I tried to find some background about this. Everything I find discusses room modes as causing uneven frequency responses due to interference and standing waves. I didn't find anything about creating harmonic distortion, so I doubt that you are right.
What I do find in the literature is that room modes occur at multiples of the wavelengths of the room dimensions.
But they are not created - they occur only if they are in the sound being played.
Do you have some source that actually mentions the creation of harmonics that are not in the original sound?
Jan
Last edited:
The writer missed one sentence:"the voltage in between just existed to allowing the current to flow though the cone." The volt is meaningless in the electric-magnetic area, the only unit used in here is ampere.
So he is not saying something of "voltage is useless".
Start with Acoustics by Beranek or in general just any experience in acoustics?
But anyway, it's not very important in this discussion at all.
Just a little side details about distortion.
Because this whole detail is quite the rabid whole in sense of how to see how standing waves are even excited in real music.
Again, totally not important in the context of CC amplifiers.
Correct, but the impression has been created that harmonics can be created by room modes. And that is incorrect.
Harmonic modes of rooms can be excited by the sound poured into them, but they are not created if they are not part of the soundfield from the speakers.
So I stood not corrected ;-)
Jan
Ehm?
I said that for LOW frequencies a CC amplifiers won't do anything.
If you quote, you should quote the entire context of a message.
In midrange, yes it most certainly helps.
I already said that before, even with a reference to a scientific paper submitted to an official acoustic research society.
Which is in line with the page you're showing, which only stated midrange frequency data, not lower frequency data.
Although those papers go quite a lot more into important details.
I feel I have to add a few things too:
Regarding adding inductors to woofers, it can be a bit tricky with passive components and transducers, the impedances interact and can 'boost' the response close to the transducers resonance for example. This is why many passive filters incorporate impedance flattening networks for the drivers. Best to simulate in crossover sims to see what happens before you stick a coil in there. The result might disappoint (or sound better) for the wrong reasons.
When it comes to the current distortion, I have a possible other way of describing it:
The driver creates the distortion voltages with excursion. The interaction with the impedance the driver 'sees' is what seems a bit confusing?
As I see it, there is 'distortion voltage' generated in the driver when it moves. If the driver 'sees' an impedance close to 0, all this distortion voltage will be over Re, generating current in Re(the voice coil) and this creates the acoustic output of distortion.
If the driver 'sees' infinite impedance, there will be no distortion/error voltage over Re, only over the amp terminals, hence no distortion current in the driver and no acoustic output of the distortion.
Taking this to middle ground, using a ideal voltage drive amp, a series resistor with the same value as Re between the driver and amp, the resulting distortion should be halved, divided over Re and the series resistor. I think this is close to what I saw when I measured the current distortion and added a resistor in series that was close to Re in value (wrote about it here https://www.diyaudio.com/community/threads/the-elsinore-project-thread.97043/post-7116385 ).
I'm not expecting to see the acoustic output distortion to go down by the same amount, since there are other distortion mechanisms too, but it should definitely do something.
It seems a bit counter-intuitive to describe current distortion as a 'error voltage' generated in the driver though.
I'm very much agreeing with what I've read in tmuikku's posts of how he sees it, and I think he's better at explaining it than I am. My post here is just to try to help people see it from another point of view.
Regarding adding inductors to woofers, it can be a bit tricky with passive components and transducers, the impedances interact and can 'boost' the response close to the transducers resonance for example. This is why many passive filters incorporate impedance flattening networks for the drivers. Best to simulate in crossover sims to see what happens before you stick a coil in there. The result might disappoint (or sound better) for the wrong reasons.
When it comes to the current distortion, I have a possible other way of describing it:
The driver creates the distortion voltages with excursion. The interaction with the impedance the driver 'sees' is what seems a bit confusing?
As I see it, there is 'distortion voltage' generated in the driver when it moves. If the driver 'sees' an impedance close to 0, all this distortion voltage will be over Re, generating current in Re(the voice coil) and this creates the acoustic output of distortion.
If the driver 'sees' infinite impedance, there will be no distortion/error voltage over Re, only over the amp terminals, hence no distortion current in the driver and no acoustic output of the distortion.
Taking this to middle ground, using a ideal voltage drive amp, a series resistor with the same value as Re between the driver and amp, the resulting distortion should be halved, divided over Re and the series resistor. I think this is close to what I saw when I measured the current distortion and added a resistor in series that was close to Re in value (wrote about it here https://www.diyaudio.com/community/threads/the-elsinore-project-thread.97043/post-7116385 ).
I'm not expecting to see the acoustic output distortion to go down by the same amount, since there are other distortion mechanisms too, but it should definitely do something.
It seems a bit counter-intuitive to describe current distortion as a 'error voltage' generated in the driver though.
I'm very much agreeing with what I've read in tmuikku's posts of how he sees it, and I think he's better at explaining it than I am. My post here is just to try to help people see it from another point of view.
Last I checked E = I * R, so E and I are linked. At least as long as we're not talking about superconductive materials.
He's (probably) fundamentally correct in that the voltage does not enter into the equations directly, but the current does. The equations probably start with F = B*I*L, where F is the magnetic force, B is the magnetic flux density, I is the current, and L is the length of the conductor (think single-turn coil). Look, ma! No voltage in the equation!! Voltage no matter! Obviously the maths explode once you move to multi-turn coils and take the homogeneity of the magnetic field into account, but I bet the fancy equations just solve F = B*I*L in a small unit area/volume and add it up at the end (otherwise known as integration).
But if E = I*R, then I = E/R, so the voltage does enter into the equations as soon as you take even the DC resistance of the voice coil into account. And the current becomes complex once you account for the inductance of the voice coil.
Fundamentally, there should not be any difference between current drive and voltage drive. An ideal current source forcing 1 A through an 8 Ω speaker will impose a voltage of 8 V across the speaker. Similarly, an ideal voltage source imposing 8 V across that same 8 Ω speaker will force 1 A of current through the speaker. The two scenarios are equivalent.
The only way I can see the two sources being different is how they handle the back-EMF. There's no way to avoid the back-EMF. If you force the speaker cone forward (either by voltage or by current) and reduce the current/voltage to zero instantaneously, the suspension on the speaker cone will move the cone back to the equilibrium position and this will generate a current in the speaker voice coil. An ideal voltage source will shunt this back-EMF current to ground, thus, keeping its output voltage constant. An ideal current source will change its output current by an equal and opposite amount to the back-EMF, thereby cancelling the back-EMF current and allowing zero change in the voltage across the voice coil. Again, in the ideal case, there is no difference between the two. Thus, if there is any difference between current and voltage drive in audio speaker applications, I bet they're due to the non-idealities of the amplifiers and not due to the different operating modes.
I encourage those who are curious about the differences between voltage and current drive to look into Thevenin and Norton equivalent circuits.
Tom
They are exactly the same if the 8Ω speaker is purely resistive, like a planar.
Most dynamic drivers are far from that They usually vary in resistance on the + and - stroke. Usually depicted as le(x) as in the inductance with respect to stroke position. If they spend $$$ on the motor they can, however, get quite close:
http://www.zaphaudio.com/Le(x)/
Compare the Scan-Speak 18W8531G ±6.5mm
To the Scan-Speak 18WU8741T00 ±8.0mm (mfg spec 6.0mm)
And notice that the 18W8531G impedance is not as linear in impedance throughout the stroke.
With a voltage amplifier, the 18W8531G will reduce output when the impedance increases, since we have a fixed voltage. The current will vary and this variance in current is extra distortion.
With a current amplifier the current is fixed. The driver will just get a little bit hotter when the impedance increases but the current will stay the same. Less distortion. Or at least less if we ignore being close to fs.
No, its discussing because I want to be clear that what you said is wrong, and I hate things that are wrong.
YMMV.
Jan
That's where your whole misconception starts, it seems.
"Back-EMF" aka microphonic voltage is a voltage to start with, not a current.
This voltage gets transformed to counteracting current via the transfer impedance which is the sum of the VC static impedance + amp output impedance. When the latter is high (infinite) no current is developed. When the latter is zero, the current is significant. And when it is negative, approaching -Re, the current is huge.
Please read my previous post for a better understanding.
