The most unique feature of the KM1 was the thermal power protection. The first version injected a little DC current through each driver. Resistance was directly measured as an indicator of voice coil temperature. The idea was to only enter a protect mode when the voice coil was right on the threshold of burning up.
This required setting up the system individually for every driver change, (it was sensitive to voice coil DCR) so a second generation KM1 used analog computer techniques to accurately simulate voice coil temperature. I had to measure 2 time constants for system calibration. The short term TC of voice coil heating and the long term TC of magnet structure heating.
David S.
ATC is a legitimate company so I would assume there is something here. Still, I'm not sure why increasing inductance is benficial. Also, they talk of reducing 3rd harmonic and measurements show that 2nd harmonic is the greater issue.
Running impedance curves on demagnetized structures is a traditional test for non-motional impedance effects such as inductance. I'm not sure how the unmagnetized state impacts the distortion. (Not doubting it. Just don't know.)
Note that the effect we are talking about is distortion of the current (with a stiff voltage source). Since F = BLi, then distortion of current gives distortion of output. Driving with a current source gives low distortion current and higher distortion of voltage, but cleaner SPL output.
I still think this is more of a magnet material issue than anything else.
David S.
Thanks for the link. That's a good, concise overview from SL.
He points out that the primary culprit is the nonlinear permeability of steel in the return circuit (e.g. the B-H curve), as the voice-coil field interacts dynamically with B field.
Seems another good reason to test loudspeaker motors with energized magnets and not miss the fallout of B and VC field interactions.
Also brings me back to how much of a liability steel itself actually represents in a loudspeaker motor.
His upshot seems to be that shorting rings are important. Unless you're ATC, that is!
He points out that the primary culprit is the nonlinear permeability of steel in the return circuit (e.g. the B-H curve), as the voice-coil field interacts dynamically with B field.
Seems another good reason to test loudspeaker motors with energized magnets and not miss the fallout of B and VC field interactions.
Also brings me back to how much of a liability steel itself actually represents in a loudspeaker motor.
His upshot seems to be that shorting rings are important. Unless you're ATC, that is!
Last edited:
Okay, before I start taking flames, I'll relent by pointing out that ATC seems to be tackling the same problem SL points out--the nonlinearity of steel--by replacing steel near the VC with an undisclosed material (SLMM) with a more linear B-H curve. So while shorting rings shield steel from nonlinear interaction with the voice coil field, SLMM simply replaces the offending material with something better behaving.
Fascinating stuff. With it's extreme SPL capability, relatively small mid drivers and dome (!) tweeter, it sounds like the KM1 was born to walk a tightrope between hell and a hot day in Bombay.
Was any of that thermal tracking used to compensate for power compression? I had a silly idea along those lines recently.
David, this is very interesting. Would you care to share more particulars--your test setup, distortion orders that were most affected, etc.? Maybe start a new thread?
Is the mechanism for this distortion reduction akin to the effect of raising inductance, ala ATC?
Seems like higher output impedance amplification has quite a bit to offer, including less power compression among other things. The major roadblock seems to be that it violates the status quo of high damping factor that lets folks casually mix and match amps and speakers with acceptable results, since pretty much every modern speaker is voiced for voltage amplification.
Raise the output impedance, and you have to throw out standard crossover design and start designing amps and speakers as interdependent systems.
But, IMO, it also opens a new world of possibilities where you can design drivers with positive attributes that would not work with voltage amplification:
How about some big, high-efficiency motors with very high flux density that raise the proportion of B to L in the BL parameter (sort of like more bias in an amp--now it takes more power to reach the B-field reversal point where hysteresis really bites.) This driver would be too low-Q for voltage amplification, but with just enough output impedance, you can dial in the system Q of your choice.
How about going further by deliberately dropping drivers' Qm so the mechanical system is damped mechanically instead of electrically, and maybe some unique design elements could be added to mechanically damp bending modes on the cone itself? I understand this might be preferable to relying on amplifier damping because I have heard it said that amp Zout can vary with frequency.
Fun to imagine...
For example, I've got a couple Aurasound NRT-18-8 bass drivers that start looking very versatile and interesting in simulations with about 16 ohms of output impedance...
Circling back to the beginning, if high Zout makes driver motors behave better, then all the better!
This was years ago.
I was experimenting with an old Fisher tube amp with a "Z-Matic" control that varied output impedance. High output impedance of course had a big effect on frequency response so I used the compressor loop of the B&K oscillator. That is, the SPL from the mesaurement microphone was used as a feedback voltage to hold response flat with different output impedance. I measured fundamental, 2nd harmonic and 3rd harmonic distortion.
Low frequency (say, below 150Hz) distortion was the same for low and high source resistance. Midrange distortion (2nd harmonic)was considerably reduced.
I still have the curves somewhere.
I think constant current drive makes a lot of sense for full range drivers, as long as you can get the right bass response. (Lower Qm, as you say) But the interaction between crossover sections of a multiway system looks like a lot of work to me. Still, some people like series connected networks, which have to deal with the same issues.
David S.
No thermal tracking, we were primarily concerned with drivers dying in a studio environment.
One interesting factor was that we had to decide whether to use ferrofluid or not. It turns out that ferrofluid had a maximum temperature spec (150 C?) while the coil was good for 210 C. (I don't remember the exact numbers here.) That sounds like a step backwards but the better heat conductivity meant the the lower temperature with ferrofluid was achieved with the same input power or output SPL. In that way the coil temp would be generally lower and power compression less.
Of course, if you used constant current drive....
David S.
Re: Z-matic--I'd love to see the curves. What was the Zout you selected for your test?
Re: ferrofluid--have you ever done/seen any studies of the useful lifetime of ferrofluid, especially in such a demanding application? How rapidly does it evaporate and polymerize? Was there a recommended "oil change" schedule for the KM1?
Bill
Re: ferrofluid--have you ever done/seen any studies of the useful lifetime of ferrofluid, especially in such a demanding application? How rapidly does it evaporate and polymerize? Was there a recommended "oil change" schedule for the KM1?
Bill
Quote from post #26 by Bill F.
(Page 18 for B-H Curves of magnets)
http://magnet.atp.tuwien.ac.at/download/fidler_euro.pdf
(page 5 for conductivity of permanent magnets)
http://www.ndt-ed.org/GeneralResources/MaterialProperties/ET/Conductivity_Misc.pdf
(page 1 for conductivity of ferrous materials)
http://www.ndt-ed.org/GeneralResources/MaterialProperties/ET/Conductivity_Iron.pdf
Best Regards
George
(Page 18 for B-H Curves of magnets)
http://magnet.atp.tuwien.ac.at/download/fidler_euro.pdf
(page 5 for conductivity of permanent magnets)
http://www.ndt-ed.org/GeneralResources/MaterialProperties/ET/Conductivity_Misc.pdf
(page 1 for conductivity of ferrous materials)
http://www.ndt-ed.org/GeneralResources/MaterialProperties/ET/Conductivity_Iron.pdf
Best Regards
George
(Some more) Magnets data
http://www.intl-magnetics.org/pdfs/0100-00.pdf
Soft Ferrites
http://www.intl-magnetics.org/pdfs/SFG-98.pdf
Regards
George
http://www.intl-magnetics.org/pdfs/0100-00.pdf
Soft Ferrites
http://www.intl-magnetics.org/pdfs/SFG-98.pdf
Regards
George
Hi!
Seeing the Lowther copper pole pieces in EX4 units I thought I'd experiment with a more lowly fullrange -
YouTube - Comparing good value speaker with phase plugs - Tannoy 12inch - Lowther PM6C
Can anyone hear the difference between the straight unit and the unit with the copper pole piece added? Possibly on the drums at 02:17 and 06:03 ?
Best wishes
David P
Seeing the Lowther copper pole pieces in EX4 units I thought I'd experiment with a more lowly fullrange -
YouTube - Comparing good value speaker with phase plugs - Tannoy 12inch - Lowther PM6C
Can anyone hear the difference between the straight unit and the unit with the copper pole piece added? Possibly on the drums at 02:17 and 06:03 ?
