. The "lure of the law" so to speak is that the higher the DF, the better ontrol the amp will have over the woofers. The short answer is not necessarily.

My tube amp has a DF of 8. Yet it has great bass slam and extension. Why?? I have no idea.

I'm still waiting for a cogent response to a question that I posted in another thread which is how can one make an informed choice between mathcing a particular amp with a particular speaker. I still don't know. All I DO know is that I am satisfied with the sonic swill coming out of my speakers even even though I'm dumb as a stone.

Hope this helps.

Happy Holidays,

Bruce
I've driven the Harbeth M-40.1's with as little as 30 tube watts with reasonable success. IME, "watts" are over-rated.
for 88db speakers at 8 ohms:
1 watt = 88db
2 watts = 91db
4 watts = 94db
8 watt = 97db and so on doubling your wattage to gain 3db.
For every 10db, you will hear a perceived doubling in wattage.
91db rated speakers should put out 3db more at the same volume.
88db speakers at 6ohms vs 91db speakers at 8ohms will be roughly 1/2 the volume (approximately).
I am puzzled by the story of your 88dB speakers. What preamp are you using? This sounds like a gain problem, not a watts problem.

Taking gain into account is an important part of system building.
I'm still waiting for a cogent response to a question that I posted in another thread which is how can one make an informed choice between matching a particular amp with a particular speaker.
Bruce, as the other thread makes clear there are many variables and matters of degree that are involved. Therefore I don't think that a one-sentence or other kind of cogent guideline can be formulated, beyond Ralph's suggestion of trying to determine what the designer's intention was.

I think that your other comments above are well said, except that with respect to this statement ...
In the case involving typical tube amps, current (amps) remains relatively constant regardless of impedance, but voltage varies. As a result, power (watts) doesn't change as much over the frequency range where impedance changes all over the place.

My ARC tube amp power rating is pretty constant regardless of impedance load, which is consistent with the Power Paradigm as described in Ralph's white paper.
... I would put it that in the case of a tube amp both current and voltage vary somewhat as a function of load impedance, with the net result being that power delivery varies significantly less than it would with a solid state amp that acts as a voltage source.

Also, the similarity of the MAXIMUM power ratings of a tube amplifier's 4 ohm and 8 ohm output taps involves different considerations, that are not directly related to the variation of power delivery as a function of load impedance WITHIN the amplifier's rated capability. The latter results from the interaction of amplifier output impedance and speaker impedance.

In considering all of this, it would probably be helpful to digest this Wikipedia writeup on the voltage divider effect. And to consider Z2 in Figure 1 as being the impedance of the speaker, which varies as a function of frequency, and Z1 as being the output impedance of the amplifier, which is essentially zero for a solid state amplifier, and typically one to several ohms for most (but not all) tube amplifiers. Then assume that in both cases Vin is being provided by an ideal voltage source (i.e., one having zero output impedance), and do a few calculations of power delivery for various combinations of Z1 and Z2. The power delivered to Z2 being equal to the square of the voltage across Z2, divided by Z2.

That is a bit oversimplified(!), because it neglects impedance phase angle, but a few such calculations should make clear what is happening with respect to the interaction between speaker impedance variation as a function of frequency and amplifier output impedance.

Best regards,
-- Al
Something that has not been pointed out is the performance of the output transformer and the load that is used.

For example the 88db speaker is also 6 ohms. If used on the 8 ohm tap the impedance is transformed by the transformer to a much higher impedance that the power tubes are actually driving. However it will be too low as to perform correctly the 8 ohm tap should be loaded at 8 ohms. The result is that the amp can't make as much power and will have higher distortion.

On top of that the bigger the transformer, the more its going to have troubles making full bandwidth. These problems are compounded on the 4 ohm tap and especially if the tap is incorrectly loaded. You might see as much as an octave of bass response lost on this account!

However we see that transistors are not immune to the issue of impedance. While it is true that many transistor amps can drive lower impedances with more power, it is also true that those same amps will have more distortion while doing so. The ear translates many forms of distortion into tonality; IOW just because a transistor amp can drive the lower impedance is not at all the same as saying it sounds its best.

So here is the bottom line, tube or transistor:

"If sound quality is your goal, it will be best served by a speaker that is 8 ohms or more, all other things being equal." If you want greater **sound pressure** there is a slight argument for 4 ohms if you have a transistor amp. IOW, there is no argument in support of four ohm speakers in high end audio.
Thank you all for your valuable input. Really good stuff.

Roscoe-
I use Rogue Audio tube preamp (with gain adjustment) and monoblocks. I don't think it's a gain issue. My phono preamp also has adjustable gain and I can easily switch it with very little additional noise, but the sound becomes rather unpleasant - I'd describe it as shrill, especially the highs. So it's not just about getting it loud; it's about getting it loud enough while maintaining the right tonal and frequency balance. And to clarify, the loudness level from the 88 dB speakers (Totem Hawk) is sufficient, (about 85 dB) but it seems it's at the top of their capabilities and requires a lot of power from 180 WPC monoblocks, which seems just crazy. (As a side note, this also demonstrates to me how important gain matching for a cartridge is. My Delos sings at 58 dB setting, but shrieks at 63 dB. The gain values are converted from mV so may not be accurate to a decibel, but still I wonder how some people use phono preamps with fixed gain and multiple cartridges.)

Bruce-
Sounds like we are looking for an answer to the same question :) Plus the issue of why so many high-end speakers have low sensitivity and low impedance. I always thought high sensitivity, high impedance speakers would be better than low sensitivity, low impedance speakers even though I didn't have sufficient technical knowledge to fully understand it. So it's very interesting to read Atmashpere's post supported by actual technical understanding of the issue.

I would like to hear a speaker designer's point of view regarding this common tendency in speaker design. If designers such as John DeVore and Alan Yun (Silverline Audio) understand this quite well, why don't others?
Ralph, Al and Actusreus -- if I reincarnate and if my I.Q. is high enough, I would love to be an EE designing speakers and amps. The problem is I have no idea where or what our hobby will be like in the future. In the meantime, I'll keep my day job and just try to pick up as much as I can on the web. Thanks Al for the hperlink -- I'll try to read and understand it.
...And if the speakers are measured at 2.83 Volts, as the impedance numbers go up, the power (Watts) of a ss amp output is reduced, as the efficiency of the speakers are correspondingly increased. All of which should work towards constant linear speaker output. Also consider that it's easier to make speakers with a more constant lower numerical impedance, than it is to make speakers with a more constant higher numerical impedance. Furthermore, wave form fidelity seems to be an elusive quality for speakers with higher numerical impedance/efficiency specs to achieve.
Ralph and Al, just reading over some of my older threads and re-read this one.

Just thinking out loud here. I wonder how many retail B&M sales people really grasp the electronic amp/speaker subject discussed in this and other similar OPs. I honestly am not making a statement one way or the other because I haven't stepped foot in a B&M store in decades to schmooze or listen to various combos of amps and speakers. I just raise the question.

Out of curiousity, I called Paradigm to ask whether the S8s (my speakers) were designed to work best with SS or tube amps. The tech service rep said the S8s soak up current, as much as the amp can push out. In fact, he said SS was the way to go.

Now, . . . my amp is tube. Would my rig sound even better with a high quality SS amp, I rhetorically ask?? Maybe, just maybe. In the meantime, the rig sounds pretty good as is. I have a great subwoofer, which back-stops the fronts, so if I have low quality bass output, it's pretty well masked.

I would be curious if any A'gon members who are or were on the retail B&M side of the hobby would chime in. I'd sure like to hear their views.
Out of curiousity, I called Paradigm to ask whether the S8s (my speakers) were designed to work best with SS or tube amps. The tech service rep said the S8s soak up current, as much as the amp can push out. In fact, he said SS was the way to go.

Now, . . . my amp is tube. Would my rig sound even better with a high quality SS amp, I rhetorically ask?? Maybe, just maybe. In the meantime, the rig sounds pretty good as is. I have a great subwoofer, which back-stops the fronts, so if I have low quality bass output, it's pretty well masked.
I have an Octave V70se (70w @ 4ohm) driving Dyn C1 Signatures (85db) and I can turn it up a lot louder than I can stand. Not because it's fatiguing but because for me it's jut too loud. That being said the Octave was designed for 4 ohm speaker and more important the Dyn is very stable at 4ohm. Now my son had Klipsch KLF30's (103db) and holy cr@p could they get loud. With either his Bryston 3Bsst or my Octave. But all I can say good about the KLF's is they could get super loud. Couldn't stand the sound except at super super low volumes. But in general when comparing anything side by side louder will win 95% of the time (short term comparisons).

So would a SS amp sound better. Yes with the right amp. I'd take a Burmester 956 over the Octave V70SE any day. But my Bryston B100sst and my sons 3Bsst were not as good sounding as the Octave V70se with the Dyn's.

But with tubes you can swap them out any change up the sound a bit (warmer - more musical or more linear and detailed).

Last I believe specs are a very general guideline. Any 2 pieces of equipment with the same specs will sound quite different. It's all about system synergy.
Xtil6 - Right and agreed. My Q is how many B&M stores really know how to mix and match, especially when there's an inherent bias (pun) that they will sell what they have.
Not many - but only the good ones. I think all dealers try their best but sometimes profit margin or price point (lower) means more sales. Well established dealers have been though it. Took my local dealer a couple of years to get it. Now that is his strong point. Total system synergy. But it does come with a price.
Actusreus wrote: "I would like to hear a speaker designer's point of view regarding this common tendency in speaker design. If designers such as John DeVore and Alan Yun (Silverline Audio) understand this quite well, why don't others?"

Because most audiophiles use solid state amps, most speaker designers don't bother with making their speakers tube-friendly. Which is fine by me, because it leaves a market niche for us little guys to exploit.

Paying a lot of attention to the impedance curve reduces your driver choices, and often makes crossover design more challenging. I could probably get away with about 1/3 to maybe 1/2 fewer components in my crossovers if I didn't mind a roller-coaster impedance curve. Also as noted already, high efficiency + decent bass extension = a big box. With audio in general moving towards small-box speakers for aesthetics, big, tube-friendly speakers are going against the grain.

While impedance dips can reduce the real-world power that a tube amp can deliver, impedance peaks can be bad too because of their effect on the frequency response (assuming the speaker was designed for "voltage paradigm" solid-state amps). Let me explain:

Suppose we have an "8-ohm speaker" with twin impedance peaks in the bass region that rise to 16 ohms, and a 32-ohm peak at the crossover frequency of 2.5 kHz. At a level equal to "1 watt into 8 ohms" a solid-state amp will be delivering 1/2 watt into those bass impedance peaks, and 1/4 watt into that 32-ohm peak at the crossover frequency. If the speaker was voiced for a solid state amp, then what happens on a tube amp is this: You get 3 dB more output into the bass impedance peaks (which can make the bass sound underdamped), and you get about 6 dB more output in the crossover region (which makes the speaker sound forward and fatiguing). Your instinct would be to blame the tube amp and conclude that tube amps are "fat and forward" because the speaker sounded fine on the solid state amp, but the real culprit is the speaker's impedance curve.

The approach I take to the impedance curve is to keep it fairly high and with no significant peaks north of the bass region, so that the spectral balance doesn't change with amplifier type. To deal with the effects of the bass impedance peaks, I tune the box considerably lower than is theoretically "optimum", so that the region boosted by the port is moved down lower in frequency. The net effect is more extended bass than we normally would have gotten, without excess upper-bass thickness. This is as close to a "free lunch" as it gets, but it's only a "free lunch" when contrasted with the same speaker on a solid state amp. (Actually my port lengths are user-adjustable, for compatibility with the various amplifier types).

In my opinion the high efficiency and benign impedance curves that make for a tube-friendly speaker are generally beneficial regardless of amplifier type, so the main drawback is the larger box size imposed by the requirement for higher efficiency.

Of course mine is by no means the only legitimate approach to designing a tube-friendly speaker. The designer can let the impedance curve do whatever and simply voice the speaker for tube amps.

I've done some over-simplifying in this post, but hopefully given enough information to illustrate some of the considerations on the designer's side of the fence.

Duke
dealer/manufacturer
Duke (Audiokinesis) - thanks for that cogent explanation. However, it riases a bunch of other questions for me.

First is, when a speaker manufacturer publishes frequency response specs, how can they possibly apply to BOTH tube and SS amps at the same time. Unless the amp in someway can sense the change in impedance and adjust its power output when encountering impedance curves, peaks and valleys, how is it possible for a speaker that is designed for a SS amp to have a flat FR if a tube amp is used, and vice versa of course?

Second, if the answer is "SS is SS and tube is tube, and never the twain shall meet (pun - LOL)," isn't it a bit of a misrepresenation on the speaker manufacturer's part not to mention that its published stats only apply in the case of [blank] type amps and that results can vary if [other blank] type amps are used?

I bet that most speakers on the market today have roller coaster impedance curves. I think I read somewher that Wilson speakers have pretty wild curves here and there too, but are also considered tube friendly. I understand that ARC uses Wilsons to check the sound of it amps.

So I guess the bottom line is what is the consumer supposed to do to make an informed decision when mixing and matching components??

Thanks for your cogent response. Hopefully we'll be hearing from some the EE/tech types too.
Recently found this...

http://www.dhtrob.com/overige/tubefriendly_lsp_en.php
Thanks Ngjockey -- that's exactly what I was looking for. In fact, I think it responds to the OP.
Bifwynne poses some excellent questions.

Bifwynne: When a speaker manufacturer publishes frequency response specs, how can they possibly apply to BOTH tube and SS amps at the same time?

Duke: If the impedance curve is fairly smooth, then there will be no significant change in frequency response when going from tubes to solid state. And the added power a tube amp delivers into the (virtually inevitable) bass impedance peaks usually only changes the shape of the curve in the low end ("fattening" it up a bit) without actually extending the -3 dB point significantly (UNLESS the port tuning is lowered). So while the frequency response curve is not going to be exactly the same, we can make an educated guess that we'll probably get a little bit more bass energy with a tube amp (and not to overly complicate things, but some tube amps moreso than others).

But if the impedance curve is a roller-coaster over most of the spectrum, then the speaker's tonal balance will change significantly with amplifier type, and the best bet is to follow the speaker manufacturer's recommendations on what sort of amps to use (if that information is hard to come by, you may be able to find out what amps he used at audio shows).

Bifwynne: If the answer is "SS is SS and tube is tube, and never the twain shall meet (pun - LOL)," isn't it a bit of a misrepresenation on the speaker manufacturer's part not to mention that its published stats only apply in the case of [blank] type amps and that results can vary if [other blank] type amps are used?

Duke: Well maybe so, but there is enough... ah... "optimism"... already in the specs of many speaker manufacturers that making comparisions between speakers based on specs is somewhat "speculative" at best. Some manufacturers assume significant contribution from the room in calculating efficiency and/or bass extension, and some do not, and that will probably have a bigger effect on specs than whether a tube or solid state amp is used. But it's perfectly kosher to ask the manufacturer.

The one thing a manufacturer can't really "pad" with optimistic assumptions is the enclosure size. Assuming you're comparing the same type of enclosure (like vented boxes), if one box is significantly bigger than the other, that speaker will either go deeper or be more efficient (or maybe a little of both).

Bifwynne: I bet that most speakers on the market today have roller coaster impedance curves. I think I read somewher that Wilson speakers have pretty wild curves here and there too, but are also considered tube friendly. I understand that ARC uses Wilsons to check the sound of it amps.

Duke: If the Wilsons were "voiced" on "power paradigm" speakers (this paper should be required reading for anyone tube-curious and/or speaker geekish: http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php), then they would work fine with ARC amps. Also if the speaker's impedance curve "zigs" where the frequency response curve "zags", a tube amp will tend to smooth out the frequency response.

Bifwynne: So I guess the bottom line is what is the consumer supposed to do to make an informed decision when mixing and matching components??

Duke: The ideal would be to have the anechoic frequency response of the speaker (preferably out to at least 45 degrees off-axis) along with the impedance curve, but unless the speaker has been reviewed by Stereophile or SoundStage, that information is unlikely to be available.

If a tube amp is involved, try to find out what the output impedance (or damping factor) of the amp is, if possible. Then contact the speaker manufacturer, tell him the amp's output impedance, and ask him if that amp would be a good match with his speakers. Or tell the speaker manufacturer the brand and model of amp you're looking at, and he can probably probably still give you good guidance. Or tell the amp manufacturer what speakers you're considering, and get his input.

Don't be fooled into thinking "bigger is better" when it comes to damping factor - the woofer's cone is never unpowered as long as the amp is on, so it never has to rely on "damping factor" to stop its motion; the audible effect of "damping factor" is on the shape of the frequency response curve, so good speaker/amplifier/room synergy is the key. Those three form a system, in my opinion. Anyway damping factor is simply 8 ohms divided by the output impedance of the amp in ohms, and is not an indicator of quality in and of itself. The circuit techniques that result in a very high damping factor (very low output impedance) often tend to be detrimental to sound quality, but that's another topic, probably for another forum.