Electrostatic Headphone Amplifier Review & Comparison

[bobbooo]

@amirm The input sensitivity for 100V out is 100mV (60dB) on the SRM-006T and SRM-007T and the SRM-313. Your plot for the SRM-006T shows 2V in, 20 times the rated sensitivity. So, if you are feeding 2V in, the volume control on the Stax could not set at maximum (as it is up front of the gain stages) if you only got 80V (32dB gain) out. So the input impedance was high as a result and the performance, both the frequency response, the noise level and distortion will all be affected.

Could you run the Dashboard THD+N/SINAD plot at rated input (100mV- with the vol pot wide open) as below on say the 007t, so we can see how close to rated spec it actually is?

006
View attachment 90436
007
View attachment 90437
313
View attachment 90438

It seems strange Stax would have held onto the old definition of line level even for products made well into the 2000s.

Doesn't this also apply to the Koss E/90X? Maybe to a lesser extent as it was measured with 1V in, 10 times instead of 20 times the rated input (also 100mV due to 60dB sensitivity for 100V output):

Running the Koss at the rated 100mV input as well might show it reaching its 0.001% (-100dB) THD+N spec.

 

[the_brunx]

These measurements seems to correlate to my hearing. I remember when I had a SRM T1, which is similar to the 006t. It had audible distortion at normal listening level, which would gradually increase with loudness,as plotted. I almost couldn't believe how this huge thing could be so disappointing in this area in comparison to smaller STAX amps I had heard before. changing tubes to some new recommended Japanese tubes helped a little. The CCS mod helped a lot. it brought it to a level where this distortion point could be comparable to my SRM323, but not quite. the difference being that the distortion on the T1 was still a little gradual and the 323 would horribly distort instantly after passing a certain level. so the problem with these amps for me was that they would clip when the bass hits when I used it with an EQ which was adjusted for increasing bass. Long story short, I'm now using a DIY Blue Hawaii and all I can say is this thing has a lot of power before clipping. That I know for sure (I think). I have never EQ'd bass so outrageously high on an electrostatic headphone without hearing obvious distortion like I do right now. I'm literally using my Sennheiser HE60 electrostatics as bass-head headphones Just my small input. <3

 

[vkvedam]

Japanese audiophile/HiFi society do seem to prefer distortion. We have seen AVRs from Sony and Marantz reiterating this. Denon seems to be the only decent offering. I am pretty sure if we measure the transducers they will have higher distortion as well.

 

[Francis Vaughan]

I’m going to agree with Kevin about the 100k load on the tube version. It isn’t a reasonable comparison between the 323 and tube based designs. Both are being run way outside of their design parameters. It is just that the 323 is inherently less sensitive to this. But out of design spec is out of design spec.
The 50k ohm anode resistance pretty much tells you that the design will be running way outside of its intended parameters. No if’s no buts, the output stage will be in significant pain and struggling to operate.
The 323 is not an inherently better design because it is less sensitive to load. It is just luck. It was never intended for such low loads, but just ends up coping better.

 

[jrosser]

https://www.ap.com/technical-library/measuring-high-impedance-sources/

 

[kevin gilmore]

i never said the keithley was on the same level as the audio-precision. What it does do is measure past 400vrms with an input impedance that
is electrostatic amplifier friendly. But if you really want to try and get .0000x numbers, then active high voltage fet probes or other similar items
are necessary. If you are adventuresome, then some of the victoreeen high voltage glass vacuum resistors work as input attenuators. I could post
pictures of these. Not sure where they can be purchased new anymore. Anything not vacuum glass enclosed is subject to arc-over and thats a fatal mistake. these things
https://resistorresisti.com/rx-4

i have measured many stax amps before and regardless of age they all measure better than .01% thd at 1khz at 100vrms. All of the stax amps except for the srm-001 and srm600 are fully dc coupled and when you add a low enough output resistance to ground you throw the entire amp out of balance and then all sorts of bad happens.

The koss does indeed measure much better. Reason is the open loop gain approaches 90db and the front end opamp uses 30db of that gain to reduce the distortion to low levels. And also the fact that its push pull. But the power supply kicks out at about 10 watts total power dissipation.
schematics have been posted elsewhere.

the ap technical note is 100% correct, other than its missing all the safety parts to prevent the ap from blowing up. Because at 1kv peak to peak at 20khz its real easy to generate a plasma. and plasmas have the ability to knock you ass across the room with relative ease.

