[etc6849]

He's paralleling the amp side, and using series on the headphone side. So the turns ratio would be closer to 7.2x4:1 = 28.8:1.

His wiring is very similar to what I drew below. The amp sees 4 coils with 1 turns in parallel. For the headphone side, tracing from L+ to L-, you see you are using 4 middle tapped coils in series with 7.2 turns each (7.2x4 = 28.8), with Vbias-/ground reference being in the middle.

I really would like to see a setup using 4 LL1630's per channel now. No reason this can't be done to raise the point of saturation even more. You can't parallel four on the amp side, but you could have two parallel sets in series to give a DC resistance closer to 6 ohms. The headphone side, I would try having all 4 LL1630's in series, with Vbias/ground reference coming off the middle.

I may just order 8 LL1630's and 2 LL9202AM's to complete my transformer collection.

One thing I didn't gather from your video, in the dual configuration, gain is still 14:1 like in single, right?

 

[gammi]

He's paralleling the amp side, and using series on the headphone side. So the turns ratio would be closer to 7.2x4:1 = 28.8:1.

His wiring is very similar to what I drew below. The amp sees 4 coils with 1 turns in parallel. For the headphone side, tracing from L+ to L-, you see you are using 4 middle tapped coils in series with 7.2 turns each (7.2x4 = 28.8), with Vbias-/ground reference being in the middle.

I really would like to see a setup using 4 LL1630's per channel now. No reason this can't be done to raise the point of saturation even more. I may just order 8 LL1630's and 2 LL9202AM's to complete my transformer collection.

Thanks for the info.

I think the issue with using 4 is you'd halve the input impedance twice and it would be quite a difficult load. Also would be getting to the point where gain is so high that volume adjustment is an issue. 28:1 is fine since the 14:1 of the single transformer always felt a bit low, but 56:1 seems like a bit much. Even at 14:1 I still only need to run my Holo Bliss at like -10dB (I think it maxes at +10dB)

Maybe some combo of parallel and serial could be interesting. Two in parallel, in series with each other or two in series parallel to each other. Like a 2x2 config instead of a 4x config. Only 28:1 gain then. Getting pretty ridiculous at this point though.

 

[icnaiwhimor1975]

Two words: futures contract.

I should have bought BTC when it was at 25..

 

[aokman]

Thanks for the info.

I think the issue with using 4 is you'd halve the input impedance twice and it would be quite a difficult load. Also would be getting to the point where gain is so high that volume adjustment is an issue. 28:1 is fine since the 14:1 of the single transformer always felt a bit low, but 56:1 seems like a bit much. Even at 14:1 I still only need to run my Holo Bliss at like -10dB (I think it maxes at +10dB)

Maybe some combo of parallel and serial could be interesting. Two in parallel, in series with each other or two in series parallel to each other. Like a 2x2 config instead of a 4x config. Only 28:1 gain then. Getting pretty ridiculous at this point though.

It would be interesting to play further in just a test setting. I get the impression that 2 x 1630 is starting to run out of drive current as a 2 probe measurement is starting to pull down the output voltage at higher frequencies, similar to what I would see on the SRM-1 when I tested it.

4 x 1630 seems viable but would prefer it to be a bit higher impedance but the amps didn’t really seem to care either. If splitting the cores creates other issues like distortion and phase shift, im not sure yet.

I would like to try the LL9202AM but we don’t seem to have many options out there that play ball without creating resistor networks. I do wonder what would happen if it were possible to do the winding config of a 1630 but in a larger core size or if everything is a balancing act.

The Edcors still aren’t out yet either, i need to drive them harder and see what happens at higher voltages then I can push into higher frequencies and check the drop off as it scales up.

 

[etc6849]

Yep, a parallel and series combo is what I'm talking about trying if I buy 8 LL1630's (4 per channel). A 57.6:1 turns ratio would give great headroom, so I'd try that as well in my testing, and of course a 28.8:1 config as @gammi mentions.

I can clip my Benchmark AHB2's when driving my X9000's if I play the most dynamic tracks I have, and listen to them at loud levels. You can't hear the clipping, but you can see the AHB2 momentarily lights up indicating clipping has occurred.

What's most interesting is the 1.97kV peak measured wrt Vbias-/ground reference... so for a fraction of a second the X9000's see nearly 4kV at a loud but listenable level!?! The peak can be seen ~23 seconds in at the video link below. Thus, I wouldn't be against trying a higher turns ratio.

Maybe some combo of parallel and serial could be interesting.

I really would like to see a setup using 4 LL1630's per channel now. No reason this can't be done to raise the point of saturation even more. You can't parallel four on the amp side, but you could have two parallel sets in series to give a DC resistance closer to 6 ohms. The headphone side, I would try having all 4 LL1630's in series, with Vbias/ground reference coming off the middle.

 

[Svperstar]

Does anyone have an updated post showing the voltage/current of various estat amps? This post on headcase is from 2012:

https://www.head-case.org/forums/topic/10215-stax-and-other-electrostatic-amp-information/

 

[jody2k]

Doing a bit of research I understood Stax tried to go "big as possible soundstage" with the first Stax Omega.
I read that there was some kind of a cost, the big soundstage came with bit of loss of instrument focus and seperation/layering.
Unsure that is true as I never heard the Original Omega.

Response to this they released the Stax omega II (007 MKII) that has a more focused smaller stage with better instrument seperation and layering.

Now with the X9000 it seems they changed direction again (going back to a big stage)
I can't keep wondering did it improve over the Omega original?
Has anyone here happen to compare the X9000 with the original Stax Omega?

 

[aokman]

Does anyone have an updated post showing the voltage/current of various estat amps? This post on headcase is from 2012:

https://www.head-case.org/forums/topic/10215-stax-and-other-electrostatic-amp-information/

So heres the issue…

Most manufacturers “rate” their voltage swing at 1Khz… but that is the easy part. What should be happening is showing maximum voltage swing into standardised load from 20Hz to 20Khz on a graph…

If you refer to my last graph, my STAX SRM-1 meets or exceeds its Vrms rating at 1Khz but quickly falls off past that frequency. I believe etc6849 had similar results with his BHSE… Also conflating this is that the aftermarket use Vpp instead of Vrms and still at 1Khz.

I would love nothing better than to put all the amps side by side in the same test environment and run them all back to back through the whole range.

Even the small LL1630 can drive high frequencies with ease…

 

[etc6849]

As far as I know, the highest any of the best estat amps can go is 2kV peak to peak stator to stator. Most will say you never need to get higher, but just normal playback of a dynamic track and loud volumes shows transients exceeding 1.9kV peak to peak, 3.8kV stator to stator.

It seems no estat amp can do what two Lundahl LL9202's driven by the Benchmark AHB2's are doing; and yes, it sounds fantastic.

Does anyone have an updated post showing the voltage/current of various estat amps? This post on headcase is from 2012:

 

[Chefguru]

As far as I know, the highest any of the best estat amps can go is 2kV peak to peak stator to stator. Most will say you never need to get higher, but just normal playback of a dynamic track and loud volumes shows transients exceeding 1.9kV peak to peak, 3.8kV stator to stator.

It seems no estat amp can do what two Lundahl LL9202's driven by the Benchmark AHB2's are doing; and yes, it sounds fantastic.

Technically anything over 1500vpps is pretty dangerous in a humid environment, which is at least some of the reasoning for many amps being at or below this (for example the new GC is not a powerhouse). T2 is around 1900vpps. The new KG amp will do above 2k.

The topology, and “things in the way”, may influence sound more than power at the 1500vpps level anyway.

 

[aokman]

This is something I have never seen a straight answer on what is the max safe voltage before the diaphragm hits the stators. I would have thought it was related to the bias voltage but most direct drive seem to be 1600Vpp max?

Honestly I rarely saw 300Vrms on my meter with music just watching and that was dam loud. Obviously transients are a different story though…

I asked Dan Clarke a while ago if he recommended pro bias to only or if his estats can take more and improve dynamic range… never got an answer

 

[gammi]

This is something I have never seen a straight answer on what is the max safe voltage before the diaphragm hits the stators. I would have thought it was related to the bias voltage but most direct drive seem to be 1600Vpp max?

Honestly I rarely saw 300Vrms on my meter with music just watching and that was dam loud. Obviously transients are a different story though…

I asked Dan Clarke a while ago if he recommended pro bias to only or if his estats can take more and improve dynamic range… never got an answer

I'm just reasoning here with some basic logic but I would assume it's not a static limit. I expect the limit is higher at higher frequencies and lower at lower frequencies due to it spending less time with the voltage applied at higher frequencies. If you apply 1600V for 1/100th of a second, the diaphragm is going to travel farther than if you apply 1600V for 1/1000th of a second.

