Pure current drive at/around system resonance(s) is fraught with problems and that is true both for sealed, ported or any other type of box (even open-baffle).
Current drive removes any electrical damping so only mechanical damping remains in place. It is not widely understood that the electrical damping is a local feedback mechanism and as such it reduces the distortion from the suspension (notably when the suspension is very progressive) as long as the motor force factor is still sufficiently linear. @lrisbo pointed that out in post #913. In the extreme case of "infinite" damping (output resistance = negative of driver's DC resistance, resulting in a Qes approaching zero) we have true motional feedback, velocity feedback in this case (giving a +6dB/oct slope over a wide range, thus the apparent "lack of bass" from ultra-low-Q drivers).
For this exact reason -- the intrinsic corrective feedback -- some compressions drivers work best around and below resonance with maximized damping (using negative output resistance) as their simple suspension is very nonlinear but the underhung motor is very linear and their high efficiency practically prevents thermal compression issues (which are emphasized with maximized damping).
Another issue with current drive is the need for a quite radical EQ notch/dip filter around resonance which is problematic as the peak in the frequency response is not stable but drifts and shifts with operating conditions.
And for ported types to work properly, the effective port Q factor must lie in a narrow range of values for the Helmholtz resonator to work as intended. Current drive decreases this Q way below useful values (whereas maximized damping results in too high port Q, giving a too narrow port response).
Finally, another less known effect is the strong tendency for chaotic behavior that occurs with any low-damping mechanical systems which have significant nonlinearity, the so-called jump resonance bifurcation phenomenon which sounds especially nasty. With typical HiFi woofers with soft suspensions you will most likely have severe jump resonance and this is one very nasty sounding distortion. Ported speakers also can have another similar type of irregularity (coil popping out aka dynamic DC offset) and the lower the damping and the softer the suspension the more so.
Summary: Around system resonance current drive will not work too well (or not at all sometimes) with conventional drivers and box types designed for voltage drive and the associated electrical damping.
perfect answer. to illustrate the motional feedback effect: the current distortion sky rockets typically around fs so that the distortion level may exceed the distortion measured acoustically. this is the result of the feedback that makes the current nonlinear in order to make the motion linear. This requires of course that the Bl(x) is constant.
I forgot to add that with current drive any EQ, even if it were perfect and stable, does nothing to prevent severe ringing at the resonant peak frequency for any external excitation. The driver's own distortion has to be considered as external here, notably recovery from excursion overdrive.
I broke a toe over the weekend 
and to help with the pain went over all 47 pages of this thread
I'm surprised that I pontificated extensively in da early days ... mostly from the correct orifice
I answer several of KSTR's (& other guru's) points in detail in my earlier posts.
and to help with the pain went over all 47 pages of this thread I'm surprised that I pontificated extensively in da early days ... mostly from the correct orifice
I answer several of KSTR's (& other guru's) points in detail in my earlier posts.
Has anyone posted THD measurements showing increased THD around/below resonance with current drive cf voltage BUT WITH FREQUENCY RESPONSE EQUALISED FOR EQUAL LEVELS? The impedance peak means current drive results in a LOT of extra drive near resonance. This extra drive may be what's causing increased THD
The answer to this is Erik Stahl's ACE bass which was discussed in earlier posts. This uses electronic voodoo to generate new stable TS parameters. It has most of the goodness of current drive, lower THD bla bla .. with one exception .. VC heating compression which current drive alleviates. Biggest advantage is that it can be used with ported boxes.
This doesn't only happen with ported speakers. Dr. Don Barlow wrote an Engineering Memo for us in da early 70s. He may have an AES paper on the subject. The issue is that as the coil moves out of the gap, the instantaneous inductance drops and VC current increases instantaneously. This is unstable and the VC jumps out with voltage drive. This leads to the non-intuitive solution where having a non-linear suspension that gradually limits REDUCES THD at high levels. I think I say more in my earlier posts. It doesn't happen with current drive.
Last edited:
Anyone have an illegal copy of Mills & Hawksford -
Probably the most important paper on Current Drive and required reading for would be gurus or pseudo gurus on the subject ... though I don't agree with da supa complicated solutions. Nice but complex true current drive amp.
I've lost my copy several HD crashes ago. As a professional beach bum, I can no longer afford to be an AES member or buy papers
Distortion Reduction in Moving-Coil Loudspeaker Systems Using Current-Drive Technology
https://www.aes.org/e-lib/browse.cfm?elib=6099Probably the most important paper on Current Drive and required reading for would be gurus or pseudo gurus on the subject ... though I don't agree with da supa complicated solutions. Nice but complex true current drive amp.
I've lost my copy several HD crashes ago. As a professional beach bum, I can no longer afford to be an AES member or buy papers
Last edited:
No, you remembered rightly. There is a later paper with the fully elaborated circuit: "Transconductance Power Amplifier Systems for Current-Driven Loudspeakers." I did not find it on my computer but I do have hardcopy. I will create a PDF.
The authors used a transconductance stage driving a common-base stage to buffer the load. Floating power supplies, similar to that employed by Vanderkooy, are used. A Hafler Transnova power amp would be a convenient platform to built upon as it has dual supplies and uses MOSFETs in common source.