 

[LTig]

https://www.ap.com/technical-library/measuring-high-impedance-sources/

This is exactly what I wanted to recommend.

 

[nj75f]

Funny.
When it comes to distortion/noise these energizers do not measure well, which should translate in only a semi-decent listening performance with the stax electrostatic headphones.
Simply put: The energizers hinder the performance of the headphones.

Still the SR-009/S, SR-007 or the likes ofSR-L700 MK2 are the best performing headphones out there, when it comes to things like detail retrieval, speed (transducer mass, inertia, oscillation) etc.
Would be interesting to know how their new energizers, Stax SRM-007S/T perform.

Johnyang wrote that tube isn't actually that bad when it comes to sinad or overall performance. https://www.audiosciencereview.com/...s-specially-for-headphones.14740/#post-458301
The Stax srm-T8000 and the stax srm-007S are hybrid designs.

It's 4- 6k+ € in europe for the last three mentioned. What are those crazy prices.

 

[amirm]

https://www.ap.com/technical-library/measuring-high-impedance-sources/

As I explained already, the impact would only be on frequency response which is not the "news" here. See the measurement again:

And this is the AP example:

As you see with 47k impedance, AP example shows a 3 dB drop at 20 kHz. I am showing only 0.8 dB which means the output impedance of the amp is much lower, resulting in negligible error if any.

 

[RayDunzl]

On top of that we are talking about distortion measurements from 20+ years ago. What did they use to measure them?

Recently, I've realized that when I say something about "20 years ago" it doesn't seem to have the same meaning as it used to, so I try to say "40 years ago" instead. "20 years ago" is like yesterday, and because almost nothing new has come along.

But then, "40 years ago" has no meaning for the younger folks.

I look back at some recent history, and what was different:

My brief History of Notable Advances Affecting Everyday Life and Ambitions:

1900 -> 1920 - cars, airplanes, radio, world war, atoms, vacuum tube, movies, long distance telephone
1920 -> 1940 - air travel, discovery of "the universe", long distance highways, movies with sound, quantum theory
1940 -> 1960 - more world war, transistors, radar, stereo, FM radio, lasers, TV, computers, rocketry, fission and fusion weapons, nuclear power, jet propulsion, Mutual Assured Destruction, antibiotics, radial tires, plastics
1960 -> 1980 - direct dial phones, answering machines, microprocessors, interplanetary probes, color TV, unleaded gas, mach 2 airliners
1980 -> 2000 - Home computers, Digital sound, mouse and GUI, data networks, genetic manipulation
2000 -> 2020 - Smartphone
2020 -> 2040 - Prediction: nothing new.

I wonder how many "years ago" will be relevant in 20 years...

 

[Vini darko]

Recently, I've realized that when I say something about "20 years ago" it doesn't seem to have the same meaning as it used to, so I try to say "40 years ago" instead. "20 years ago" is like yesterday, and because almost nothing new has come along.

But then, "40 years ago" has no meaning for the younger folks.

I look back at some recent history, and what was different:

My brief History of Notable Advances Affecting Everyday Life and Ambitions:

1900 -> 1920 - cars, airplanes, radio, world war, atoms, vacuum tube, movies, long distance telephone
1920 -> 1940 - air travel, discovery of "the universe", long distance highways, movies with sound, quantum theory
1940 -> 1960 - more world war, transistors, radar, stereo, FM radio, lasers, TV, computers, rocketry, fission and fusion weapons, nuclear power, jet propulsion, Mutual Assured Destruction, antibiotics, radial tires, plastics
1960 -> 1980 - direct dial phones, answering machines, microprocessors, interplanetary probes, color TV, unleaded gas, mach 2 airliners
1980 -> 2000 - Home computers, Digital sound, mouse and GUI, data networks, genetic manipulation
2000 -> 2020 - Smartphone
2020 -> 2040 - Prediction: nothing new.

I wonder how many "years ago" will be relevant in 20 years...

Next 20 years : direct interface , civilan autonomous robots , havesting resorces from space , first humans on mars. Just to name a few things i expect to happen.

 

[Helicopter]

"Clearly" not accurate? Based on what? You have measurements like mine that show different results? Or are you trusting single number spec value from Stax from two decades back with who knows what setup?

And what are the odds of both tube amps measuring as they did?

Here is a random tube amp distortion vs output:

See the sharp rising curve? Now here is the Stax tube amp:

You see the family resemblance?

So no, I don't know that the measurements are faulty and certainly not "clearly." You have independent measurements that show that, I love to see it and will analyze. For now, we have data with reasonable accuracy given what we have to work with.

Next 20 years : direct interface , civilan autonomous robots , havesting resorces from space , first humans on mars. Just to name a few things i expect to happen.

Yeah. Those humans on Mars will be martyrs.

 

[Newman]

@RayDunzl 2020 -> 2040: pandemic acceleration ( and social tear and refabrication), decline of US, driverless transport, VTOL taxis, VR entertainment, ... and I’m probably understating it.

 

[RayDunzl]

direct interface

There's a little bit of that now, in prosthetics, no?

civilan autonomous robots

An extension of microprocessors, data networks, and other existing technologies, but I'll give you a point for it.

havesting resorces from space

An incredibly inefficient activity, I think, given what it takes to maneuver in space, and what we have for propulsion. Needs something "new".

first humans on mars

We did men on the moon, so, that's just an ambitious extrapolation.

Just to name a few things i expect to happen

I just expect more "too many people" and all the problems that come with it.


.

 

[Vini darko]

There's a little bit of that now, in prosthetics, no?



An extension of microprocessors, data networks, and other existing technologies, but I'll give you a point for it.



An incredibly inefficient activity, I think, given what it takes to maneuver in space, and what we have for propulsion. Needs something "new".



We did men on the moon, so, that's just an ambitious extrapolation.



I just expect more "too many people" and all the problems that come with it.


.

Hahahaha yeah people and thier damm breeding.

 

[Helicopter]

i never said the keithley was on the same level as the audio-precision. What it does do is measure past 400vrms with an input impedance that
is electrostatic amplifier friendly. But if you really want to try and get .0000x numbers, then active high voltage fet probes or other similar items
are necessary. If you are adventuresome, then some of the victoreeen high voltage glass vacuum resistors work as input attenuators. I could post
pictures of these. Not sure where they can be purchased new anymore. Anything not vacuum glass enclosed is subject to arc-over and thats a fatal mistake. these things
https://resistorresisti.com/rx-4

i have measured many stax amps before and regardless of age they all measure better than .01% thd at 1khz at 100vrms. All of the stax amps except for the srm-001 and srm600 are fully dc coupled and when you add a low enough output resistance to ground you throw the entire amp out of balance and then all sorts of bad happens.

The koss does indeed measure much better. Reason is the open loop gain approaches 90db and the front end opamp uses 30db of that gain to reduce the distortion to low levels. And also the fact that its push pull. But the power supply kicks out at about 10 watts total power dissipation.
schematics have been posted elsewhere.

the ap technical note is 100% correct, other than its missing all the safety parts to prevent the ap from blowing up. Because at 1kv peak to peak at 20khz its real easy to generate a plasma. and plasmas have the ability to knock you ass across the room with relative ease.

I love tubes, and I see you are highly regarded, but your talk about tubes for science outside radio transmission strikes me as heretical.

I suspect the talk about the 100kohm load is groupthink, but I don't know enough about audio to be absouletely certain.

 

[bobbooo]

I almost bought the Koss ESP/95X at one point when it was on large discount due to closing of Drop's EU warehouse - I'm glad I didn't.

Why? I hope not due to anything you've seen in this review. Because the results show, if accurate and the 100kohm load used here is a good enough approximation of the near-infinite resistance of an actual estat headphone, that the Koss E/90X amp will add at most an inaudible 0.1dB additional distortion to that of the ESP/95X, as combining the ~-80dB distortion of the former with the latter's distortion of -65dB you get a (worst case, in phase) total distortion of -64.9dB. Another amp with lower distortion will not offer an audible benefit. The oft-repeated audiophile 'wisdom' that the E/90X is the limiting factor for the ESP/95X and you need to spend hundreds to thousands of dollars on a Stax amp for the headphones to 'achieve their full potential' is evidently just another audiphool myth.

 

[JohnYang1997]

Easy way out is to measure the headphones with different amps. The total distortion is what matters.

 

[Xulonn]

I just expect more "too many people" and all the problems that come with it.

 

[maverickronin]

100kohm load used here is a good enough approximation of the near-infinite resistance of an actual estat headphone

It isn't at all, but the fact that it's a far harder load than a real headphone makes the amp's performance even more impressive.

FWIW, when I had a long term loan of one I was quite impressed and didn't feel the need to track down the odd parts that would be needed to make an adapter for one of my Stax amps. (And even then people say you need to modify the Stax amp for an extra ~3.3% bias voltage (580 vs 600) for it sound its best...)