Current drive for Loudspeakers

Arcgothic, a true current drive has infinitesimally hi source resistance, of course impractical. We generate the source impedance by feedback of load current. So let us set it to 40 Ohms and connect a 4 Ohms loudspeaker. The loudspeaker may have 8 Ohms at 20KHz and 16 Ohms at 40 KHz, because of voice coil inductance, but at 80 KHz stay at 16 Ohms and fall to 8 Ohms at 160 KHz, due to voice coil capacitance. At even higher frequencies the cable starts to show, too. Hence response of amplitude versus frequency of feedback signal is bent, and it is hard to keep the amplifier stable. One may come up with one size, which fits all, but active loudspeakers (amplifier and loudspeaker as a unit) are also possible.
 
I thought I'd bump this thread with something I came across recently. This speaker came out 2 years ago.
electronics
Unique to the implementation used in the Kii THREE is a combined voltage/current control loop that goes beyond merely a better amp – it actively improves the distortion performance of the drive units which contributes significantly to the extreme resolution of the speaker.
So are they using current sensing for kind of a MFB or something? I would think that this would effectively yield some sort of medium source impedance?

In this speaker, any deviations related to Q could easily be corrected using the DSP already present and in use for generating the speaker's polar response.
 
I thought I'd bump this thread with something I came across recently. This speaker came out 2 years ago.
electronics

So are they using current sensing for kind of a MFB or something? I would think that this would effectively yield some sort of medium source impedance?

In this speaker, any deviations related to Q could easily be corrected using the DSP already present and in use for generating the speaker's polar response.

Bose has been using a combination of voltage and load current feedback in their products for some time now.
Analog Devices will soon announce their Class-D current amplifier which together with their SigmaDSP processors (SSM3525 + ADAU1462)would be able to linearize the response of a loudspeaker. I believe they showcased this at CES in Jan 2017.
 
Last edited:
Hi Guys,

I ve been experimenting current drive , after some reading of the Eisa M Book.

some outcomes;
the thd is reduced (around 10db or more) and the typical H3 @ f breakup/3 is dramaticall reduced, meaning you can use the speaker up to say f breakup/2 instead of 3.
Current amp is equivalent to a very high impedance in serie, resulting in a very big Qts (around 10!) and then the speaker (in its closed box ) cannot be used as this generate massive ringing at low frequency.
Then you have to use a LT upfront or an Eq to reduce the Q to say 0.7.
But then you will see that the THD of the speaker below its resonance frequency are worst than in Volatge mode.

On way to solve this is to add in the feedback loop of the current drive amplifier a band pass signal, to "null" the qts (back to 0.7) via inherent feedback.
you can read this famous paper which gives some direction to combine acceleration and velocity within the feedback loop.
Index of /Audio/Papers/Loudspeaker Distortion Reduction
(paper already reffered by forr in the post here : http://www.diyaudio.com/forums/vend...lateral-mosfet-amplifier-241.html#post3636397)


BTW, to get stability on a current drive amplifier you need also a smal 1/1+S correction in the feedback loop, for example 1k//4n7+1k-ground :p

one way to better control all of that is to use a transconductance amplifier, driven upfront by a servo comparing the speed of the coil to the input signal (integrated via a bandpass to represent a speed too).
the tricky part is the velocity measurment, which you can chose from
- acceleration sensor+integ (probably good enough for Sub)
- motional impedance transfer function on the voltage seen at the current sensor resistor (very good approach dependent on the mechanical parameter of the driver and box, with potential shift over time)
- back emf measurement by nulling R-voice coil but highly dependent on the tempco....

good luck! i m working on this for a woofer (35-800hz) and might publish more this year, if Im succeeding.
 
Last edited:
Bruno P has been using current drive mainly for midrange driver and not for DXT tweeter where as I remember correctly it wasn't beneficient at all and tweeter is mainly voltage driven. Unfortunately I don't know how LF transducers are driven. He frequently put emphasis on fact that THD reduction using servos is efficient mostly with current drive.
 
Bruno P has been using current drive mainly for midrange driver and not for DXT tweeter where as I remember correctly it wasn't beneficient at all and tweeter is mainly voltage driven. Unfortunately I don't know how LF transducers are driven. He frequently put emphasis on fact that THD reduction using servos is efficient mostly with current drive.

I am sorry for double posts but I was trying to find some supporting info on this subject:

a great thread which got derailed unfortunately:
Current-Driving of Loudspeakers

and this thread is also interesting:
aktives-hoeren.de • Thema anzeigen - Kii Three - Bruno Putzeys' neue Kreation

and 11 years before now someone very clever threw here a great idea so ahead with thinking that noone catched it up then:
http://www.diyaudio.com/forums/multi-way/78481-mix-current-drive-voltage-drive-lf.html

I myself was trying to get "quasi current drive" tweeter just inserting resistor after amp before DXT tweeter and indeed I was no clue if it was changing anything in sound then SPL level.
 
Last edited:
Has been anybody still working about this specific subject here?

Has someone already wrapped current-control around Hypex amplifier? Bruno P. clearly has challenged you back in 2012. Eight years have been passed and...? There are some DIY-ers who need to build current-drive in order to utilize motion-control on woofer mechanical response and to drop-down nonlinearities in the midrange. Today we are extensively using DSP systems for controlling various things and we are free from passive systems constraints.

Unfortunately my knowledge is poor in this subject and I count on you.


I am talking about specific systems only, where this kind of control is beneficial. Please not weight here whether we need it or not and please focus on subject from practical implementation side of view.


Included whitepapers for starting the discussion:
Audio Design Principles
 
Last edited:
Has been anybody still working about this specific subject here?

Has someone already wrapped current-control around Hypex amplifier? Bruno P. clearly has challenged you back in 2012. Eight years have been passed and...? There are some DIY-ers who need to build current-drive in order to utilize motion-control on woofer mechanical response and to drop-down nonlinearities in the midrange. Today we are extensively using DSP systems for controlling various things and we are free from passive systems constraints.

Unfortunately my knowledge is poor in this subject and I count on you.


I am talking about specific systems only, where this kind of control is beneficial. Please not weight here whether we need it or not and please focus on subject from practical implementation side of view.


Included whitepapers for starting the discussion:
Audio Design Principles

I am at the beginning of a project along these lines. I had planned on posting to this thread once the project is a bit more mature.

Project is a 6 channel amplifier using Hypex modules (most likely the OEM 400UCD). It will drive a tri-amped stereo set of line array speakers. Dipole Bass and Dipole Midrange, ribbon monopole tweeter. The long term plan is to create "switchable" current drive / voltage drive amplifier with the ability to "dial up" different output impedance for each amplifier. The current drive schematic is not decided yet as I am only at the point of ordering the custom chassis and building the 1200 Watt linear power supply.


I hope to have 2 channels running in a month. By then I may have the "current drive" plan decided and can share something.
 
I built a current drive amplifier based on concepts from Esa Meriläinen. I’ve been using it for about a year now matched with Alpair 12pw drivers in a simple closed box and a MiniDSP 2x4HD for response correction. The 12pw seems to be a good fit for current drive especially where the current drive inherently levels its response above 10kHz. I realize sound is subjective and I don’t have “high end” systems to compare to, but I would say as a system this sounds very good. A very natural/realistic sound which is most important to me.

Thanks, Dave

IMG_0331.jpg
 
seems someone else did achieve a good step on this:
Why current-drive? - An active current-drive audio systemAn active current-drive audio system


there are 3 problems to solve with the current drive approach:
- the loudspeaker has to be mounted in a closed box
-you have to damp the high qts either with upfront filter or damping material and choosing the speaker.
-you have to correct the curve in the lows to suit a deeper bass extension than what you ll get with the loudspeaker in the closed box...

I ve made progress on solving the high qts issue by adding in the feedback loop of the amp a bandpass filter. the principles can be wrapped from this paper already mentionned here.

Using the minidsp upfront is a good way to have the LT and to smooth the highs inherently increased by the current drive which cancels the loudspeaker self
 
Last edited:
Bose has been using a combination of voltage and load current feedback in their products for some time now.
Analog Devices will soon announce their Class-D current amplifier which together with their SigmaDSP processors (SSM3525 + ADAU1462)would be able to linearize the response of a loudspeaker. I believe they showcased this at CES in Jan 2017.

seems its finally announced for CES 2019:

"Linearization/Speaker Management with ADI Class D Amplifiers
Discover a full turn-key speaker enhancement solution incorporating a mid-power Class D amplifier with I/V sensing (SSM3525) along with an audio processor to take full advantage of high performance current and voltage monitoring. We’ll demonstrate how a low-performance speaker can perform to the level of a much more costly speaker."
 
Current drive means no electrical damping of unwanted cone movement. You have to do it all mechanically somehow.
No, check out this URL:
Impedance Compensation
If one have frequency independent 8 ohms (or 4 ohms), that means even an equalizing around fo of the bass transducer, what difference remains in this case between voltage drive and current drive ?
What audible differences remain at a listening test between both modes with perfect impedance equalizing of the whole speaker system ?

check out also this URL's:
Applying Impedance Equalization Circuit without knowing voice coil inductance
total impedance presented to amplifier
Speaker Zobel / Impedance Equalization Network Circuit Calculator
https://assets.ctfassets.net/4zjnzn...ments_of_High-quality_Loudspeaker_Systems.pdf
high impedance load
 
Last edited: