Are there technical ideas for designing a loudspeaker especially for current driven amplifiers? This could be special CFA amps or voltage driven amps with resistors in line.
I have some propositions on that topic.
Current drive with voltage driven amps can be realized with the "simple resistor method" as described by Esa Merilainen (website current drive info).
Nelson Pass tested this method with higher value resistors like 50 ohms. Merilainen describes first benefits putting a resistor with the same R like the driver 4 / 8 ohms in line with it.
Many voltage driven amps work with an impedance like 3 to 4 ohms giving most wattage.
A loudspeaker - for example a fullrange driver should have 1- 2 ohms voice coil resistance so that 2-3 ohms added push the amp into "beginning" current drive.
Putting several drivers parallel as a line driver would do the same concerning impedance but not everyone wants a line driver array.
Voltage driven amps with the simple resistor method can be a choice with loudspeakers having a Faraday / copper ring on the pole piece which brings distortion already down. Adding a resistor in line makes the sound better - very close to real CFA (current feedback amplifiers) amplifiers what I found out in a listening test.
Often Multi way speakers have resistances in line with tweeters and midrange drivers. That is the reason why many do not feel the need to try out current feedback amplifiers.
CFA amps would profit from special loudspeakers which have a lower mechanical Q. As the bass resonance is less controlled in this type of amps.
However all my tests with basic current feedback amps did not show any necessity for taming the resonance.
I have some propositions on that topic.
Current drive with voltage driven amps can be realized with the "simple resistor method" as described by Esa Merilainen (website current drive info).
Nelson Pass tested this method with higher value resistors like 50 ohms. Merilainen describes first benefits putting a resistor with the same R like the driver 4 / 8 ohms in line with it.
Many voltage driven amps work with an impedance like 3 to 4 ohms giving most wattage.
A loudspeaker - for example a fullrange driver should have 1- 2 ohms voice coil resistance so that 2-3 ohms added push the amp into "beginning" current drive.
Putting several drivers parallel as a line driver would do the same concerning impedance but not everyone wants a line driver array.
Voltage driven amps with the simple resistor method can be a choice with loudspeakers having a Faraday / copper ring on the pole piece which brings distortion already down. Adding a resistor in line makes the sound better - very close to real CFA (current feedback amplifiers) amplifiers what I found out in a listening test.
Often Multi way speakers have resistances in line with tweeters and midrange drivers. That is the reason why many do not feel the need to try out current feedback amplifiers.
CFA amps would profit from special loudspeakers which have a lower mechanical Q. As the bass resonance is less controlled in this type of amps.
However all my tests with basic current feedback amps did not show any necessity for taming the resonance.
You post shows that 'current drive' amps have a HUGE range of Zo; from 4R to 50R. NONE of these are current drive amps in my book which would be Zo > 1k from 20 - 20k Hz.
Speakers would sound COMPLETELY DIFFERENT. cos widely different frequency responses on each of your pseudo current drive amps ... as Dave sorta hints.
I'm still waiting for a true Current Drive amp that is easy to build to appear. Then we can discuss the beast you've asked for.
Speakers would sound COMPLETELY DIFFERENT. cos widely different frequency responses on each of your pseudo current drive amps ... as Dave sorta hints.
I'm still waiting for a true Current Drive amp that is easy to build to appear. Then we can discuss the beast you've asked for.
True current drive maybe easily obtained by means of a regular voltage amp in a closed-loop configuration with a non-intrusive current sensor to pickup the speaker current. Some of these have precision and linearity suitable for audio purposes with bandwidths often exceeding 100kHz and current ranges upto 100A.
https://www.allegromicro.com/en/pro...-to-fifty-amp-integrated-conductor-sensor-ics
which is exactly the reason why the amplifier doesn't matter much as long as it's in good working order
if speaker (driver) is allowed to function to it's true merit, assuming driver was designed for voltage drive dominating circuit impedance affecting it's own frequency response, it will emit it's own impedance/current related distortion acoustically as well. High circuit impedance will reduce drivers effect on circuit impedance, which is kind of the whole current amplifier thing, which also means no real current amp is needed: Armed with this knowledge one can tailor circuit impedance in series with the driver to reduce distortion and have desired frequency response, with off the shelf drivers and amplifiers. It's true that if a driver and amplifier were designed together, or current amplifier with active filtering was used, the performance could be even better.
If one must have current amplifier, then it's output impedance must be known to design a suitable driver for it to give flat acoustic response, or use DSP to vary current without reducing impedance in series with the driver. If using passive network to EQ acoustic response with current amplifier, it now reduces the circuit impedance in series with the driver and the net result is that the acoustic performance (distortion) is not much better than with low output impedance amplifier with other kind of passive network, and was relatively unnecessary exercise.
yes it is exposed to thermal variation, but the high circuit impedance outside the driver makes the current through voice coil constant. This means voice coil impedance, including resistance going up with heat, doesn't ideally affect acoustic sound at all. That's the whole point of current amplifier, the amplifier (impedance outside of the driver) dominates current through voice coil, not the voice coil itself. Now voice coil impedance can vary all day long and you wouldn't notice any change in acoustic domain.
