Stax Interview and Factory tour 12/12/12 - Page 11

If you're talking about the big Fujiya Autumn Headphone Festival the date is not decided yet but usually late October.

If you're around long enough the week after (early Nov), Arnaud & I usually hit the Tokyo International Audio Show (home speakers event). You'll get the usual suspects for this one too - Avalon, B&W, Sonus Faber, McIntosh, Accuphase, etc.

what do you call low and at what VDS? quite a few of the parts I mentioned were SiC Jfets and bipolars, but I agree its an area with the mosfets certainly that is a big problem. another thing was the fact that a Semisouth Engineer is working to understand what parameters are important to audio use of these power devices and we may well end up with a new power device for audio. Ciss/Coss/Crss are definitely on the radar and Mike Mazzola is putting some serious time into helping us/help him understand their use for audio, by ameliorating some of the more destructive components of operation by setup/operating conditions.

here

there is also some talk of lower power high voltage, low noise jfets, like one would use for a beefy VAS


perhaps Kevin is not counting the RF/Coms devices because they are so obscenely expensive and/or need hardcore PCB layout and tricky compensation for stability at audio frequencies/gains? I would expect high end distortion anywhere near audio frequencies might be a problem for RF amplifiers
Edited by qusp - 2/20/13 at 8:11pm

I think some of these types of devices are now starting to be used in some audio devices. Metrum Acoustics use comms chips in their DAC's http://www.nosminidac.nl/Octave_English.html  They have very good performance, having compared them to similar products costing 5 or 6 times the price. 

 

 

nope, sorry I think you have misunderstood what we are talking about. there is nothing vaguely like a high power RF fet in the Octave, it uses Industrial Coms DACs.

i'm talking about crazy stuff like these LDMOS N-Channel MOS-FET and J-FET normally used for plasma cutters, Broadcast RF and Avionics transponders output Cap of ~100-150pf, silly amounts of power and come in Dual huge Ceramic and gold SMD packages. running $100-1000 each. they are designed for MUCH higher frequency than typically used for audio, but I wonder if they could be used in a Modulated Carrier class D type Electrostatic amp.

this one from Microsemi has ~10pf output capacitance above 200VDS and will handle 500VDS. its a dual common source push-pull N-channel pair. I havent looked much into this area, but it seems to me there might be some DIY Gold there somewhere. watch out for the BeO though

I would think Kevin is already looking at, or has looked at such devices given his work, but has maybe decided against them for cost, difficulty or just deemed them too crazy even for him to use in an audio amp

there are other similar devices from Freescale and STM, as well as probably some more exotic manufacturers. some are very high power but only low voltage, but there are a few 400-1000VDS capable parts and all have reasonably low Coss, some have pretty low Ciss too

freescale was the only one with a decent pic


Edited by qusp - 2/21/13 at 7:27am

The capacitance of Stax drivers is around 110pf so the device capacitance has to be very low indeed.  The C4686 was 2.2pf if I'm not mistaken which pretty much ideal and on par with tubes. 

 

We have looked into some crazy parts but our goal is always to make the design available and as easy to build as is possible.  That means no to the 200$ each parts...   

I think you have misunderstood my comment too.  I know they are not power amplifier chips, but its the first case I know of where industrial comms silicon has been used in a credible audio product.  The other point is the designers background I believe is in industrial electronics.  I think he has had some involvement in electrostatic speakers though.  If we get more open minded designers crossing over from industrial to commercial audio perhaps we might see designs using the types of components you are discussing.  Given that Metrum are starting to build a line of audio components, it may not be long before they develop a power or preamp.

Edited by complin - 2/21/13 at 11:34am

 

Would it need an output filter inductor also?

TBH I cant say I was very impressed with the design for the octave, the digital input and power supply is pretty ordinary IMO. hes hardly the first, comms instrumentation amps, comms logic chips, coms low noise jfets etc have all been used for many years, certainly in DIY. 'made for audio' is not always an endorsement of quality.

I did think it was a bit novel, I just didnt think it was all it was cracked up to be. I guess we'll see what he does with the high end model
Edited by qusp - 2/22/13 at 5:09am

depend how you implement the feedback and modulator. its just a rough idea, not even a back of the hand scribble, but it should work with the right devices. and thats all i'm saying

if for instanvce you look at probably the highest quality class D modules for pro and high power high end these days, fthe NCORE, doesnt have an output inductor and that has led to very nice sounding amps. as with the new high power NCD1200
Edited by qusp - 2/22/13 at 5:13am

I think you have perhaps missed the whole point of the Quad and Octave designs, which is to provide very high quality Musical performance at reasonable cost.  This I believe has been achieved is spades!

The design is deliberately simple, no analogue output stage as such for example, and as I often find in audio less is more.  Yes it does have some compromises as does most audio kit, unless you have very, very deep pockets.  The power supply IMHO is as good as if not better that similar products costing 4 or 5 times it's €800 price.  

I agree the digital input could be improved upon and its input and output facilities are limited.  I have personally found the SPDIF input very cable sensitive and needs to be a certified 75 Ohm cable.  Also it would have been better if it had been fitted with a BNC plug rather than the RCA.

 

DIY using comms based silicone is very different from commercially marketed products.  Often buyers look for well know chips or manufacturers like Woolfson, TI etc 

Edited by complin - 2/22/13 at 8:48am

 

Okay. Albeit dealing with a low voltage IC class D amplifier, this paper describe some consequences of a filter less class D amplifier. The switching square waves can overheat, for instance, tweeters. 

 

I also read that the diaphragm mass add inductance to electrostatic transducers. 

 

I know you are not going to say anything else, but here are my questions anyway: 

 

a) A 2 micron diaphragm mass is already enough to filter this switching content? This may lead to other choices of diaphragm material by Stax. 

 

b) Switching square waves in a filterless class D amplifier can overheat the stators? There will be a concern regarding the stator temperature (if I recall correctly, the SR-009 does already have that center hole for cooling, doesn´t it?).

 

c) How do you deal with EMI when using a filter less class D amplifier? Shielding the headphone cable? I would not be worried about EMI effects in sound quality,but the radiating EMI from the stators and headphone cable interfering with radiofrequency apliances.

Edited by jgazal - 2/22/13 at 5:54pm

I personally have no experience with class D amps - no measurements, etc. Did hear a couple of class D designs ( no idea of the make/brand ) - and they sounded like they still have a LONG way to go till something palatable to audiohille ears avustomed to class A ESL amps driving good electrostatics will eventually emerge. I am FAR from not endorsing progress ( class A is enormously inefficient and wastes electricity, something not acceptable en masse regarding enviroment ) - but the sound of class D I did get to hear was so poor that a mere prospect of attaching that to some good transducer sends chills up my spine. The major problem would be removing switching HF garbage - switching causes enough sonic trouble when used for power supplies to make me switch to linear supply or batteries - NEVER that switching monstrosity for powering my recorders.

 

Until I see an unretouched square wave signal from an independent party coming from a class D amp - not interested. Also not interested in output devices capable only 400 V or so at $ 100/+ each - I am not tube/valve oriented as such, but if there is a place in audio circuitry that is tailor made for tubes/valves, it is the output device for direct driving / without output transformer ESL amp. 

 

Still, all the best in getting this new technologies first to work at all as they should, then bring them to reasonable price level. 

Well here is an informed opinion of class D in audio - 3 or 4 year old presentation on the subject from Martin Colloms HIFICritic

 

 

 

 

Class-D.pdf 127k .pdf file

 

 

 

So, it seems like the 009 was developed and voiced and tested by Stax with the 007t/ii amp. I sure like the way it sounds. I did need to fool around with cables (power and interconnect) and get a warmer-signature power conditioner to

tame some upper mid brightness, but that's the same thing you need to do with revealing speakers, and it's pretty easy to do with a little judicious shopping, and now it is not bright and that was the only sonic issue.

I feel better that Stax seems to confirm my findings, that I'm not just imagining how good it sounds, and I suspected Stax developed the 009 around its own amps (and probably not a T2).

I doubt Stax would spend a lot of money in R&D to develop a pair of expensive and flawed earspeakers, just to make up for the deficiencies of their amplifiers.

Is not that the 009's don't need as much power as the 007Mk1; it doesn't mean they don't sound better with more power.

Even with the SRD7Mk2, the 009's sound wonderful.