Yes H3 is affected the most, but at frequencies >100Hz
At LF(<60Hz) H2 is affected more than H3.
Yeah, OK - I see that. It's just that I don't consider that driver usable <100Hz. The sharp rise in H2 and H3 is what I often use as an indication of crossover point - or "don't go this low" point.
Agreed.
I was just commenting on speaker dave's statement in post #22 that distortion at and below resonance did not change with drive source impedance. H3 definitely seemed unaffected in this region, but H2 did change.
http://www.diyaudio.com/forums/mult...ver-qts-you-cant-tuna-fish-3.html#post2745854
Thanks Bolserst, that does make it easier to compare the 3 curves.
There is some variation visible in the 2nd harmonic curve, but its hard to make much sense of the variation. Higher Z makes the distortion greater at 50Hz but less at lowest frequenies and less in the midrange. You certainly can't say the "high Z is better" (or low Z). It almost looks to me like there are some competing 2nd order effects and they can cancel or add depending on the impedance.
It is also on the order of about 3dB for the impedance variation tested, whereas the midrange 3rd harmonic drops by about 10dB for the same range.
In the end, the midrange hysteresis effect is an instantaneaous nonlinearity of impedance (occurs every cycle of a sine wave) so, since F = BLi, constant i means lower distortion than the constant v of low source impedance.
Interesting stuff.
David S.
Here a post of Thorstenl. Posted in 2003 He explains it all here:
I think we are sorta talking crosspurposes.
My point is that Speakers of the conventional sort are current conrolled devices. This means if you feed the speaker with a voltage, meaning from a low impedance source (theoretically a zero impedance source), then the current flowing through the Speakers voicecoil will be a result of the voltage applied and the drivers impedance. Here comes the trickey part. The Drivers impedance (even if we discount mechanical resonances) is highly non-linear and signal dependent.
For one, the drivers voicecoil is normally filled and surrounded with a solid iron/steel chunks. If we take such a coil and apply a voltage from a low impedance source we will observe a distortion of the current flowing through the coil. This means that current through the vociecoil will be no longer directly proportional to the Voltage applied to the driver, but will contain a further non-linear element following in fact the cube (exponential function) of the current through the voicecoil.
Most common drivers will have motors (Magnet & Voicecoil) exhibiting 0.1 - 0.5% 3rd harmonics caused mainly by this phenomaena (eddy current distortion from here on) for around 1 Watt RMS applied power. As the distortion follows the cube of the current it also means this distortion follows the square of the power. Thus a tenfold increase in power will be accompanied by a hundertfold increase in eddy current distortion. Near full rated Power many speaker driver have (motor) distortion levels exceeding 50% composed of ODD ORDER (or in musical terms dissonant) Harmonics!!!!
All of this is a direct, unavoidable and absolutely NECCESARY result of driving the speakers voicecoil from a near zero impedance voltage source. We may apply zero distortion zero ohm source impedance voltage to the driver, all we get back for our efforts in making an amplifier with 0.000001% THD and a damping factor of 10,000 is that we have succeeded in maximally distorting the current in the drivers voicecoil and thus to maximise the distortion of the Driver/Amplifier system!!!!
If we now design an Amplifiers that feeds the driver from a very high impedance (theoretically an infinite impedance) we will directly control the current in the voicecoil and the current in the voicecoil will become directly proportional to our input signal. If we where to measure the voltage on the speaker terminal we would find the voltage to be now VERY HEAVILY distorted, but as the voltage is completely and utterly inconsequential to the forceexcerted on the cone and thus the movement of the cone we would attain an acoustical signal essentially free of certain types of distortion.
Just to complete the list of issues, another source of distortion in the motor (assuming of course a magnet gap geometry that ensures a magnet field that remains strictly constant strength across the whole intended voicecoil travel) is the "stiffness" of the magnetic field. This is related to the "willingness" of the magnet material to change it's magnetic properties under the influence of external fields. In our case the external magnetic fields are from the voicecoil.
Modern cermaic magnets are pretty bad in this respect, being able to accept much less sheer fieldstrength and being quite "soggy", being very notably modulated by the voicecoil mangetic field. The classic Alnico (Aluminum Nickel Cobalt) Alloys are by a factor of around five (depends upon the exact alloy) better than ferrite/ceramic magents with modern Neodymium magnets being better by a factor of up to twentyfive.
Just for kicks, using a currentsource fed fieldcoil would result in a magnetic field that cannot be externally modulated and also allows much greater fieldstrength than any permanent magnet....
Anyway, the distortion caused by this "magnet field stiffness" follows the square of the magnetic field modulating and thus the square of the current. Current drive cannot do anything about this. Luckily the square law favours even harmonics whch tend to lend warmth and musicality to the music if not being in too great a proportion.
Another happening in speakers is the so called thermal compression. I shall re-quote my text from above:
"given that the voice coil gets hot and is wound with copper or aluminum (instead of constantan) and thus increases it's resistance"
What this means is that with more applied voltage (and thus power) the voicecoil gets hotter and thus will result in less current being drawn (as an aside, the Re and thus Qe in the T/S Parameters will be effected too - that just for those misguided souls that assume T/S parameters are constants).
This means that for a given (say 6db) increase in drive voltage the actaul SPL produced by the driver will increase by less than increase in level of the driving signal. Socalled "High Fidelity" Speakers showing 4 - 6db thermal compression at full rated power are not uncommon, this means that near full power the speaker is actually 4 - 6db LESS sensitive than at low power. This is very audible, trust me.
What has all the above missive to do with the discussion here?
Ultrasound claims a Damping Factor of Zero and thus employs what is best described as current drive. It matters zip how this current drive is achieved (feedback or not), what happens is that current through the drivers voicecoil is controlled, not the Voltage across the voicecoil.
Having experimented with similar configurations (using chipamps) in the mid 80's of the last century I can attest to the rather surprising transformation of the sound. I ended up using current drive only on wideband midranges and tweeters (so from around 200Hz upwarsd), ending up with negative feedback via electret mike capsulas for the woofers (this was a fully active 3-Way speaker).
IF and that is a BIG IF I could design my "dream" Speaker driver from scratch I suspect it would look like this:
1) Fieldcoil Magnet with constant current actuation
2) Linear Magnet Field Geometry (Similar to JBL's SFG style arrangement)
3) Qm of the woofer cone section controlled by electromagnetic breaking (eddy current in the aluminum voicecoil former) and a voicecoil winding that makes the Qm the woofers dominant Q, not as usal Qe.
4) Coincident (in the same magnet gap) ICT style 2" Tweeter Dome, induction from the Woofer voicecoil coupled to the tweeter dome acting in effect as transformer to drive a very low resistance "voicecoil". The tweeter requires a form of phaseplug to extend the treble past 10KHz by avoiding cancellation, maybe a ring radiator arrangement would be good.
The above driver would be technically speaking "full-range" (or rather wideband) without electrical crossover as such, fully phase/impulse coherent. I'd probably aim at around 105db/2.83V/1m with a range covered of around 125Hz-12.5KHz, meaning it would likely have to be a 12 - 15" unit.
Together with that it would be nice to have a 20" or larger Subwoofer driver similar in constrction to the above and of course to drive the speaker drivers actively with "current drive" on open baffles. Add a nice ceramic or ribbon supertweeter....
Well, we all can dream.
Sayonara
(thorstens brains are OK)
Last edited:
That is an interesting post and I would agree with all of it except that we know that the conversion to ferrite magnets in the 70s led to higher 3rd harmonic distortion that current drive (proven by my measurements some pages back) will reduce.
(I also don't like the repetition of the comon audiophile falacy that 2nd harmonic distortion is good and 3rd is bad...way oversimplified.)
David S.
(I also don't like the repetition of the comon audiophile falacy that 2nd harmonic distortion is good and 3rd is bad...way oversimplified.)
David S.
Is it? The tritone (one and a half octave above the fundamental) sounds quite dissonant, while an extra component at exactly one octave higher than the fundamental is masked to a large extent by the fundamental. My ABX listening-tests with music samples with added distortion have made it clear to me that pretty high levels of third order harmonics can be tolerated, but the second order ones can be much higher still before they are even audible.
You're overlooking something important.
2nd order non-linearity doesn't just produce 2nd harmonic distortion products - in the presence of two or more simultaneous frequencies (eg all music) it also produces 2nd order intermodulation products.
Whilst the 2nd harmonic products might be considered "musical" since they're always the same note one octave up, 2nd order IM products will be dissonant, as they are sum and difference frequencies.
Now imagine music composed of dozens or hundreds of frequency components at once all inter-modulating with each other in every possible combination and you have a huge mass of dissonant non-musically related distortion products even if you only had pure 2nd order non-linearity in the system.
Last edited:
Some very interesting insights and experiments about the mechanisms why this is so : http://igitur-archive.library.uu.nl/med/2011-0204-200555/Book Heerens de Ru EN.pdf
Most important aspect : Our hearing is "plagued" by TONS of IMD all by itself!
Author's website where the generator software can be downloaded (also useful to explore the effect of polarity / phase on harmonics) :
three compartment cochlear model
Also 3rd harmonic distortion in the bass tends to fatten up the sound and is liked by a lot of listeners. IME.
Would there be any way to modify a Gain Clone type amp for higher output impedience without just using seriese resistance?
Yes, current feedback.
A small resistance in series with the driver creates a voltage in proportion to current through the driver. It is feed back to the amp input and increases the effective output impedance without losing a lot of power in the process.
Elliot Sound Products has a clear description of it.
Variable Amplifier Impedance
This is the technique used in the variable Z Fisher amp I described (and other amps of the era).
David S
What I found interesting in that article is that Rod appears to be quite sceptical of the effects of source impedance on driver distortion:
"It has been suggested that loudspeaker intermodulation distortion is dramatically reduced by using a high impedance source (but by whom I cannot recall - I do know that one site I looked at was Russian, and a reader sent me a translation many months ago). I have experimented with this idea to some extent, but have been unable to prove that this is the case - at least with the drivers I tried it with.
This does not mean that the claim is false, but I am unable to think of any valid reason that could account for such driver behaviour. It is interesting anyway, and some of you might like to carry out a few experiments of your own. I would be most interested to hear about your results should you decide to test this theory - preferably more thoroughly than I did."
To be fair the page appears to be written around 2003 so may be out of date, but I thought Le(x) modulation was a reasonably well known phenomena even at that time.
Apparently no readers of that page rose to the challenge of sending in results of their own experiments, or for some reason the page was never updated...
Edit: opps, I thought I was posting this in the flux modulation thread. 😱
Last edited:
I don't recall seeing any other measurements of distortion vs. output impedance, other than my own. Its always much easier to speculate on "what would happen" than to do the tests. Did Hawksford give measurements?
David S.
(And Jmmlc in this thread.)
True enough. I guess I was focusing on his comment "I am unable to think of any valid reason that could account for such driver behaviour.", when we know that force is proportional to current not voltage, Le varies with excursion, therefore impedance varies with excursion.
So using voltage drive will result in distortion due to distortion of the current waveform with excursion. I'm surprised that nobody wrote in to respond to the above quote in 8 years either on theoretical or practical grounds.
Maybe the effects of Le(x) just aren't that widely appreciated outside of those who design drivers ?
A lot of parameters vary with excursion but most of them are not linearized with current drive.
For example driver stiffnes always increases with excursion and Bl always drops off with excursion, but current drive doesn't seem to help either. My curves may show a slight change in 2nd harmonic, but the much stronger 3rd harmonic, due to the typical symmetrical suspension stiffening and Bl falloff, sees no effect.
The key is whether the physical parameter results in an instantanteous (every cycle) impedance nonlinearity. I have no doubt that if you measured current under low Z drive conditions that you would see 1% 3rd harmonic distortion of the current at mid frequencies. That is why current drive helps. At LF it makes no difference so the nonlinearity of the parameters must have no effect on current purity.
I haven't fully digested the Klippel paper but I believe that is what he is referring to when discussing current distortion.
Also, note that the mid frequency distortion and its cure are related to ferrite magnet material nonlinearity (hysteresis) and would still occur in a motor with flat Le vs. excursion.
Its true that only driver designers worry about flux mod distortion, because they can deal with it, as long as there is some money to spend on fancier motors. Adjusting amplifier source impedance is generally not a commercial solution.
David S.
Someone may have, but it didn't get posted. I've written to Rod a few times to point out minor mistakes or errors and never gotten a reply or acknowledgement.
I would image he gets 100X more mail than he can handle.
Fwiw, there is actually a choice “c” that is within the diy’r grasp and I am not sure it was mentioned. Pardon the redundancy if it was.
Background;
In the way old days I used to work on acoustic levitation systems (supports a small object with sound, without contact at a high temperatures for manufacturing in space). I was using a very large “speaker magnet” although it wasn’t a speaker and the magnet assembly was actually a Samarium cobalt magnet stronger than anything I had ever seen before. This was before Neo magnets so my first exposure to rare earth magnets was this whopper that had about 20lb of magnet and 40LB of steel.
Anyway, this was supposed to make the sound source louder (levitation begins at about 155dB) but compared to the electromagnet version, wasn’t as powerful.
In fooling around with the sound source (called a St Clair vibrator) , I did find an odd thing, if I spaced the resonant bar which was he source a bit further away from the magnet, I got a little bit more level.
I though, weird, maybe the magnet is too strong????
I went to the machine shop and cut up 8pc of 1/2 inch square steel bar stock that was the thickness of the magnet assy (about 3 inches) and put those on the outside of the magnet structure.
Sure enough, the level went up some more. By the time the entire perimeter was covered in these ½ inch “shorting bars”, the output had increased 5dB.
The magnet was much too strong!, I later designed a ceramic magnet that gave 10dB more and much later a piezo driven source that was about 20dB more efficient.
Anyway, the choice “c” is to add iron on the outside of the magnet which shorts out some of the gap flux by providing a “better “ path. One can simply “tune” by adding more shorts.
Hope that helps,
Best
Tom Danley
Danley Sound Labs
Background;
In the way old days I used to work on acoustic levitation systems (supports a small object with sound, without contact at a high temperatures for manufacturing in space). I was using a very large “speaker magnet” although it wasn’t a speaker and the magnet assembly was actually a Samarium cobalt magnet stronger than anything I had ever seen before. This was before Neo magnets so my first exposure to rare earth magnets was this whopper that had about 20lb of magnet and 40LB of steel.
Anyway, this was supposed to make the sound source louder (levitation begins at about 155dB) but compared to the electromagnet version, wasn’t as powerful.
In fooling around with the sound source (called a St Clair vibrator) , I did find an odd thing, if I spaced the resonant bar which was he source a bit further away from the magnet, I got a little bit more level.
I though, weird, maybe the magnet is too strong????
I went to the machine shop and cut up 8pc of 1/2 inch square steel bar stock that was the thickness of the magnet assy (about 3 inches) and put those on the outside of the magnet structure.
Sure enough, the level went up some more. By the time the entire perimeter was covered in these ½ inch “shorting bars”, the output had increased 5dB.
The magnet was much too strong!, I later designed a ceramic magnet that gave 10dB more and much later a piezo driven source that was about 20dB more efficient.
Anyway, the choice “c” is to add iron on the outside of the magnet which shorts out some of the gap flux by providing a “better “ path. One can simply “tune” by adding more shorts.
Hope that helps,
Best
Tom Danley
Danley Sound Labs
- Status
- Not open for further replies.