"Thus, there is nothing indispensable in electrical damping; and in principle, there cannot be a difference in the driver’s resonance behavior, either in the frequency or time domains, whether the damping is accomplished by a low-impedance amplifier or mechanically."
(Esa Meriläinen: https://www.edn.com/loudspeaker-operation-the-superiority-of-current-drive-over-voltage-drive/)
"volume: 100l
fill: 4,375 kg sonorock (idea: Esa Meriläinen)
resistor: 20R"
(Me: https://www.diyaudio.com/community/threads/loudspeaker-for-current-drive.403936/page-5#post-7475291)
But...
no one would be able to do it because there is no microphone suitable for it.
(Esa Meriläinen: https://www.edn.com/loudspeaker-operation-the-superiority-of-current-drive-over-voltage-drive/)
"volume: 100l
fill: 4,375 kg sonorock (idea: Esa Meriläinen)
resistor: 20R"
(Me: https://www.diyaudio.com/community/threads/loudspeaker-for-current-drive.403936/page-5#post-7475291)
I'm absolutely with you on that!
But...
no one would be able to do it because there is no microphone suitable for it.
Last edited:
Last edited:
Damn it!
The small >> o << followed gravity 😆 However, the assignment of the colors to the resistors is correct.
Dark green is D3o, with 0R
light green is D3, with 20R
All you've "proved" with your test is that the more constant you make the current source, the more constant will be the current. And the horse says "neigh"...
Correct. Just that mechanical damping is not any easier to practically realize with good precision, let alone frequency selective.
Best thing would be an additional moving coil in a separate magnet circuit, shunted at resonance only. It should be more linear than the VC so that the damping stays intact even when the main VC is popping out of the gap already.
Hi,
yeah I think you must do some imagination, start asking what the impedance peak actually is and then imagine how the driver actually works out, how the electrical and mechanical workings are connected. For example, why adding damping material to closed box would reduce impedance peak? When you really have asked all questions and got answers so that it all fits together you can reason with the backEMF very effectively.
BackEMF does not make boost (at resonance), as it makes current that opposes motion it can only reduce output. But, if we reduce backEMF it appears as a boost, reduction of reduction, kind of magic!🙂 This is important to understand, the loudspeaker drivers are designed at factory to work with voltage amplifier, which means they are assumed to dominate circuit impedance, a driver would reduce it's own output by affecting circuit current and that is factored in designing drivers: The flat frequency response you see on datasheet includes a lot of backEMF at resonance and voice coil inductance above, both reducing current and force on the voice coil affecting the frequency response. If you now take driver's ability to affect circuit current away by increasing series impedance, you'd reduce this reduction, you'd get boost in frequency response compared to what datasheet shows!
So, as KSTR already explained nicely, Bl is involved to make the cone move (F=Bli) in the first place but it is also involved how much backEMF voltage gets generated when the cone moves. The driver is motor and generator at the same time, simultaneously*. Now as Bl reduces with excursion, the backEMF voltage would drop due to Bl reducing, and dropping backEMF voltage means less backEMF current which means reduced impedance peak. As voltage from amplifier is still the same like a moment ago it means more current flows as impedance peak drops, which kind of compensates the original F=Bli, which made the cone move in the first place. So, basically distortion reduction would be through less distortion in first place (less variation in force). This would concern the main resonance only, and it's harmonics. Varying Bl would contribute distortion above the resonance, as current cannot compensate dropping force being perpendicular. Perhaps some confusion here? Great displacement on low frequency would change driver parameters on all frequencies above, for the whole bandwidth, so drop in Bl on a kick drum hit would distort the whole midband even though the fundamental had less harmonics due to current compensating for reduced Bl.
At least this is how I think about it, I have no complete picture of driver motor workings so I do not know how accurate this is and what else is involved. This is the simple idea though, at least it is logical and makes sense.
* this seems to have been one hard to understand thing in other threads. backEMF is literally the driver behaving as generator due to any excitation force that makes it move. As amplifier makes voltage, current flows and the cone moves, and the driver works now as another voltage source in the same circuit. Voltages do not make acoustic output though, but any current through voice coil makes. This is how the series impedance works reducing distortion, as voltage from backEMF is kind of always the same for given acoustic output, we just increase impedance of the circuit (and amplifier voltage) so that driver generated voltage has less effects on the circuit current, and thus less effect on acoustic output. It is multiple kind of different systems acting simultaneously affecting each other so no wonder could be hard to grasp, although it's easy to understand and reason with in practical level after first realization.
Last edited:
What's your problem with microphones? Just use a good one.