Tom, I bold-ed a section in your post because I don't understand. Isn't the defining thing of a current source that it does not change no matter how much (EMF) voltage or current you throw at it?
Jan
Basically, yes. But current drive does not address all motor distortion effects and therefore the net improvement is limited. The best results are obtained with drivers that have massive Le(x) fluctuations, i.e. cheaply constructed drivers. For a well-designed driver like a Purifi woofer, the improvement in distortion is negligible. Tailored impedance still is a good thing even for those as it allows you to dial in a arbitrary system Q, and high source impedance is always good for less thermal compression (an aspect that is very effective for small tweeters, especially small AMT's).
I've wrote it before...
I've conducted experiments with current drive vs. voltage drive (and even negative output impedance drive) where the terminal voltage of the driver (vs. frequency ) was dialed in to be exactly the same for any choice of source impedance, with the help of a DRC package.
Obviously, when the driver terminal voltage is the same, the SPL output must be the same, too, and it fact it was (as is the current). Therefore it sounded and measured really almost identical, notably at low to medium excursions. But, the fine-print was different and especially the large signal behavior / overload recovery was very different.
It turns out (I don't know how many times I've written this before) that for a given driver in a given enclosure/alignment there is an output impedance vs frequency profile that is the best compromise overall in measured properties as well as in listening tests. It never was pure current driver or pure voltage driver for the (limited amount of) drivers I've tested, except for AMT's.
I've conducted experiments with current drive vs. voltage drive (and even negative output impedance drive) where the terminal voltage of the driver (vs. frequency ) was dialed in to be exactly the same for any choice of source impedance, with the help of a DRC package.
Obviously, when the driver terminal voltage is the same, the SPL output must be the same, too, and it fact it was (as is the current). Therefore it sounded and measured really almost identical, notably at low to medium excursions. But, the fine-print was different and especially the large signal behavior / overload recovery was very different.
It turns out (I don't know how many times I've written this before) that for a given driver in a given enclosure/alignment there is an output impedance vs frequency profile that is the best compromise overall in measured properties as well as in listening tests. It never was pure current driver or pure voltage driver for the (limited amount of) drivers I've tested, except for AMT's.
Pentodes were always regarded as inferior sound to triodes. But significantly more efficient and less hum/buzz; that's why "mainstream commercial". See Radiotron Designer Handbook, F Langford Smith, "The Relation Between the Power Output Stage and the Loudspeaker". Much "High Fidelity", US and UK, work promoted triodes, or triodes with overall NFB, culminating in Williamson.
The trends Langford Smith reported may have been wrong, but his paper was widely printed and highly influential.
The losses are still there, but now provided by the amplifier (drivers with Qt is close to Qe
dave
Malcolm Hawksford "Transimpedance Power Amplifiers Systems for Current-Driven Loudspeakers"
Note: AES papers are behind a paywall unles sprovided by the author, whixh is th ecase for this one (i can’t find the original link.
dave
Yes, you will get lower distortions for such setup with current drive 99% of the time. Maybe except woofers around fs. Mostly higher order harmonics are reduced with current drive. The spectrum is cleaner .Current drive is no brainer compared to voltage drive in open baffle active system. As reletively no external forces are present on driver like in box. So you EQ current drive and get perfect impulse as with voltage drive. No ringing as someone thinks. Damping doesn’t matter here as the driver is minimum phase system so can be equalized perfectly. Full perfect inversion posible for linear time invariant systems.So only non linear part is effected by current drive. Only for externaly aplied forces current drive will respond diferently compared to voltage.
Why people say that current drive amps are dificult to build i don’t understand. Electrons doesn’t care about the mode, they folow the potential diference. So the circuit can folow current or voltage or any other feedback formula. There are many circuits in industry wich uses current drives and nobody sees it dificult to do. At my work we use only current drive to drive laser diodes with impulses many hundred amps and the circuits are nothing special. I use only current drive amps or mixed mode for audio also. It works its done and it sounds good.
Take lm3886 on a breadboard, spend 15min and you will get current drive amp. EQ to the same as voltage drive and compare for yourself. Just compare on the naked driver not on box speaker with filters designed for voltage drive. And if you like it, build the system acordingly.
This post for inspiration for you guys😀👍