Why are low impedance speakers harder to drive than high impedance speakers

For headphones, why do we need good, high, powered amp to
move headphones with very high impedance (300-600 ohms).

You don't. Many phones of this impedance are easily driven by a few milliwatts. Some 'phones require a bit of power and others don't. How much power they need is independent of their impedance.

+1 on that for sure :)

I thought the notion that speakers, being transducers, vary much more in sound that do amps was universally agreed upon, and should therefore be selected first. The notion that power amps vary in character as much or more than do speakers is one I disagree with.

If you have no preference for tube or solid state then this is the way to go. If you have found that you prefer one over the other then you will need to get the amp first and then find a speaker to match. That is why in most cases, you start with the amp. It is tricky- to know that amp you have to hear it in a variety of circumstances to make an informed purchase. But if you do that, you are less likely to flush more $$$$$ down the loo trying to get the system to sound the way you want it.

if your tube amp has 4, 8, and 16 ohm taps (typical in tube amps, though the RM-200 offers 2, 4, and 8 ohms), and you hook up an, say, 8 ohm speaker to the 4 ohm tap, the power available to the speaker will be less than it would be if connected to the 8 ohm tap. But, says Music References Roger Modjeski, a tube amp so employed will usually be producing not only less power, but also less distortion, and better sound. As Al mentioned, Roger calls this tactic "light loading". In addition to lower distortion, an additional benefit of using a lower impedance tap is that the amps output impedance will be lower---it will have a higher damping factor, and will interact less with the varying impedance characteristics of the speaker load, resulting in a more predictable frequency response.

While generally true, a problem that can turn up when doing something like this is that the transformer can 'ring' if insufficiently loaded. In addition, with such a load, it will not be as flat across its bandwidth, as the transformer will tend to express less of its turns ratio and more of its inter-winding capacitance. The 'lighter' you load the transformer the more of a problem this becomes.

So the result, while possibly reducing distortion in the output tubes, will be to **increase** distortion from the transformer (ringing) and degrade the frequency response. Of course, if the amp employs negative feedback some of this will get sorted by that, but a problem with negative feedback is that while reducing distortion overall (in particular lower ordered harmonics), it actually **introduces** higher ordered harmonics that otherwise may not have been present at all! Its best not to give feedback too many places to screw up.

 

 Ralph, you mentioned 'former' twice in the last part of your post.

Ach!

That should read:

Again, this all comes down to intention. Is your intention to get the system to sound as good as it can or is it more important to simply play loudly? If the latter, than some of the lower impedance speakers and higher power transistor amps will be of interest; if the former, then you will be very careful to be matching the speaker to the amplifier (and not the other way ’round) and most likely avoiding lower impedances in general.

-as Al corrected.

So while I can say with precision that most solid state amps are more accurate, as far as measurements are concerned

Erik, I don't think this statement is correct, and here's why. If you look at the specs, the lower distortion and apparent constant voltage characteristics of most solid state amps looks great! The problem is, that bit of paper ignores how our ears perceive sound.

This takes a bit to grasp! To give you some idea, most of us know that the ear employs a logarithmic approach to sound pressure. This is why we use the VU scale of decibels.

So take this concept, but apply it to harmonics. The ear seems to use something that looks very much like a logarithmic approach when it comes to how sensitive it is to harmonics- being less sensitive to lower orders and far more sensitive to higher orders on what looks much like an inverse logarithmic function.

The fact that the ear is more sensitive to higher orders has been known for decades and should not be a matter of debate! This is very easy to prove with simple test equipment.

Add to that the fact that the ear is tuned to be most sensitive to bird song frequencies (Fletcher-Munson). This fact arises out of evolution and is millions of years old- birds are the first warning of a predator in the area!

So the fact is that if the ear does not care about the lower harmonics so much, then logically we should be designing to eliminate the higher orders, especially since the tools that the amplifier designer has in the tool box all have certain limitations. For example, as I stated earlier (and as been stated by Norman Crowhurst, a universally recognized sage), loop negative feedback is known to add additional harmonics and IM distortions (the harmonics can go as high as the 81st and the intermodulations occurring at the feedback node in the amp). In this way an amp with feedback will usually sound brighter than an amp without, even though on the bench they both measure flat.

So what is more 'accurate'?  Low distortion on paper is meaningless unless we also know what it is that makes it 'low'. Its one thing if we can see the lower orders in the harmonic distortion spectrum. But if we are to take how the ear perceives sound into account, the higher orders should really be a lot lot lower than they are currently with all 'low THD' amps. And by that I mean **at least**  2 orders of magnitude! 

Just seeing 'low THD' doesn't cut it.

****This is ignoring how the ear works!!!**** (fist bangs tabletop)

The fact is that as far as the ear is concerned, the distortion of most amps with seemingly really low THD is that the distortion is higher. Its easy to hear too- which is why tubes still exist in the marketplace 60 years after being declared obsolete. Its why the tubes/transistor thing has been going on longer than the internet!

(if the tubes weren't doing something right, they would have been gone long ago. How many flathead V8s are still in production? If you got 'none' then you probably also know its because they are obsolete. There is a huge difference between being declared obsolete and actually **being** obsolete!)

