So then I get the idea that speaker current drive is not preferred because it drives the speaker 'better' (whatever that means), but because the amp high Zout prevents the non-linear back-EMF to circulate and create non-linear distortion. And you pay for it with a non-flat freq response.
Now this is a clear engineering problem, for which a solution should be available with the benefits but without the penalties.
So let's look for that.
Jan
Now this is a clear engineering problem, for which a solution should be available with the benefits but without the penalties.
So let's look for that.
Jan
I'm not sure what the reasoning behind that is, but it's certainly a pity that the dual crossover filter has a notch in its output impedances at the crossover frequency.
The conventional, voltage driven filter has a large peak in its output impedances at the crossover frequency, so in a conventional set-up, the woofer and tweeter are actually current driven around the crossover frequency.
It's instructive to read the whitepaper on Putzeys' Kii3 active speaker. Current drive is used with specific drivers in specific freq bands, morphing to voltage drive at other places.
A practical test of such a mixed-drive setup is here: https://www.kiiaudio.com/media/GENERAL/docs/reviews/audioxpress012017low.pdf
Jan
A practical test of such a mixed-drive setup is here: https://www.kiiaudio.com/media/GENERAL/docs/reviews/audioxpress012017low.pdf
Jan
Last edited:
There are some AES articles from the 1980's that combine current drive with motional feedback. The motional feedback is mainly there to provide damping.Now this is a clear engineering problem, for which a solution should be available with the benefits but without the penalties.
So let's look for that.
Jan
Right. In theory everything is possible after enough bending.
But how practical is that?
Having a low Qm means an awful lot of mechanical losses.
No thanks.
Anyway to me it still sounds using a technology the wrong way.
In the paper and research I posted before it's very clear that CC works very well for (roughly) 2-3 fs and above. From a theoretical point of view it also doesn't make a lot of sense to correct motional non-linearities this way.
Which is fine because for those lower frequencies we could rather use NFB with a motion sensor.
Thank you again for all your input, special thanks to KSTR for showing up the phenomenon of bifurcation - a thing to keep in mind for current driven projects.
My idea of understanding is usually related to a practical project, so lets go to a real project:
Starting point is a 4way dipol-speaker with folded dipol in a W-frame (2 drivers) at the low end and 2 tweeters
mounted back to back on the high end. This setup is active driven - one very low THD composite amp for each speaker.
All is controlled by a miniDSP 4x10HD. All drivers were chosen for their low distortion (at least) in their operating range.
So the highest distortion actually comes from the speakers themselves. The question that arises: can the distortion of the
complete setup get lower when current driven?
The few transconductance amps i have seen by now all have relative high THD (compared to the composite amp used), so i doubt
there will be any significant improvement when using them. The next step is to find a low THD current amp (or build one by myself).
Bernd
My idea of understanding is usually related to a practical project, so lets go to a real project:
Starting point is a 4way dipol-speaker with folded dipol in a W-frame (2 drivers) at the low end and 2 tweeters
mounted back to back on the high end. This setup is active driven - one very low THD composite amp for each speaker.
All is controlled by a miniDSP 4x10HD. All drivers were chosen for their low distortion (at least) in their operating range.
So the highest distortion actually comes from the speakers themselves. The question that arises: can the distortion of the
complete setup get lower when current driven?
The few transconductance amps i have seen by now all have relative high THD (compared to the composite amp used), so i doubt
there will be any significant improvement when using them. The next step is to find a low THD current amp (or build one by myself).
Bernd
Before switching amplifiers simplest way to try if there is some audible benefits is to add inductors for each woofer needing a low pass. As its dipole there is 6db octave rolloff per driver, use big inductor in series and you'll get flattish response with reduced distortion. If they have bad cone resonances you could add parallel notch (in series with the circuit) but this is two parts more than simple inductor in series. Both increase impedance and reduce distortion current. Just adjust the DSP and you'll have exactly same response as before but with reduced distortion.
In this case.
Go for current feedback for the mids. Maybe the tweeters, but the advantage will be small.
For the woofers just go for CV (aka a regular amplifier)
If you want to go fancy, NFB with a motion sensor can be used.
But keep in min that not only the threshold for distortion for lower frequencies is much higher but also that those pesky room modes are ruining most of the lower frequencies.
??
I assume you know what room models are? If so, you know what I mean.
Even if you will get zero distortion from the woofers, without room correction room modes will ruin that quality in sound completely.
By either creating massive dips and peaks in the frequency response or by the fact that some have quite the delay in them. In other words they can still be heard even when the transient is long gone.
Technically speaking both can be seen as a form of distortion I guess. Room modes themselves are highly dependent on the Q factor of the walls and such. As well as the shape of the room. So to some extend they will be acting different compared to an ideal situation. This could most definitely seen as a form of distortion.
Although that's not what I meant obviously.
Silly response
Any standing wave (which room modes are) creates harmonics.
A room is nothing more than just a pipe.
Put a sound source on a piece of pipe and measure the harmonics.
So in a general sense it's actually very misleading what you're saying yourself.
It all depends how you want to use words like distortion.
Because in general distortion means anything that transforms the signal so it's different than the original.
Phase shift distortion or (group)delay shift distortion is a good example of this, which also doesn't create harmonics.
Yet we still call it distortion.
(yes if we again bend the example a certain way, it sometimes can create IMD, that's not the point)
If people only mean linear distortion, than they need to be specific about it.
It's rather informative than misleading in this context.
It's quite misleading to just omit much more important issues like room modes and care about the few tiny little percentage you win. Maybe at most.
Previous research has already shown that you won't win anything with a CC amplifier.
In fact, if people here are advising speakers with low mechanical Q, you will introduce an awful lot more distortion compared to low Qm speakers.
... speaking about misleading context.......
And yet Pavel (PMA) has shown you can win quite a lot http://pmacura.cz/speaker_dist3.htm. Is it worth the effort is the bigger question..
OK, I stand corrected. In general, when people speak about distortion they mean non-linear distortion, the one that creates harmonics.
Much less often people speak about linear distortion which normally means amplitude/phase distortion or freq response distortion.
So I think we agree on that.
The other thing I wonder is how voltage drive versus current drive does influence room acoustics, or whether that makes any difference in this context.
Jan
There is reverse connection at least, sound pressure in room is "microphoned" by the drivers as already someone mentioned in this thread at some point. I've got no idea how severe /audible this is but nevertheless something that is connected to system distortion performance. Current drive amp, or high impedance load for the driver, should not turn this into any distortion current and acoustic output while the cone would still move (I assume). On driver resonance this could be an issue I think. For example a tweeter on waveguide, thats like vintage hearing aid scoping sounds in the room. I guess such tweeter would benefit electronic damping on its resonance to keep it still in order not to get affected by sound pressure from the other drivers. Coaxial speaker. Anyway, something to wonder about.
Last edited:
Based on this page http://pmacura.cz/speaker_dist3.htm to me a good way to start seems to be the measurement of the Intermodulation a speaker produces when driven with two-tone-signals...