So rather than defining the limit as a static Vpp it's probably a rate of Vpp/s.
This could explain how @etc6849 is seeing very high instantaneous Vpp and not blowing up his STAX.

 

[aokman]

I'm just reasoning here with some basic logic but I would assume it's not a static limit. I expect the limit is higher at higher frequencies and lower at lower frequencies due to it spending less time with the voltage applied at higher frequencies. If you apply 1600V for 1/100th of a second, the diaphragm is going to travel farther than if you apply 1600V for 1/1000th of a second.

So rather than defining the limit as a static Vpp it's probably a rate of Vpp/s.
This could explain how @etc6849 is seeing very high instantaneous Vpp and not blowing up his STAX.

Interesting, thanks for sharing that as I had not considered it. Makes me ponder if STAX factor this into their amps at all or its still the rule of linear is linear and voltage / amplitude as flat as possible across the audio spectrum…

 

[padam]

Doing a bit of research I understood Stax tried to go "big as possible soundstage" with the first Stax Omega.
I read that there was some kind of a cost, the big soundstage came with bit of loss of instrument focus and separation/layering.
Unsure that is true as I never heard the Original Omega.

Response to this they released the Stax omega II (007 MKII) that has a more focused smaller stage with better instrument separation and layering.

Now with the X9000 it seems they changed direction again (going back to a big stage)
I can't keep wondering did it improve over the Omega original?
Has anyone here happen to compare the X9000 with the original Stax Omega?

https://www.head-fi.org/threads/stax-sr-x9000.959852/page-89#post-16992372

From memory, an X9000 can project an even "bigger sound" with the right recording with many elements, like movie soundtracks.
When doing a deep-focused listening (meaning, I kill off all lights and unnecessary electronics in a quiet environment, closing the eyes is insufficient) I can pinpoint more 'complex directions' in the stage, contributing to a more 'detailed' feeling of spatiality.
However, as far as improving over the Omega, not sure it has achieved it with "more normal" recordings. Does not have a uniform feel, separation is over-exaggerated, as if there were gaps in the soundscape. This was partially caused by tilted outside grilles, eliminating some reflections.
A bit "technically over-engineered", less tuned by ears (imho). So, for me, the direction pretty much set in what they need to do in the next version.

The 007 has a fairly uniform feeling of air and depth, but also laser separation, so it does have a more unique presentation than the X9000, which the very first review on Youtube (now moved to the Long Tech channel) was smartly pointing out.

In quite a few recordings, I actually like the "simplified" NB Lambda stage, also don't mind the SR-X stage either, which some people may feel as small and tight after getting used to an "earspeaker-like" presentation instead of "headphone-like". Freeing the music of some of these 'spatial gimmicks', bringing it closer to the listener can lead to a more pure, tactile and uplifting listening experience imho. NB Lambda still has "classic Stax air rendering", just a less exaggerated version. But, for instance, lower treble can be quite prominent compared to other models.

So, personally, I just wouldn't be overly obsessed with stage beyond a certain level, which was one of the main points of that review you quoted.
I just find that there are more interesting aspects.

But if you prefer to do more research on that, there is an updated review here mainly focusing on the X9000, but he also still owns the Omega as well:
https://www.head-fi.org/threads/stax-omega4-compared-to-omega1-omega2-omega3.971637/


Stax wasn't specifically aiming for the stage. Main point of the Omega was to substantially evolve and upsize the circular mesh driver (SR-X Mk1). Big driver = big stage.

With the SR-007 Omega II following after the bankruptcy (don't confuse SR-007 Omega II Mk1 and SR-007 Omega II Mk2, there are different generations not including various versions amongst them, different Omega versions with slightly different staging as well, let's not got into that either...), they needed to make something more financially viable and reliable (meaning: SR-Omega and SR-007 had exactly the same retail price, but driver technology in the latter was significantly cheaper and easier to make) which was the main driving force behind it. And the byproduct of using a smaller transducer and designing the rest around were the actual consequences, and not specifically focusing on 'improving' those things.

 

[ufospls2]

Curious to see how this turns out