The authors used a transconductance stage driving a common-base stage to buffer the load. Floating power supplies, similar to that employed by Vanderkooy, are used. A Hafler Transnova power amp would be a convenient platform to built upon as it has dual supplies and uses MOSFETs in common source.
Ouch! Wish you a quick recovery.I broke a toe over the weekend
Good point. It is paramount to any comparison, be it a measurement or a listening session, that the SPL response must be exactly the same for any kind of drive option, to better than 1dB (+-0.1dB if possible). For true current drive this is comparatively easy with today's tools like convolvers (to realize any kind of FIR filter): you simple record the current's impulse response with a sense resistor in normal voltage drive (with any EQ applied to achieve the required SPL target, and at low levels to keep things linear. This means a large sense resistor for good S/N ratio which should be inside the feedback loop of the voltage drive amp to remove its effect) and then use that IR as a convolution kernel for the current drive. In other cases (like mixed impedance drive) I recorded the driver terminal voltage impulse response for both amps, standard voltage drive with EQ as above) and the test amplifier and obtained the correction kernel by division. A DRC package like Acourate is very useful for this but today we have excellent and free tools like REW and Rephase to do the same thing.
ACE Bass is as a clever mixed feedback giving a varying output impedance vs frequency for the amp and thus has little to do with pure current drive, see https://www.diyaudio.com/community/threads/servo-amp-ace-bass.356977/#post-6269626. Of course, at the time it was an out-of-the-box way of thinking which was novel (and thus worth a patent).
Well, OK.... I wasn't to imply dynamic DC offset only happens with ported (or other setups without an air spring down to DC), but with ported speakers it is observed most often. Current drive avoids that but may give rise to the jump resonance issue, so it's a two-sided sword.
And true, the PA guys have figured it out after many burnt up voice-coils that making the suspension very stiff and rather progressive is the way to prevent the issue... and to get a low enough Fs they had to increase mass, and in turn motor strength to retain decent efficiency. Another solution is partial shorting rings at both ends of the VC acting as eddy current brakes when the coil leaves the gap (https://audiotechnology.dk/faq/, section "What is the difference between Kapton and aluminum voice coils?")
Thanks for this svjeff. When I saw this 30+ yrs ago, I was put off by da complexity cos I kunt reed en rite. It's only now I see how to modify Vanderkooy's amp to get the same results without Mills & Hawksford's BS
Last edited:
Thanks for your contributions too KSTR.
Looks like 'pure current drive' has no general application cos no speakers are designed for it. In that case, I regard current drive as just another electronic voodoo to integrate amps & speakers in special applications. eg subwoofers. I've done a lot of that da previous Millenium in various iterations of my Powered Integrated Super Sub technology ... using current drive, ACE, black magic, linear & non-linear electronics, analogue & DSP EQ, bla bla ...
Only this month I'm thinking about designing speakers for current drive. For this, I need a pure current drive amp. A modified Vanderkooy amp may do the trick
But I'm a real beach bum so don't hold your breathe
Define "defeat."
They are certainly designed so that they can be used with current sources (current-drive) and anybody wants to hear it for themselves, my front door is open.
As Ståhl said himself "due to voice-coil impedance ...an upper bound between 100 to 500 Hz occurs, when AC-bass is used".
/örjan
I meant as much as to return the response to what we'd call flat, despite the impedance variations.
https://www.diyaudio.com/community/threads/the-elsinore-project-thread.97043/post-7123797
Another effect is that drivers share the current with the conjugate, so the non-linearity of the conjugate components is to some degree mirrored in the driver current.
Last edited:
Something is being overlooked. That is not a complete description of what is happening in totality in the Elsinores. Look at it and maybe see that there is something that makes your analysis incomplete. There is a key component that you are overlooking. But that's OK, I will leave it there.
It has been done before and as I can recall, the results were good. I am trying to remember who it was. I think it was Nelson Pass? Also, ribbons should as a whole produce less "voltage distortion" than drivers with coils. In theory, this should result in a more stable impedance under dynamic conditions and less current modulations.
Re-reading Mills and Hawksford is making my headache worse.
They came up with an LCR model of the speaker impedance, so that when it heats up (approximately in sync with the speaker) a voltage divider circuit automatically adjusts ("compresses") the gain, so they can claim to use current drive down to DC, while re-introducing thermal compression as a safety mechanism. And then maybe incorporate a soft-clipping circuit to make the un/compressed transition smooth and imperceptible?
Whyyy?? Pointless complexity for the fun of it is the definition of a Rube Goldberg machine.
They came up with an LCR model of the speaker impedance, so that when it heats up (approximately in sync with the speaker) a voltage divider circuit automatically adjusts ("compresses") the gain, so they can claim to use current drive down to DC, while re-introducing thermal compression as a safety mechanism. And then maybe incorporate a soft-clipping circuit to make the un/compressed transition smooth and imperceptible?
Whyyy?? Pointless complexity for the fun of it is the definition of a Rube Goldberg machine.