With voltage amplifier, where the driver voice coil dominates whole circuit impedance, the thermal increase in resistance would reduce current (with constant voltage) in the circuit, which means reduced acoustic output. This also means the driver protects itself from meltdown by regulating current, if voice coil burns it's due to operator not noticing the compression and turning voltage up. On current drive you wouldn't notice anything in acoustic output before smell of magic smoke.
With voltage amplifier, where the driver voice coil dominates whole circuit impedance, the thermal increase in resistance would reduce current (with constant voltage) in the circuit, which means reduced acoustic output. This also means the driver protects itself from meltdown by regulating current, if voice coil burns it's due to operator not noticing the compression and turning voltage up. On current drive you wouldn't notice anything in acoustic output before smell of magic smoke.
Last edited:
Yes, if the current is controlled, that means that the amp is compensating for the increased resistance of the VC by putting more voltage across it causing more heat (which the amp compensates religiously), creating a vicious cycle that could possibly result in damage.
This may not happen if the VC copper has reached steady state (thermal equilibrium) when the heat gained is equal to the heat lost and the temperature, in turn, remains constant thereafter.
Alternatively, a zero temperature coefficient material could be used for the VC. And with current drive, it doesn't matter even if such a material (if it exists) has slightly higher resistivity when compared to copper or aluminium, as long as it doesn't change its value.
This may not happen if the VC copper has reached steady state (thermal equilibrium) when the heat gained is equal to the heat lost and the temperature, in turn, remains constant thereafter.
Alternatively, a zero temperature coefficient material could be used for the VC. And with current drive, it doesn't matter even if such a material (if it exists) has slightly higher resistivity when compared to copper or aluminium, as long as it doesn't change its value.
Last edited:
Exactly, and if system is working near it's limits, it's sound is not very good regardless of amplification. For good sound one would need better fit, bigger, speaker system anyway. If sound quality is a factor then the thermal effect is almost irrelevant in this regard. If a system needs to be run at it's limits then voltage amplifier would save money, less broken drivers.
Last edited:
Back to the original question (call it 'quasi / semi-current drive', or even 'high output impedance' if you prefer ), then you're essentially arriving at what the pioneers were working with by default, and in fact took further by providing variable output impedance for system tuning in some of their more advanced / expensive designs.
Assuming you don't have that luxury available, and you don't know the precise output impedance of the system + any relevant series R in circuit from wire loop, connections etc., then as Dave says what you need is a speaker system with a flat impedance, which if you chuck sufficient input Zobels at it, and you have close tolerance, consistent drive units & enclosures, isn't actually all that difficult. It just needs more components in the filter; I did this for the last version of the Edingdale commercial speaker (currently out of production but may return next year) as most buyers were wanting to use it with SET amplifiers. The Elsinore does more or less the same, as does, to an extent, any speaker where the designer has accounted for SETs etc. by providing an optional input LCR Zobel to flatten the impedance rise at / around the crossover frequency / frequencies. Nothing particularly mysterious here.
), then you're essentially arriving at what the pioneers were working with by default, and in fact took further by providing variable output impedance for system tuning in some of their more advanced / expensive designs.Assuming you don't have that luxury available, and you don't know the precise output impedance of the system + any relevant series R in circuit from wire loop, connections etc., then as Dave says what you need is a speaker system with a flat impedance, which if you chuck sufficient input Zobels at it, and you have close tolerance, consistent drive units & enclosures, isn't actually all that difficult. It just needs more components in the filter; I did this for the last version of the Edingdale commercial speaker (currently out of production but may return next year) as most buyers were wanting to use it with SET amplifiers. The Elsinore does more or less the same, as does, to an extent, any speaker where the designer has accounted for SETs etc. by providing an optional input LCR Zobel to flatten the impedance rise at / around the crossover frequency / frequencies. Nothing particularly mysterious here.
^^ Yeah that is a way to get the frequency response rather constant regardless of amplifier output impedance. What it means, both ideal voltage and ideal current amplifier would result quite similar acoustic frequency response with a particular driver, but also quite similar distortion performance, because due to the passive network that makes impedance constant for the amplifier it also makes rather constant impedance in series with the driver as well, to have it's varying impedance manifest varying current through voice coil and into acoustic domain regardless of amplifier used. The distortion performance might be bit better than with voltage amp and conventional crossover, but not as good as the current amplifier without any shunt network. Homogenization both ways, and the benefit is mostly in utility.
Zobels and what not are a good thing in a way, in usability perspective, and good utilization of circuit impedance for what it is, ability to use any amplifier.
