Experiments with the current drive

[tktran303]

Do you have access to COMSOL?

 

[Kravchenko_Audio]

Do you have access to COMSOL?

I use custom software. We have compared it regularly against COMSOL. What is custom about our software is the wrapper. We use FEMM4.2 for the magnetics, and Calculix and Z88 for different parts of simulating a cone or the spider.

A legitimate COMSOL license is prohibitive to pay for as a small company. We've been running our software since 2018 and designed nearly a hundred drivers since then. Our simulations come out very close to the real article. And we have save the butt of a few errant COMSOL designed drivers as well.

With a bunch of research to create our software we found some interesting information regarding simulations on the Opensource software versus COMSOL. With attention to the input details the answers are identical. The math behind the scenes to create the simulation is identical. The only difference is that FEMM 4.2 simulates in 3D but displays in 2D.

 

[tktran303]

Well, if anyone else has already done a deep dive on the various ways of inductance demodulation eg. copper cap on the pole piece to make the voice coil inductance low and symmetric eg. original Philips drivers in the market by 1973, Ragnar Lian SD patent 1973-1993, vs 3 copper rings (Ragnar Lian’s patent for “SD-1” in 1994-2014), 2 copper rings eg. SEAS’ old Excel, vs 1-3 alu rings or 1-3 rings of variable resistivity eg 2 alu, 1 copper eg. KEF

And/or effects of current drive on them, I’d like to hear about…

 

[capslock]

@TNT

My hypothesis goes something like this-

A bare driver like an alu cone with a full copper sleeve eg. SB15/17 N(B)AC may not minimal to no real improvement.

I have the SB17NAC-04 here so one fine day...

b) A driver with one (or two) air cored inductor in series will show some improvement. The level of improvement may depend on the total series resistance and how much it linearizes the impedance.

I have the Scan-Speak 18W/8535 with the SD-1 (3 copper rings) as part of the ProAc Response 2.5 clone, with 3rd order low pass filters using 2 air core series inductors, so one fine day...

c) To linearize the response back to original response without the inductor. Why would one want to do that?


2.


@Kravchenko_Audio

Dear Mark,
I agree.
In theory, there is no difference between theory and practice. But in practice, there is.
( Benjamin Brewster)

Unfortunately I will not be able to do that experiment. That picture is an old one, after I had to repair a damaged driver. I've since reassembled it, and it's in a completed speaker now. I won't be disassembling it again.

Re: simulation vs observation-
That reminds of the time I wrote to @jackocleebrown at ASR
https://www.audiosciencereview.com/.../kef-r-meta-series-release.41420/post-1487851
asking why he didn't have a fully copper sleeve in the TOTL Reference/Blade META drivers ie. coax and woofers.
He was curious and wrote to me directly. We compared notes. With thanks @5th element

In the end we did not reach a consensus. But we stand by our research and I shared with him our findings*.

Hopefully we can look forward to the next generation KEF Reference/META UniQ transducers to have even lower distortion from the MF unit. (IMHO the HF is already top of the class. He did complete a PhD on Phase Plug Design for compression drivers and applied that knowledge when he went to work for Celestion, and KEF, both subsidiaries of (OEM battery giant) GP (Gold Peak))

Full disclosure: "*In the interest of transparency, we confirm that no financial compensation, funding, or material support was exchanged in relation to the research findings presented. All conclusions and interpretations are based solely on objective analysis, independent data assessment, and professional integrity. This disclosure underscores our commitment to unbiased inquiry and the ethical dissemination of information."

Thank you for those measurements. I makes sense that if the motor is already really well demodulated, you are at the mercy of the nonlinearity of mechanical compliance at or below resonance, so current drive may make things worse. The is the infamous low damping factor for better control thing. With current drive there is no electrical damping at all and close to no motor force needed to get the driver to oscillate. The only thing that can improve it over the voltage driven scenario is MFB.

I am surprised that there is no benefit of current drive higher up in frequency. The Purifi folks seem to have done a really thorough job. There is a region between 100 and 200 Hz in what I suspect is the plot of 3rd harmonic (it is also labelled 2nd like the preceeding graph) where current drive seems better, but it is not systemetic, i.e. the same region in HD5 is worse, and in the midrange for both uneven HDs, current drive is identical or worse.

I would have expected uneven midrange to benefit from current drive even with the sophisticated motor. Now 20 R series resistance is only an approximation of current drive. Is there a chance you can use a larger resistance or acutally modify an amplifier for real current drive?

As for the SB15/17N..AC, they have a pretty good motor but their sleeve is not thick enough to be effective at LF, and there is no Neo magnet on the pole piece to saturate it. I bet they'll benefit from current drive.

I'm not sure I get what you are trying to tell us here:
"b) A driver with one (or two) air cored inductor in series will show some improvement. The level of improvement may depend on the total series resistance and how much it linearizes the impedance."

Do you mean that the series resistance of the inductors is already similar to current drive so there is little to be gained from full current drive?

 

[tktran303]

@capslock
Good catch~ Your are correct in noticing the typo error in the 2nd graph again title "...2nd HARMONIC" is incorrectly labelled (the titles are manually entered by me), and that the legend in the bottom left, and the table in the top right corner stating 3rd harmonic is correct. Apologies.

Is there a chance you can use a larger resistance or actually modify an amplifier for real current drive?

Currently I am using a 20 ohm cement wire wound resistor rated at 10Watts...
There's little chance of me using a 40 ohm resistor because I would need to acquire some high power resistors. I could easily acquire 2x 80 ohm 10W resistors to run in parallel, but the power handling would unlikely to be sufficient.

No chance of the latter- at this stage my interest lies mainly in seeing what would happen if future all-in-one integrated manufacturers implement partial current drive in their products, or what happens with passive crossover networks...

 

[TNT]

all-in-one integrated manufacturers implement partial current drive in their products

Do we know of such examples...? Kii?

//

 

[tktran303]

Australian manufacturer SGR Audio:

https://www.sgraudio.com/collection/convex-series

Recently French manufacturer Focal:

https://www.focal.com/catalog/pro-audio/monitoring-speakers/utopia-main

there may others that am unaware…

 

[Kravchenko_Audio]

Currently I am using a 20 ohm cement wire wound resistor rated at 10Watts...

So this resister is inductive already. Have you measured it's inductance?

 

[Boden]

@mark: Take your time and then show us the Borresen too.

BTW: Is Steve Mowry no longer doing his own sims? I was under the impression the Steallus model range was his own work..