That's a good one Gerhard, however the AC voltage of the primary voice coil would be the same as the output voltage of the amplifier (in case of active filters of speakers), which is already corrected by amplifier feedback and which in case of properly designed amplifier has negligible distortion. Would you like to make a correction for a speaker cable voltage drop?
(I am joking of course)
When the coils are tightly coupled - and bifilar in a magnetic short is about as close as it gets - then the amplifier & cable affect both coils equally.
That "which is already corrected by amplifier feedback and which in case of properly designed amplifier has negligible distortion" sounds much like the "feedback runs and runs and..." fault.
In a previous life, some 35 years ago, I tried to do an active speaker design after St°ahl, that goes into the opposite direction: perfect voltage drive. You need an amplifier with negative output impedance to cancel the ohmic resistance of the voice coil. Then you can do funny things to the TS-Parameters: synthetic mass etc. and you can achieve a very low frequency response at the expense of efficiency. But the first acoustic watt is enough, at least at home.
I gave up when I noted that the voice coil resistance is a moving target. Someone kicks a drum and the VC resistance soars high. You cannot ignore the TC. That makes the system completely time-variant. Maybe one could measure/model the effects with today's technology.
After some detours I decided finally to shove some grand to B&W, never regretted it.
Gerhard
(the circle belongs above the a in Stahl, probably from Sweden.)
That "which is already corrected by amplifier feedback and which in case of properly designed amplifier has negligible distortion" sounds much like the "feedback runs and runs and..." fault.
In a previous life, some 35 years ago, I tried to do an active speaker design after St°ahl, that goes into the opposite direction: perfect voltage drive. You need an amplifier with negative output impedance to cancel the ohmic resistance of the voice coil. Then you can do funny things to the TS-Parameters: synthetic mass etc. and you can achieve a very low frequency response at the expense of efficiency. But the first acoustic watt is enough, at least at home.
I gave up when I noted that the voice coil resistance is a moving target. Someone kicks a drum and the VC resistance soars high. You cannot ignore the TC. That makes the system completely time-variant. Maybe one could measure/model the effects with today's technology.
After some detours I decided finally to shove some grand to B&W, never regretted it.
Gerhard
(the circle belongs above the a in Stahl, probably from Sweden.)
Last edited:
Just a thought, would not the transformer coupling be only between L1==>L2, excluding Re drop?
The idea is to remove the microphic (velocity, "back-emf") voltage from the whole equation which renders a result very similar to current drive basically, and JN confirmed this.
IMHO it's a sort of "indirect" current drive, the current isn't impressed by the amp directly, rather a voltage is forced accross the isolated transfer impedance (static VC impedance) which is quasi-constant and therefore transfers the voltage into a current which nicely corresponds to that voltage, to first order at any rate. The isolation of the static transfer impedance by subracting an exact copy of the microphonic voltage is a nice new trick in the book that I never saw before nor did I ever think about something like this. The exactness of the copy is guaranteed by design of the bifilar VC alone and not determined by circuit components, doesn't need trimming etc. That's the clever idea behind it, at least as I understand it. The second coil isn't used to help extract the velocity voltage, rather it does the exact opposite, it removes that voltage. Also, it lowers/eliminates many secondary terms, like Le influences etc. The effective V->I transfer impedance will be really close to a stable linear resistance, the plain coil wire's resistance. That's the "100% in-phase" argument John has made.
If we remove the back-emf voltage from the transfer we also remove any effects of its nonlinearities, that's whole the point. But we also loose any electrical damping...
If this "indirect current drive" actually gives better results overall than true current-drive remains to be seen and given that it's more complicated to implement we'd like to see quite an improvement to justify the effort. Current drive, be it direct or not, has its own set of issues and these still need to be adressed, notably how to implement proper damping at resonance (if required by chosen acoustic loading)? We could take the terminal voltage in that frequency range and mix in into the amp's feedback just as we would do with a current drive amplifier core... or find a way to extract the true microphic voltage and use that to degenerate the apparent output impedance
IMHO it's a sort of "indirect" current drive, the current isn't impressed by the amp directly, rather a voltage is forced accross the isolated transfer impedance (static VC impedance) which is quasi-constant and therefore transfers the voltage into a current which nicely corresponds to that voltage, to first order at any rate. The isolation of the static transfer impedance by subracting an exact copy of the microphonic voltage is a nice new trick in the book that I never saw before nor did I ever think about something like this. The exactness of the copy is guaranteed by design of the bifilar VC alone and not determined by circuit components, doesn't need trimming etc. That's the clever idea behind it, at least as I understand it. The second coil isn't used to help extract the velocity voltage, rather it does the exact opposite, it removes that voltage. Also, it lowers/eliminates many secondary terms, like Le influences etc. The effective V->I transfer impedance will be really close to a stable linear resistance, the plain coil wire's resistance. That's the "100% in-phase" argument John has made.
If we remove the back-emf voltage from the transfer we also remove any effects of its nonlinearities, that's whole the point. But we also loose any electrical damping...
If this "indirect current drive" actually gives better results overall than true current-drive remains to be seen and given that it's more complicated to implement we'd like to see quite an improvement to justify the effort. Current drive, be it direct or not, has its own set of issues and these still need to be adressed, notably how to implement proper damping at resonance (if required by chosen acoustic loading)? We could take the terminal voltage in that frequency range and mix in into the amp's feedback just as we would do with a current drive amplifier core... or find a way to extract the true microphic voltage and use that to degenerate the apparent output impedance
Yes, also saw this in my experiments. Once you compensate more than 7/8 of the VC resistance you'll end up with negative drive impedance when VC resistance increased just 10% or so.