For high SPL I have 1/4" GRAS microphones - these don't show THD at 130-140dBSpl (can do >160dBSpl, good for nearfield measurements). For normal levels my Earthworks M50 does a perfect job down to <0,01% THD (cause then noise kicks in, but no mic THD). But even a cheap MicW M215 will do that perfect. And even better a high sensitive 1/2" ref mic.
What ****** mics do you use?
Last edited by a moderator:
1/4 inch capsule is the obvious choice when lowest distortion ir required from the mic. I'm happy with my Gefell MK301.
A lot of remarkable work has been done here:
https://www.diyaudio.com/community/threads/multitone-distortion-measurement-on-drivers.391826/
and here:
https://www.audiosciencereview.com/...tion-higher-than-speakers-yes-no-maybe.33747/
money quote:
https://www.diyaudio.com/community/...urement-on-drivers.391826/page-4#post-7185358
...and now I'll continue taming the happy camping parrots ;-)
Leo L. Beranek, Albert Neville Thiele and Richard H. Small have done a great job.
But...
Wolfgang Klippel et al. extended this to include movement outside the zero position.
Total destruction is the result of any electrical dampening approach.
Especially with voltage drive!
The less the nonlinear! electrical dampening is, the less is the jump out penomenon.
But of course current drive (or resistor drive ;-) doesn't work without other means of dampening.
For example the "Sonorock" (heavy mineral wool) mechanical damper does not have any of the electrical dampening nonlinearity problems at all. It works without deflection and is therefore linear.
The "Drive Current Distortion Measurent" shows: Rockwool dampening does no harm.
The impedance measurements show the low Q at resonance.
Last edited:
I had read that. You have described the BL/EMF problem very correctly and very well. It quickly just seeped into my knowledge base. Without paying attention to where it came from. Thank You!
I would like to add: This causes the voice coil to jump out of the gap
Sorry for my late reply.
Last edited:
Are you sure? to me it seems ypu've got some misunderstanding? The paper linked in post #40 shows jump resonance happens when damping is low, and they seem to mean when electrical damping is low, which means current drive. Also the graph shows no jump happens with high damping, iow voltage drive.
One can work around random noise limit, though.
Use synced record-while-playback, play signals that are periodic (repeat after N samples) and use test tones at FFT bin centers, then do time-domain averaging to reduce the uncorrelated noise. This is brute force but very effective.
With REW you can simply use "coherent averaging". Larger FFT sizes also help.
Looking down 20dB lower than without averaging is often feasible (needs > 100 averages, though).
'Jump' resonance seems pretty obscure and unlikely either way. It would need a high power frequency sweep to traverse that vulnerable zone, so the sensitivity jumps down from the overhanging peak, resulting in a one-off glitch. Or perhaps in cases of extreme abuse there would be repetitive glitches around 100-120Hz in the example above, where the bass resonance can't make up its mind what frequency it wants to be at.
*targeted editing by me, sorry.
I am pretty sure that there is no experimental result for high electrical damping (real voltage drive amplifier) with the real vifa in this paper. 😉
Last edited:
Maybe. @KSTR suggested to look at it from a time domain perspective.
The whole thing really calls for a programming exercise in one's favourite programming language:
-a spring loaded mass,
-energy loss from velocity based friction,
-repetitively ping the particle with energy whose values are taken from a *.wav file or something like that,
-make the ping force, spring constant, and opposing friction vary with displacement,
-with each 'ping' append the new particle position to a *.wav file for later viewing.
My guess up front would be that the coil "popping out of the gap" would mostly be the inertia carrying the mass on a kind-of ballistic trajectory, and it would be unresponsive to other signals until the mechanical spring pulls it back. So that would be a source of IMD for higher frequencies riding on top of bass signals.
Anyway, that would be a very, very rough "1st order" simulation. There's basically no end to the rabbit hole, and you can keep adding complexity and detail.
A big factor could be how the motor's field gets pushed around by the coil. Normally I think of the field lines as preferring to take the shortest path through the least air. But if the coil gives the magnet a good shove, the field lines may take longer path, and we have to look at the geometry in the given moment to figure out if the interaction between the coil and magnet will be weakened, or whether some kind of positive feedback occurs (magnet field getting pulled towards the coil).
Not so sure about that - perhaps for an underhung motor, I suppose.
Mostly, the effects should be more gradual, and apparent early on in displacements from the rest position. You could probably model with Bl reducing linearly with displacement.
Another aim would be to find the best mechanism for controlling intermodulation distortion in the presence of significant low-frequency excursion...