Best wishes
David P
The Z-matic curves were taken at the two extremes. One would be the usual damping factor of 20 or so. The other was probably on the order of 100 ohms out?
Not studies but I have run into some of the issues with its use. Some drivers can get "sludgy" after long periods of rest. I think the iron particles pack in. Heavy excercise seems to help rejuvinte in this case.
Part of the choice to use ferrofluid, when the power input wouldn't be increased, was that we thought the temperature limit when using it was a soft limit that could be exceeded occaisionally. The voice coil temperature was a hard limit. Exceed it and quickly burn out the coil.
David S.
At KEF we used current drive on more than one occasion. We did experiments with a modified Quad 405 amplifier module. It's output impedance was I believe greater than 1000 ohms.
It certainly has the capability of lowering midrange distortion, but as noted plays havoc with the low frequency response because of loss of electrical damping. You are now totally dependent upon mechanical damping, which is typically not very great in conventional woofers. Additionally the mechanical parameters are not well controlled, so the low frequency response (resonance freq, Q) becomes very variable.
One way around this is to add considerable mechanical damping, though this is difficult to implement. One method used in the past was a cloth layer immediately behind the diaphragm. However, as excursion increases, the effects of this damping are not always consistent.
The other way is to have mixed feedback, voltage drive at low frequencies and current drive at high, but this also is not easy to implement.
I believe Prof. Hawksford has published papers comprehensively discussing and evaluating current drive.
We had a project at KEF to investigate rooms acoustics where we installed over 30 speakers in an anechoic chamber fed from a 32 channel custom built DSP system. Each of these speakers was an active, bi-amplified, current-output amplifier configuration. I built all the drivers for these, along with the active EQ. It was exceedingly complex to measure and individually adjust the active xover for each driver, made much worse because of the current drive.
All things considered, through the years I have come o the conclusion that it is better to design a good driver, than to "fix" it with special amplification.
This applies especially in the case of motional feedback!
Andrew
It certainly has the capability of lowering midrange distortion, but as noted plays havoc with the low frequency response because of loss of electrical damping. You are now totally dependent upon mechanical damping, which is typically not very great in conventional woofers. Additionally the mechanical parameters are not well controlled, so the low frequency response (resonance freq, Q) becomes very variable.
One way around this is to add considerable mechanical damping, though this is difficult to implement. One method used in the past was a cloth layer immediately behind the diaphragm. However, as excursion increases, the effects of this damping are not always consistent.
The other way is to have mixed feedback, voltage drive at low frequencies and current drive at high, but this also is not easy to implement.
I believe Prof. Hawksford has published papers comprehensively discussing and evaluating current drive.
We had a project at KEF to investigate rooms acoustics where we installed over 30 speakers in an anechoic chamber fed from a 32 channel custom built DSP system. Each of these speakers was an active, bi-amplified, current-output amplifier configuration. I built all the drivers for these, along with the active EQ. It was exceedingly complex to measure and individually adjust the active xover for each driver, made much worse because of the current drive.
All things considered, through the years I have come o the conclusion that it is better to design a good driver, than to "fix" it with special amplification.
This applies especially in the case of motional feedback!
Andrew
I'm going to blame the 7mo-old who kept me awake from 2-6am this morning, but I'm having a hard time thinking today. What is the distortion mechanism treated by raising Zout? Le(x)? Le(i)?
I largely agree, but there are an interesting subset of existing low-Qt drivers that become more interesting for a range of alignments with a modest amount of additional Zout (like the Aurasounds I already mentioned).
IMO, Zout should help achieve the desired, damping--it need not be either a micro-ohm or a kilo-ohm.
I largely agree, but there are an interesting subset of existing low-Qt drivers that become more interesting for a range of alignments with a modest amount of additional Zout (like the Aurasounds I already mentioned).
IMO, Zout should help achieve the desired, damping--it need not be either a micro-ohm or a kilo-ohm.
There is a difference among all of the variations. The issue is when listening through other devices, it's ot going to mean anything. Only the person actually listening to them directly can make a judgement that means anything.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.