As per opening post, if exact output impedance of used amplifier was known, and a driver was designed with that so that no zobels or other shunts are needed to hit some acoustic frequency response , then there could be real difference in performance where driver motor does not make acoustic distortion much at all. Or just use current amp with off-the-shelf driver and deal with frequency response before power amp, with DSP for example. Or use modern very low distortion driver so there is no special benefit of current amplifier, and I'd imagine no real difference in sound. The whole current amplifier thing really is kind of a curiosity, and not a real issue. Amplifier can be abstracted away, use low distortion driver, or reduce distortion of any driver by increasing series impedance.
Zobels and what not are a good thing in a way, in usability perspective, and good utilization of circuit impedance for what it is, ability to use any amplifier.
As per opening post, if exact output impedance of used amplifier was known, and a driver was designed with that so that no zobels or other shunts are needed to hit some acoustic frequency response , then there could be real difference in performance where driver motor does not make acoustic distortion much at all. Or just use current amp with off-the-shelf driver and deal with frequency response before power amp, with DSP for example. Or use modern very low distortion driver so there is no special benefit of current amplifier, and I'd imagine no real difference in sound. The whole current amplifier thing really is kind of a curiosity, and not a real issue. Amplifier can be abstracted away, use low distortion driver, or reduce distortion of any driver by increasing series impedance.
Last edited:
What exactly is it? The title says current drive, which is what you get when you put a resistor in series and take feedback from that. But this is not the same as a current feedback amplifier and that is again different from a current driven amplifier...
A while ago I couldn't find any low-value resistors lying around to connect my speakers to, but I did manage to dig up a pair of 6.3V incandescent lamps among the stegosaurus bones. The lazy flicker was vaguely correlated with the loud parts, but I could not discern any obvious distortion or compression effects.
Is there a description how to use this technique somewhere?
I dream of converting a Behringer NU3000dsp amplifier to current drive for example.
@kgrlee he complains about availability. It's my desire too not to have to work myself as a "non-electronic" converting chip amps to current drive.
But even simple chips like TDA2003 bring great sonic benefits with fullrange drivers. So I am not in urgent need of a ready made amp. I do them for little money myself now.
Presumably, you've built an amplifier with one of these sensors and measured Zo over 20 - 20k Hz. What did you find? Was it unconditionally stable with load?
@Havoc
Simple resistor method:
https://www.current-drive.info/9
Besides is basic current feedback method I realized in this thread:
https://www.diyaudio.com/community/threads/chip-amp-modification-to-current-drive.389985/
Simple resistor method:
https://www.current-drive.info/9
Besides is basic current feedback method I realized in this thread:
https://www.diyaudio.com/community/threads/chip-amp-modification-to-current-drive.389985/
That is using an amplifier with voltage feedback but by deriving the voltage for the feedback from the current in effect making the output of the amplifier a current proportional to the input voltage. This is current drive. But the feedback is still a voltage.
Current feedback as understood in electronics is where the current in the feedback loop is what controls the amplifier. The output of the amplifier can still be a voltage or a current. Mostly used in video amps for high frequencies. https://en.wikipedia.org/wiki/Current-feedback_operational_amplifier This is what all manufacturer of electronic components will understand with current feedback.
A current driven amplifier is where the input of the amplifier is the current into its input terminal. The output of the amplifier can be a voltage or current proportional to this input current. This is what you will find in an LM37000 where the + and - input are low impedance and it is the current into the opamp that controls the ouput.
Terminology is important because these describe very different things.
Current feedback as understood in electronics is where the current in the feedback loop is what controls the amplifier. The output of the amplifier can still be a voltage or a current. Mostly used in video amps for high frequencies. https://en.wikipedia.org/wiki/Current-feedback_operational_amplifier This is what all manufacturer of electronic components will understand with current feedback.
A current driven amplifier is where the input of the amplifier is the current into its input terminal. The output of the amplifier can be a voltage or current proportional to this input current. This is what you will find in an LM37000 where the + and - input are low impedance and it is the current into the opamp that controls the ouput.
Terminology is important because these describe very different things.
Yes, they're all different things but the OP has tried to have a common name for two things 1) current controlled voltage amplifier and 2) voltage amplifier with added output impedance, and somehow ended up using the term CFA for convenience.
Just as in your chip amp thread, wire the speaker through the current sensor (instead of the sense resistor) and use the sensor output as feedback signal for the error amplifier. Since the sensor output is isolated, this works for both BTL and single-ended amps. Note that you may need to add compensation to the forward path of the amplifier.
No, I haven't built or measured anything like that, but am reasonably sure that it could be done without any trouble. In principle it would just be an 'AC battery charger'. Stability needs to be addressed by adding compensation if necessary.
However, the main problem is the thermal runaway that several others have also mentioned.
Last edited: