I don't think this is correct. Because of the high impedance in the circuit, the back EMF is not damped in any way, and the cone just keeps on flapping at it's own resonance frequency, undamped.
The non-linearities definitely turn up in the acoustic domain!
Jan
If I recall correctly we listened to both a true full-range (no EQ) and to a 2-way where a full-range driver had some help in the LF (so with EQ). The drivers were Mark Audio Alpair in some of Dave's (Planet10) enclosures.
Tom
Hi Jan,
yes, forgot to add this in the sentence you quoted. We got reduced current so the damping is also reduced or eliminated, at driver resonance with current drive amp. I think this is also what makes current drive systems marginal in comparison to voltage drive stuff, amps and speakers. Also it is one reason why I think its not very useful to think too black and white about the voltage vs. current amplification as both have pros and cons. Damping could be also done by the cabinet alignment for example. Definitely driver resonance needs to be damped somehow as the resonance would make huge/unwanted excursion which is the prerequisite for motor related distortions to realize into acoustic domain.
I think it is more productive to think about the impedance and what are the pros and cons of each type of drive and why, as system level issue. For example voltage drive amp with series inductor between it and driver leaves low impedance for the driver resonance and higher impedance for distortion current, best of both. But, this of course works as low pass filter so not applicable in every scenario.
Last edited:
I'm just a hobbyist but intuition says because its not compatible with 99% of loudspeakers... Perhaps loudspeakers are more economical to make voltage amplifiers in mind, impedance and box (alignment) can be almost what ever (costs cut) and paired with almost any voltage amplifier gives predictable frequency response.
Last edited:
Yes. The FE127e has a similar impedance response and we could dial in the variable tranconductance amp so that just the right amount of extra bottom & top were added.
dave
Tom. i am sure you have also heard LM3886 amplifiers that sound bad and i know you make some that sound fantastic.
A sample of one only tells us about the specific amplifier you heard.
Speaker matching when using a current amp (or a near current amp) becomes much more critical.
dave
The electronics stoff is a bit over my head, but in a current amplifier the EMF is zero so hard to damp it down any further.
dave
That's what I thought. We know that with a voltage drive amp, the output impedance determines how well the EMF is absorbed by the amp so as to damp the speaker motion resulting from its own resonance. Damping factor - and ideally the output impedance is zero so the speaker EMF is 100% absorbed and the speaker tightly damped.
With current drive, the output impedance is (ideally) infinite and there is no damping of the speaker movement by absorbing the EMF.
You can see this in for instance how the speaker freq response follows the speaker impedance response, necessitating (sp?) response equalisation.
A speaker under current drive will ring like a bell when excited mechanically. The same when excited electrically due to for instance a transient signal.
No?
Jan
With current drive, the output impedance is (ideally) infinite and there is no damping of the speaker movement by absorbing the EMF.
You can see this in for instance how the speaker freq response follows the speaker impedance response, necessitating (sp?) response equalisation.
A speaker under current drive will ring like a bell when excited mechanically. The same when excited electrically due to for instance a transient signal.
No?
Jan
So one should build speakers with Q=0,5 critical damped. Then no "flapping" should occur. Smaller boxes, less material, tighter bass - all great properties.
I wish there were a proper good power amp alternative available for a true CD amp, without any resistors i series, based on modern technology and components.
Can it be made in class D?
//
Jan, the mechanical Q of a loudspeaker is only 2 or 3, and it can be acoustically damped further if needed, so that would be a very poor bell.
You do need equalization to get rid of the peak in the response, but that just requires a simple filter. Getting rid of the distortion and compression you get under voltage drive is more difficult.
You do need equalization to get rid of the peak in the response, but that just requires a simple filter. Getting rid of the distortion and compression you get under voltage drive is more difficult.
What is the EMF thru an infinite resistor? The math i saw was pretty convincing.
But in practice getting that kind of Rout is not acheivable, like low Rout starts to become so close (exponential) anything over damping of about 20-40 becomes moot, once one reaches sufficiently high output impedance the voltage across the amplifier's output Z is practically zero.
dave
But in practice getting that kind of Rout is not acheivable, like low Rout starts to become so close (exponential) anything over damping of about 20-40 becomes moot, once one reaches sufficiently high output impedance the voltage across the amplifier's output Z is practically zero.
dave
Or a loudspeaker specifically designed for current or near current amplifiers (ie FE206 horn), but as the loudspeaker’s impedance varies more widely EQ can become a positive contributor.
dave
I'm sure the feedback-masters at Purifi could do it, with a voltage feedback after-the-output-filter class D design and some sort of current sense scheme, like a sub-ohm resistor. Probably comp for the output inductor hysteresis too...
I did it using line matching transformers. Think 70V winding driving the speaker, the amp driving the 8 Ohm winding through a multi-ohm series resistor, which converts the amp's output voltage to a current into the transformer winding. Presumably, a current of less magnitude due to the turns ratio comes out the 70V winding. It's horrendously inefficient, but lets you experiment with the idea using a very simple design.
Directly drove a FR with that arrangement, so no xover impedance to worry about. (I believe it did like the tranny being close to the driver; i.e. no length of speaker cable being driven by the HiZ winding) I also crossed via in-amp DSP well above the FR's Fs, so no behavior anomaly there to be concerned with. Unsure if I achieved anything sonically, but at least I gave it a shot.
Then, I let things get too far when I started thinking how class A tube amps always have a magnetic bias and most P-P tube amps are not perfectly biased to zero DC magnetic flux in the OPT, so I started experimenting with injecting current into the primary winding to throw the quiescent flux one way. Thought I heard something. But couldnt pass my own A/B test with a controlled relay arrangement from my listening position.
I tore it all out and am now back to straight voltage output from the class D amps. Started playing with a Polk SDA arrangement...
IME, pure current drive around resonance doesn't work well, especially not for woofers where any resonance peak is fully within the desired passband.
- While the resulting response peak can easily be EQ'd out, the woofer still reacts undamped to any external excitation, and that includes its own motor/suspension distortion etc... that leads to spongy "one-note bass", there's always some ringing on the resonance frequency especially after a large signal event.
- Even worse, weakly damped mechanical systems with large nonlinearities (which speaker drivers do have at higher excursion) very often exhibit bifurcation, leading to jump resonance (see attached paper). Sounds extremely awful.
Last edited:
