The Stax Thread III

[kevin gilmore]

BIAS supply questions.

Again, if my design question belongs in another thread, please alert me and I'll repost. Thanks.

The questions are:

1) Regulated bias supply or unregulated?
• Advantage Regulated: The UHV-power regulated supply module, model USA24600S-4, available from lcsc.com, is quite cheap from China, $33 each.
• Disadvantage Regulated: The UHV is is not adjustable and is available with a fixed 600 V output (580 V is not an option to my knowledge)
• Advantage Regulated: The Bertan PMT model is very clean, has very low ripple and is adjustable.
• Disadvantage Regulated: The Bertan PMT is very expensive, well over $100 from Ebay and potentially over $300 new
• Advantage Unregulated: The unregulated DC-DC converter, Q 06-5 from Emco (XP Power) is quite expensive, well over $100 each retail. But it can be adjusted with an adjustable low voltage supply input. The LT 3081C regulator can be easily adjusted to circa 5V, so producing a predictable output voltage from the Emco into a known fixed load.

2) Is 600 VDC bias dangerous to the phones?
* Dr. C. of Audeze tells me that the CRBN can take a much higher bias voltage without any damage, so 600 V is not a problem with the Audeze. The question is this dangerous for Stax phones?

3) If 600 VDC is dangerous for the Stax models, how can we effectively attenuate a 600 V regulated supply?
• The higher the resistive values in the attenuator, the higher the impedance feeding the 4M7 bias resistor and the more unpredictable and harder to measure is the output voltage. It pays to start with a low impedance source to accurately measure the voltage, even with a 10X probe. The lower the resistive values in the attenuator, the bigger and more cumbersome wattage resistors have to be employed. This is a dilemma, I would like suggestions as I'm attracted by the price of the UHV module.

1)
the bias supply on all my desktop headphone amplifiers is a voltage doubler from the + regulated supply which makes 700v-800v. don't care what the voltage is because it is followed by a 10m90s and a 580v string of 1% zener diodes. then 2 stage filter. this produces a regulated 580v +/- 5v that is not adjustable. over time and temperature and long term the measurements drift less than 1%

the 600v of the uhv power is too low to do the 10m90s, but 2 resistors bring it down to 580v easily and sure looks to be extremely stable, less than 1%
(have it on a long term measuring rig right now)

you could use the emco module and set it for about 700v and then do the same thing with the ixys and zener diodes.

the one problem i know of for sure is that as the bias current changes mainly due to humidity, the emco module output voltage will drift even with a fixed input voltage.

alternatively you can wrap the emco module in a fully regulated differential discrete amplifier similar to the grhv.

2) personal opinion is that the stax headphones will not be damaged with 600v as long as you use the 5Meg resistor

3) correctly measuring the bias voltage is easy with any standard 10M input resistance dvm. just have to do it before the 5Meg resistor.
(or get one of the 1Gohm input impedance voltmeters, which is much more expensive)
(new versions of these are hard to find, have to go vintage vtvm)

If the headphones are new, the inital charging current is 50 micro amps which drops very quickly to under 10 microamps (5 minutes)
using 1gohm input impedance voltmeter measuring voltage accross the 5M resistor. i have seen really trashed headphones continuously draw over 100 microamps which does not decrease with time.

doing a fully regulated everything with feedback directly after the 5M resistor is probably a bad idea because it will try to force the headphones to charge faster.

doing a quad esl63 charging circuit which includes a neon light in series with a 47k resistor across the 5M might be a way to charge headphones safely and much quicker.
since it works reliably for 20+ years on my quads, it must work.

 

[bobkatz]

I was just worried that for a low value resistor divider, the wattage resistors required would be very high, and for a high value resistor divider, then you run into precisely measuring the actual voltage even with a high impedance VM or a 10X probe. The higher the value of resistors in the divider, the more that the 4M7 loads down the attenuator as well, though that's not the real question, the real question is what is the voltage if the source resistance of the divider is rather high. What values did you end up with? I certainly know that I cannot reliably measure the Bias voltage at the headphone end of the 4M7, but I know, that's an extremely high resistor

 

[gammi]

I was just worried that for a low value resistor divider, the wattage resistors required would be very high, and for a high value resistor divider, then you run into precisely measuring the actual voltage even with a high impedance VM or a 10X probe.

I was able to set mine fine with a $20 multimeter from Amazon. Just measure before the 4M7.
I used a 604K resistor and 25K trimmer.

 

[kevin gilmore]

the posted original values were way wrong.
half watt resistors fine with 604k and 25k pot

the super high impedance meter is only used to measure the charging current by measuring the voltage across the 5meg resistor.
otherwise unnecessary.

inline microamp meter is another solution for measuring the current.

 

[bobkatz]

I was able to set mine fine with a $20 multimeter from Amazon. Just measure before the 4M7.
I used a 604K resistor and 25K trimmer.

OK, thanks. So the source impedance is going to be hopefully less than 25k so a 10 Meg TVM will not meaningfully load down the measurement. Well done! I will steal those values from you!

 

[TheAbyss2022]

The CRBN has arrived and it's going to be a dogfight between all three.

Impressions hopefully in a couple days once the CRBN has had some burn in.

 

[mulveling]

The CRBN has arrived and it's going to be a dogfight between all three.

Impressions hopefully in a couple days once the CRBN has had some burn in.

Awesome - but just go ahead and listen, man! Can't say I've ever noticed burn-in effects on an electrostatic headphone, yet. And really in general, there was just one weird L3000 I had that seemed to show significant improvement from new (and yes, I had an older L3000 to compare).

I sort of missed the CRBN release, and now that I'm playing with my T2 again I've realized I may be looking for something different than 009 - finally, after 10 years. The Lambda 404LE is hitting the spot for now, but I'd like to try something else eventually. X9000 may not be the right fit given what I've read about them. I would be looking more for fun / interesting / engaging sonic perspectives rather than pursuing the ultimate in resolution / technical acuity.

Going up against my speaker rigs is the challenge because they excel in "engagement" factor, and that is hard for headphones to approach - especially while keeping listening fatigue to a reasonable level. I've found some dynamics can be quite engaging, but lord they murder ears at higher SPLs - not good for longer term listening.

 

[gammi]

the posted original values were way wrong.
half watt resistors fine with 604k and 25k pot

Yes, that's why I corrected them to 604k and 25k in the original post.

 

[TheAbyss2022]

Awesome - but just go ahead and listen, man! Can't say I've ever noticed burn-in effects on an electrostatic headphone, yet. And really in general, there was just one weird L3000 I had that seemed to show significant improvement from new.

I sort of missed the CRBN release, and now that I'm paying with my T2 again I've realized I may be looking for something different than 009 - finally, after 10 years. The Lambda 404LE is hitting the spot for now, but I'd like to try something else eventually. X9000 may not be the right fit given what I've read about them. I would be looking more for fun / interesting / engaging than ultimate resolution.

I'm listening to the CRBN right now but I won't do any comparisons with the Stax until I have put a few more hours on them.

009 is a very special headphone and it still is top tier imo.

 

[mulveling]

009 is a very special headphone and it still is a top tier imo.

Absolutely agree - and to your findings on 009 vs. 009S, I do listen to a good bit of hard rock & metal. BUT after 10 years, I've moved on everywhere else in audio (sometimes several times over), and am simply itching for a different transducer. My role for headphones has changed too; it's now a "side piece", no longer an equal partner to speakers. So just plain technical acuity alone isn't gonna do it for me. I'm not looking to marry

 

[TheAbyss2022]

Absolutely agree - and to your findings on 009 vs. 009S, I do listen to a good bit of hard rock & metal. BUT after 10 years, I've moved on everywhere else in audio (sometimes several times over), and am simply itching for a different transducer. My role for headphones has changed too; it's now a "side piece", no longer an equal partner to speakers.

I haven't gotten into speakers due to space, noise, room treatment etc. too much hassle plus it's a different ball game in pricing

 

[mulveling]

I haven't gotten into speakers due to space, noise, room treatment etc. too much hassle plus it's a different ball game in pricing

Yes, that was my reasoning too, for the longest time. Also when I started looking at speakers circa 2000s, the big multi-driver units just sounded awful - completely lacking in "coherence" compared to headphones. Complex music (like metal) falls apart. Of course Diana Krall demos sounded great, but who cares! When I found Tannoys, they sounded fully coherent (dual-concentric drivers), and my journey began. The good news is I found even small-ish rooms can work (Tannoys are room friendly), and you usually don't need a ton of acoustic treatments (less is more). The bad news is you really can't enjoy speakers if you share a wall / floor with neighbors, and everything new & nice these days costs a fortune. Some good deals to be had used, but you have to already know what's good - even Tannoys have really bad models - the Definition DC10A always ran me out of the room. I've now heard other good speaker brands I like too, but their asking prices are very dear. Got a life supply of Tannoys, and happy to stick with that - never get tired of 'em

 

[TheAbyss2022]

Yes, that was my reasoning too, for the longest time. Also when I started looking at speakers circa 2000s, the big multi-driver units just sounded awful - completely lacking in "coherence" compared to headphones. Complex music (like metal) falls apart. Of course Diana Krall demos sounded great, but who cares! When I found Tannoys, they sounded fully coherent (dual-concentric drivers), and my journey began. The good news is I found even small-ish rooms can work (Tannoys are room friendly), and you usually don't need a ton of acoustic treatments (less is more). The bad news is you really can't enjoy speakers if you share a wall / floor with neighbors, and everything new & nice these days costs a fortune. Some good deals to be had used, but you have to already know what's good - even Tannoys have really bad models - the Definition DC10A always ran me out of the room. I've now heard other good speaker brands I like too, but their asking prices are very dear. Got a life supply of Tannoys, and happy to stick with that - never get tired of 'em

That's the issue as I share a wall and he bought a sub woofer and out of nowhere started to play movies through it and I had enough and told him to sell it!

 

[mulveling]

That's the issue as I share a wall and he bought a sub woofer and out of nowhere started to play movies through it and I had enough and told him to sell it!

There is definitely no way I can share a wall anymore without going crazy. Ironically, my old college dorm (on-campus apartment) at GA Tech had the best sound insulation of any structure I've ever lived in. Everything since then has been flim-flam construction.

 

[bobkatz]

More discoveries: Linearity of frequency response of my transformer-based amp, which uses a 15 watt toroidal transformer. I have reported in this thread that there is a rising HF response at the secondary. The response is -0.5 dB at 20 Hz, flat by 50 Hz or so, flat at 1k until it's +0.25 at 10k, +0.75 at 20k and +2 by 40 k. The 1k square wave has a corresponding overshoot. So, a good friend advised me to try terminating the secondary of the transformer, which is essentially working into an open circuit. Some of it could also be interaction of the transformer with the headphone capacitance. My measurement rig has a 100 pF capacitor simulating the headphone load.

So I tried various resistors across the high impedance secondary. They absolutely help the overshoot and the flatness. 100k was ineffective. A 10k resistor was somewhat effective and dropped the level at 1 kHz by 1.6 dB, which I don't mind. A 1k resistor damped the frequency response and the level too much although it did a very nice job on the square wave. I settled on 5k, with a measured response +0.1 dB at 10k, +0.2 dB at 20k, just fine in my book, and a meaningful improvement on the square wave overshoot. I forget what the voltage loss was, but I thought it was liveable when I measured. Will measure again after the resistor is in permanently.

Since I have amplifier and transformer headroom to a bit above 1000 v RMS, what about the termination resistor wattage? I settled on a nice small 100 watt 5k 2% resistor. Ideally it should be 200 watts, but not only is 1000 volts very loud, it could only happen on a very brief transient peak, not enough to start heating the resistor. When I get the resistors from Digikey I'll leave the top open and watch the temperature on the resistor and see how it's "cooking". I don't think 100 watts will be a problem.