In essence, the bench specs are an excellent example of the Emperor's New Clothes. This is because you have to ignore the obvious coloration of brightness/harshness/brittle in order to really say that its more accurate. The bench spec thing still has its roots in the 1960s and has not changed much since then (its mostly based on an idea of low distortion and flat frequency response while totally ignoring what the ear perceives; its actually tuned to the eye rather than the ear).

Put another way- we like to think our amps are low distortion because that is how they look on paper. That appearance is false- we're not measuring the right thing. Try to wrap your head around the fact of the ear's crazy sensitivity to higher ordered harmonics and use **that** as a baseline instead. If you can make that translation, you will see that most amps are fairly high distortion and not accurate at all.

When this is true, the 4 Ohm is usually 3 dB more sensitive. The low impedance causes extra current flow which provides for increased force against the same magnet.

This seems to require clarification!

Given two drivers of the same **efficiency** (1 watt, 1 meter), if one is 8 ohms and the other is 4, the 4 ohm unit will be 3 db more **sensitive** (sensitivity is measured at 2.83Volts at 1 meter; if this is 8 ohms that is one watt, at 4 ohms its two watts; two watts is double one watt and there is your 3 db ).

That current will not flow as expected by the speaker designer if the amplifier output impedance is higher than about zero ohms. This is because as the output impedance is increased, more and more power will be dissipated in the output section of the amp rather than the load- dissipated in the form of heat.

That’s a lot of amps! In particular, this is especially true of tube amps, whose output impedance can often be measured in ohms rather than fractions of an ohm. One might ask, ’what is the point of such an amplifier?’ and the answer has to do with how humans perceive sound.

In a nutshell, we perceive volume, or sound pressure, by listening for the higher-ordered harmonics. This is because pure tones do not exist in nature, and apparently nature sorted out millions of years ago that listening for the higher-ordered harmonics is more expedient, as it would millions of years before anyone invented pure fundamental tones :)

BTW this is very easy to prove with simple test equipment and is not a matter of debate. I’ve posted the way to prove this a number of times on this site.

Since this is the case, a good number of designers (myself included) prefer to design amps that by intention do not make the higher ordered harmonics. To do this often requires a higher output impedance, because to do that means avoiding loop negative feedback (which is known to enhance higher orders and generate more of them at the same time; see Norman Crowhurst). Loop feedback lowers output impedance; without it the output impedance is therefore higher.

It also happens that it is far easier to design such an amplifier by avoiding the use of semiconductors in the signal path. Semiconductors have non-linear capacitive elements inherent in their junctions (magnified by current through the junction) and these are known to create higher ordered harmonics in the distortion structure of the device (FETs and MOSFETs far less so than conventional bipolar devices; a particular device known as a varactor takes advantage of this aspect and is used as a variable capacitance to tune radio receivers).

The bottom line is that if you are dealing with an amplifier designed to not make higher ordered harmonics (as opposed to just low THD in general), the usual voltage rules as defined in the quote above simply don’t work (and I explained why in my second paragraph). Put another way as a speaker designer you have to pay attention to driver efficiency rather than sensitivity.

This is why back in the old days, many speakers had mid and tweeter level controls. They were not there to adjust the speaker to the room, they were there to adjust the speaker to an amplifier of unknown voltage response (high output impedance).

The approach is trickier, but has the advantage of less overall audible distortion (which the ear converts to tonality, often favoring that tonality over actual frequency response errors!).

In most cases this design approach is to avoid ’brightness’ and ’harshness’; two audiophile terms used to describe the presence of trace amounts of higher ordered harmonic distortion.

The way I see it, if a system **always** has brightness (which will be found to not be toned down by a treble control because it does not arise from a frequency response error) then the best it will sound will be like a nice stereo rather than real music.

Again, this all comes down to intention. Is your intention to get the system to sound as good as it can or is it more important to simply play loudly? If the former, than some of the lower impedance speakers and higher power transistor amps will be of interest; if the former, then you will be very careful to be matching the speaker to the amplifier (and not the other way ’round) and most likely avoiding lower impedances in general.

you took my statement out of context. Here is what I actually said:

Yes- it was not to contradict you, but to use the phrase as a talking point, as I see that approach recommended a lot. Then later people sell off the gear in search of that holy grail... I think if you start with the grail first you're less likely to sell and waste the cash.

Atmashere, the OP said nothing of ESL's. I did make mention of : "Most typical speakers...."

Correct- in fact by 'most typical' is about 95% of the market. In high end, its still about 85% and so is 'mostly accurate' :)

For those outside of that percentage (Charlesdad's speakers are box speakers but they are an example) that voltage thing just does not work. This is entirely due to the designer wanting the speaker to work with a particular kind of amp. In Charlesdad's case, the speakers were originally envisioned on a set of our M-60s and later that manufacturer started making SETs, which work much the same way. SETs in general are much happier on higher impedance loads despite often having 4 ohm taps and they tend to make constant power rather than constant voltage owing to zero loop feedback.

In a tube amp, the 4 ohm tap is not an efficient way to use the output transformer, which often means not only less power (lost due to heat) but also less bandwidth, sometimes up to an octave lost on the bottom end. So its often really worthwhile to avoid low impedance speakers with tube amps even if you have the taps on the output transformer!

So why would a designer go the other direction?

Lower impedances have appeared as a means to get more power out of solid state designs. But this is very different from getting lower distortion! So a lot depends on what your goals are!

I think the trend towards speakers with lower impedance corresponds to the trend for speakers to be smaller yet more full range to fit into people’s lifestyle.

The size of the speaker has nothing to do with its impedance. The two are unrelated design aspects.

Impedance at port frequencies is always low.  Check any ported speaker impedance curve and see.

This statement is mostly false. Ports are usually placed in the peak of a cabinet resonance (which will be seen by a peak in the impedance) as a means of reducing the peak and spreading it out. If properly placed, two lessor peaks with a dip in between will be seen, but overall usually represents a higher than nominal impedance.

Usually it comes down to approach: picking the speaker you love and then finding the right amp

I don't buy that this works! Often people have a preference about tubes and transistors- the speaker **must** be chosen to take that preference into account!! Otherwise you may never get satisfaction and a lot of money down the loo.

 ^ Lest anyone get the idea that the increase/decrease of power output by ss amps into decreased/increased impedances would suggest that frequency linearity would be compromised, the opposite is actually the case, they actually provide better frequency linearity. Most typical speakers will decrease/increase their sensitivity in direct proportion to the increased/decreased impedance changes.

In the specific case to which this quote refers, a Quad ESL was the speaker and most definitely does ***not*** fit this rule! The Quad has impedance curve arising from capacitance and not resonance of a driver in a box (for starters, there is no box). Planars in general don't fit that rule either.

I feel like several points need clarification and at least one needs to be made. First the clarification:

If sound quality is your goal, your amplifier investment dollar will be best served by a speaker of 8 ohms or more, all other variables being equal (and unfortunately they almost never are).

By this I mean that no matter what amp you have, its not in your best interest to make it work hard if you want the best sound out of it that is possible! It does not matter what sort of amp- tubes, transistors or class D. The problem is the same- as you decrease impedance the distortion in the amp goes up. Unfortunately the kind of distortion that is going up is the kind you really don't want- because its the kind that is pretty audible! You don't hear it as breakup or crackle though, you hear it as tonality. Usually this means that the sound will be harsher and less detailed. This is because the distortion components are usually recognized by the ear as harshness and it take a vanishingly small amount of distortion to do this!

This is because the human ear uses higher ordered harmonics to gauge how loud sounds are, and as a result is evolved to be far more sensitive to them. This is in fact why distortion can be so low as to be hard to measure, yet it can still manifest as brightness and hardness.

So to reduce distortion and make the amp sound smoother, use a higher impedance load. Steve McCormick, a well known solid state amplifier manufacturer, sent a note to Paul Speltz, who is known for the 'anticables' but also for an autoformer known as the ZERO. The ZERO allows you to drive a 4 ohm speaker while the amp is loaded at 16 ohms. What Steve said in his letter was that even though his amps had no worries doubling power into 4 ohm loads (and BTW are very well built and designed IME) that the fact of the matter is they sound better driving 4 ohms via the ZEROs.  This is simply due to reduced distortion.

One other point not previously mentioned is the effect of speaker cables! At lower impedances they tend to be far more critical, where shorter distances and much larger gauges are required to prevent loss of definition and impact. This can be a pretty big deal as the series resistance of the speaker cable can have a pretty dramatic effect on the effective output impedance of the amplifier! By contrast the speaker cables are far less critical at 16 ohms- which is why a lot of us kids got by with hardware store zip cord  in the old days.

The reason 4 ohms is such a big deal has more to do with transistors than anything else. Back when tubes were the only game in town, speakers were usually either 8 or 16 ohms. They were often a lot more efficient too, as acoustic suspension hadn't been invented (Henry Kloss, who was the inventor of that was a co-founder of Acoustic Research, who in turn made the AR-1, the world's first acoustic suspension speaker about 1958).

Obviously several things at once were happening at the end of the 1950s. Acoustic suspension (far less efficient) was getting started, and so were transistors. At first the two were not significant. But as silicon transistors (and consequently higher power) became more available (which really started in the late 60s or early 70s) the need to get a bit more output out of the amp started to take off. BTW, high efficiency loudspeakers are a **lot** harder to build, by nearly a factor of 10 over acoustic suspension, so you can see that speaker manufacturers sensed a profit motive (as did the amp manufacturers by going solid state- a similar motivation is occurring now with class D). However by getting a possible 3 db more output out of an amplifier suddenly became a big deal; acoustic suspension allowed a smaller size but paid the price in efficiency. Anything that might make them appear easier to drive was helpful.

But in the world of high end audio, where tiny little things can make a difference, the need for 4 ohms is almost non-existent (at least not if sound quality is the goal). There really isn't a speaker technology (like planars for example) that actually **need** to be 4 ohms as opposed to a higher impedance. In most cases, this simply happens because the speaker manufacturer does not realize the smoothness and transparency benefits that are possible with the same amp if the impedance is increased!