This is addressed with a servoed bridge invented by David Birt (see pointer that @Soundbloke gave in the past). It work even better with a dual VC driver because much less fitering is needed to extract the DC voltage (or current) which represents VC temperature which modulates one resisitive leg of the bridge. The AC component of that sensed entity can be made common-mode with a dual VC. The AC component itself (true velocity voltage) which is what we want to regulate upon in this application, still only is as good as the bridge can extract it. It uses a replica of the static driver impedance which can only be good to first order. And with the sensor being limited at large signal (BL drop) itself any control loop that is "too tight" yields heavy overshoots. The measured velocity is lower than real, hence the control drives more current into the VC until measured velocity matches. Actual velocity is too high, then. Once the coil starts leaving the gap this gives a nice clusterfuck.
Last edited:
Exactly. Still it may render less overall distortion especially when degenerated to fit a driver's damping needs. Unless we use any sort of MFB which largely depends on the linearity and dynamic range of the sensor mechanism we can only better the motor design in order to improve something. An addition to the VC would be useful: add sections above and below the main coil and use that for damping only, by terminating it with the same apparent drive impedance that the main coil sees. By this we can control the ill-effects under large signal / excursion overdrive much better, the damping required by the acoustic loading will be kept intact even if main Bl has gone down. Eddy current brake of sorts.
Some driver from AudioTechnology of Denmark can be ordered with shorting ring sections on the former above and below the coil winding which just do that, keeping the driver safe. Some PA drivers with their symmetric motors even inject counterfource once the VC leaves the gap...
Last edited:
No, the two coil voltages will not be the same. The drive coil is the one which supports all the energy transfer.
As a simplistic model, consider each coil as an ideal Ls/Rs. If you use a meter to measure this inductor, the meter will give you the Ls and Rs value. The Ls is the measure of all signal energy returned to the meter, it's current is 90 degrees out of phase with the voltage drive. The Rs is the portion of the current that is in phase with the drive. And, it reflects the energy that is NOT returned to the meter.
If you measure an air core inductor, watch the Rs (dissipation) as you put a conductor near the coil. Rs will increase, showing you the power loss due to the eddy dissipation in the conductor. In the case of a voice coil, the Rs is due to all losses in the coil, IR loss of the wire, eddy loss in the irons, and most importantly, acoustic "loss", or more precisely, energy going into the air.
Within a bifilar doc, both Ls components couple magnetically 100%. The Rs components are totally independent.
If you drive one coil with power, it will have a terminal voltage dependent on the sum of the Ls reactance and the IR drop of the Rs. The second coil will see the exact same flux, so it's Ls reactive voltage will be the exact same as that of the first coil. However, it has no current within it, so has no direct drive loss.
It will, however see the eddy loss and the Le(x) variation due to excursion.
The difference between the two coils in my configuration is the dissipative component of the energy into the speaker.
Jn
Your understanding is quite accurate.
However, I'm not sure if damping is lost entirely.
And as to simplicity, I require zero additional electronics, just redirect the feedback of the amp.
Jn
Except for some residue I see no way how your approach could implement damping. Damping means the back-emf has a lever to work upon, transfering the voltage back into a current. If you regulate effectively to zero current with zero input signal at the amp the lever is removed.
Re simplicity, you need to recone a driver with that added sense coil. Really good DVC midbass drivers, let alone with bifilar coils, are scarce if not completely non-existent, some car woofers mainly, that's it.
We've talked about 18sound's AIC before, a secondary stator coil which helps reduce the distortion still present with current drive with an ordinary motor, an active flux compensator (yeah, finally ;-) Any thoughts on this, as there are drivers available right out of the box that would be applicable to HiFi midbass duties?
Active flux capacitor????
Oh, wait..compensation..
Damping. Not sure, I've also been down your train of thought road...and you have viable points to consider.
Damping historically uses the amp stiffness as a wall to brace against. In this case, the brace is loosy goosy (sorry for the really technical words there, just try and keep up
) but the wall is moving to comp through the brace. I've done this in motion control, but the time constants are about 4 orders of magnitude different.It will be great to see what 1audio finds, his driver in open air is a good test of this.
I personally would have to recone. However, a true Manu would already have the winding capability for bifilar coils if they make DVC speakers, this simply puts a smaller gauge on one of the feed spools. Yes, it's not quite that easy, but I've already worked this problem out mechanically for doing it in my basement.
The most difficult aspect of the winding is actually the climb of the wires from layer 1 to layer two. How the smaller Gauge transits to layer two without getting crushed by the larger wire is the question. I haven't worked out how to do that simply on an automatic machine. Complex, yes it's trivial, by hand, easy as well, but simple and cheap, well that's why they pay the engineers the big bucks. (Well, so I've heard..)
Jn
But this system does not control current per se, it is controlling the voltage across the in phase resistance. Any unintended cone motion will still try to impress a voltage on that resistance, so I suspect there will be control.
While conceptually I've understood this E/M design for six years, the attempt to use it with a dual voice coils speaker is exactly one week old. I've about ten minutes more than all of you at thinking this through.
The near future will be fun. I hope this is actually an advance.
Jn
- Status
- Not open for further replies.