• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Electrostatic Headphone Amplifier Review & Comparison

OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,384
Location
Seattle Area
Hey Amir.... how about a real level matched blind ABX test between the tube and SS Stax amps.... do you think you could tell the difference with music? It would be an interesting test I think.
I know at high volumes it is obvious due to distortion. As to performing the AB test, would need to build a high-voltage capable AB switch to switch over 5 conductors. I worry about the transients doing harm to the amp also. In other words, not for me to test. :)
 

pavuol

Major Contributor
Forum Donor
Joined
Oct 2, 2019
Messages
1,562
Likes
3,950
Location
EU next to warzone :.(
Beware the bezoars!

sidenote - I didn't know Stax offers inears and portable battery amp as well. Apparently they didn't want to give up the "audiophile on-the-go" market segment :).
 
  • Like
Reactions: Tim

Veri

Master Contributor
Joined
Feb 6, 2018
Messages
9,596
Likes
12,036
This is a review and detailed measurements of a number of amplifiers/energizers/drivers for electrostatic headphones. I own three of them which are rather vintage Stax units: the SRM-311, SRM-006t and SRM-007t. The first one is solid state and the other tube. The last unit for comparison is the Koss E/90X energizer that is on kind loan from a member.

Here is the gang:

View attachment 90410

The Stax units harken back to the great days of Japanese hi-fi with nice enclosures, switches and volume pots. The Koss in sharp contrast could be mistaken for a give-away kids toy in McDonald's happy meal.

I did not have the plug for any of these so made a twisted wire probe that I simply stuck in the Left+/- connection. For this reason, the noise level may be higher than it would be with a more proper connector.

Stax Electrostatic Headphone Amp Measurements
I had to modify my headphone amp dashboard as it uses a 600 ohm load which doesn't work for electrostatic headphones which essentially have infinite resistance. The highest my Audio Precision goes to is 100,000 ohm so that is what I used. The other major difference is that electrostatic headphones require far higher voltages to produce sound. So instead of my usual 2 or 4 volt output, I used a whopping 80 volts! I thought this would be within the comfort range of all the headphone amps as it is in the middle setting of the Stax headphone amps. Alas, this did not turn out to be the case but we are getting ahead of ourselves. Let's start with the Stax SRM-311 where I developed the tests:

View attachment 90411

Second harmonic is at -88 dB which more or less sets our SINAD since distortion is more dominant than noise (SINAD is a relative sum of noise and distortion).

Frequency response had a slight low frequency boost:

View attachment 90412

And droop at 20 kHz. The variation probably gets lose in the frequency response of the headphone itself.

Last test that I prepped was the power test but now into 100K ohm:

View attachment 90413

Let's jump to the other extreme and test the SRM-007t which is a differential tube amplifier (balanced in, balanced out):
View attachment 90414

Now this is a bummer. Distortion at around 80 volts was huge as you see above. I thought it might be clipping or something but that is not the case:

View attachment 90415

No wonder I always heard this amp distorting like heck at higher volumes.

Its frequency response is more flat though:
View attachment 90416

Here is the SMR-006t which is also tube based but its output is not differential:

View attachment 90419

Frequency response is the same as SRM-007t:

View attachment 90420

It has the same distortion as well although noise floor is higher yet:

View attachment 90421

So this rules out anything being broken in the tube energizers other maybe higher noise of the SRM-006t.

KOSS E/90X Energizer Measurements
The E/90X lit its LED in red in protest well before 80 volt out. So I decided to take it easy on it and turn up the level to just before LED turned red:

View attachment 90423

Running our sweep we see that the KOSS is much more well behaved albeit, with much lower output voltage:

View attachment 90424

Both distortion and noise are low but then we run out juice at nearly 10% of what the Stax amps can produce. I noted that the output level was not that high in my review of the KOSS ESP 95X headphone about this issue.

Frequency response test was interesting for two reasons:

View attachment 90425

Notice that the low frequency response changes with output level! Is this some kind of loudness compression? If so, the KOSS headphone measurements will be level sensitive.

The second issue was instability at frequencies below 20 Hz. I set the frequency to 16 Hz and as I turned up the volume gradually the output suddenly jumped from something like 10 volts to 160 volts! The LED would then flicker between red and green. Suggest using a high pass filter with this headphone/amp to make sure it doesn't see anything below 20 Hz.

Conclusions
Anyone want to buy a pair of hardly used Stax tube headphone amps? 'Cause after seeing how poorly they perform, I don't think I want to be seen in the same house with them! The heck were they thinking? A tube amp mates well to electrostat headphones because it naturally have a high voltage output. Why screw that up with tons and tons of distortion? I guess they decided to use no feedback.

I was surprised that the KOSS E/90X had very low distortion. Alas, it must use much lower voltage rails/output transistors so not able to generate nearly as high a voltage as the Stax units can. Alas, if it is providing adaptive bass boost, using a different amp without that feature (the norm) would cause different audible results.

Will be interesting to test other after-market amps to see how they compare to these.

-----------
As always, questions, comments, recommendations, etc. are welcome.

Dropped by the garden (what's left of it) to see if the persimmons are ripe enough to pick and my heart fell to the ground. The one fruit I was eyeballing everyday was gone! Fearing the worst, I searched for the rest and found a few remaining ones:

View attachment 90430

We waited nearly 10 years for this little tree to bear fruit and it finally bore fruit a couple of years ago. Alas, last year some varmint at them all. :( So despite losing some, I am happy to have rescued the above. They last a long time since they are the Japanese variety.

Appreciate any kind donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
Been waiting for this kind of information for a long time, and we get four measurements in one!! Many thanks Amir :)
I sure wonder about the entry-level Stax SRM-252S and "mid" range SRM-353X now! Hope you get hands on them some day.
Oh and since you measured the iFi iDSD already, perhaps the iFi iESL some day :D
 

brimble

Active Member
Forum Donor
Joined
Oct 18, 2020
Messages
159
Likes
206
Location
Australia
Thank you so much for this review!

I spend about 27 hours a day with Stax headphones on my head, so it's a very important subect for me.

I've listened to many Stax amps in stores, although I haven't always taken complete notes, and I own two myself (both solid state). My observations have always been that the solid state ones sound identical (although admittedly this is from listening, not only without measurements but also sometimes in less-than-ideal listening conditions) but that the tube amps sound completely different.

Good to have my impressions confirmed by the measurements.

As to which way round the difference is, what you found is what I expected. It's very good to have that confirmed, and it may have saved me hundreds of dollars, as now I won't be even tempted to buy a tube Stax amp to check for myself.

One grateful reader.
 
Last edited:

brimble

Active Member
Forum Donor
Joined
Oct 18, 2020
Messages
159
Likes
206
Location
Australia
Hey Amir.... how about a real level matched blind ABX test between the tube and SS Stax amps.... do you think you could tell the difference with music? It would be an interesting test I think.

Admittedly I haven't done this with proper level matching, but I have done it with my best effort at level matching by hand, FWIW. The difference was very audible, and I much preferred the sound of the solid state amps.
 

test1223

Addicted to Fun and Learning
Joined
Jan 10, 2020
Messages
503
Likes
511
Very nice test.
Are there better amps which are not to expensive for the stax headphones?
@amirm the level dependent channel balance should also be an issue with the old porti design. Have you measured it?
 

SpaceMonkey

Active Member
Joined
Jan 11, 2020
Messages
225
Likes
214
Ok, this is unique content and you need to continue like this. Measure only stuff noone else measured before. But I am not sure whether methodology is correct (100k load)
 
Last edited:

milosz

Addicted to Fun and Learning
Forum Donor
Joined
Mar 27, 2019
Messages
583
Likes
1,643
Location
Chicago
I wonder about the validity of driving a fixed, purely resistive load with these amps. 100k Ω may seem high, but it's probably an order or two of magnitude lower than an electrostatic driver at audio frequencies. Think about measuring an amplifier designed for an 8 Ω load at 0.08 Ω ....

According to a quick web search, Stax specs the SR-Lambda at 122 pF including the cable; I assume that is per driver. This would have a reactance of 65 megohms at 20 Hz, 1 megohm at 1 kHz and and 65 k Ω at 20 kHz. This characteristic is different electrically from a 100 k Ω load, and I wonder what difference this might make in the measurements.
 

SpaceMonkey

Active Member
Joined
Jan 11, 2020
Messages
225
Likes
214
I wonder about the validity of driving a fixed, purely resistive load with these amps. 100k Ω may seem high, but it's probably an order or two of magnitude lower than an electrostatic driver at audio frequencies. Think about measuring an amplifier designed for an 8 Ω load at 0.08 Ω ....

According to a quick web search, Stax specs the SR-Lambda at 122 pF including the cable; I assume that is per driver. This would have a reactance of 65 megohms at 20 Hz, 1 megohm at 1 kHz and and 65 k Ω at 20 kHz. This characteristic is different electrically from a 100 k Ω load, and I wonder what difference this might make in the measurements.
I wonder if one could extrapolate from measuring using AP with inline resistor in megaohm range.
 

vkvedam

Addicted to Fun and Learning
Forum Donor
Joined
Apr 12, 2019
Messages
583
Likes
807
Location
Coventry, UK
Well, I am not surprised at all. This has just confirmed my assumptions. Thanks Amir!
 

Fishie

Member
Joined
Sep 1, 2020
Messages
15
Likes
10
Location
Hong Kong
Sennheiser HEV 90 and HEV 1060 review next?
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
I wonder about the validity of driving a fixed, purely resistive load with these amps. 100k Ω may seem high, but it's probably an order or two of magnitude lower than an electrostatic driver at audio frequencies.

It is an interesting question. I don't think the answer is clear cut, it isn't quite the same as asking a conventional amp to drive into a nail.

Looking about for some examples, and there isn't a lot of info. A few schematics about, but not exactly the units being tested.

It looks as if the SRM 006t tube design is similar to the original transistor SRM design but with tubes in the output stage. The output devices in both forms use about 50kΩ anode resistors. The 006t is a differential output, but like the SRM-1 mk2 and its ilk, is a single ended differential topology.
The SRM 007t is a very different beast, in particular it has a push pull differential output. Hence the 007t has 4 tubes (each a dual triode) whilst the 006t has two tubes.

No SRM 311 found, but the SRM 323 is looks as if it has a much more conventional topology, and seems to have its output impedance defined with about 5kΩ in series with the output.

So, the 006t tube amp will have its wings noticeably clipped running into 100kΩ. Probably has a hundred or so volts stripped from its output swing.
Not sure about the 007t. But what maybe worse, the load line of the tubes in both amps will be shifted driving the lower impedance, possibly enough to make a real difference to operating conditions and distortion levels. There isn't room for a great deal of forward gain to allow feedback to reign this in, so I do suspect that the load is probably making life hard for the tube amps. The defined gain of at least the 006t is 100. I don't see a huge amount of forward voltage gain in the circuit, so feedback looks limited.

The SRM 323, and I assume, if the 311 is similar, is much more robust, and I would expect that it hardly notices a 100kΩ load.
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
Here is the SMR-006t which is also tube based but its output is not differential:
I think this is not quite right. Assuming I found the right schematic, it looks as if the difference between the 006t and 007t is not that the 006t is not differential, but rather that the 006t is singled ended differential, whereas the 007t is push-pull differential.

ETA. Ugh. Now I'm not so sure about the 007t. In the murk of fuzzy unreferenced schematics, it may just be that the 007t parallels up the halves of the output tubes, and is otherwise much the same as the 006t in topology. Either way, they are differential outputs.
 
Last edited:

30 Ounce

Active Member
Joined
Apr 9, 2019
Messages
155
Likes
217
How old are these? Looks like mains leakage in the 006 (old capacitors?) is adding a lot of noise. Have you replaced the tubes recently? Wonder if new tubes and capacitors or whatever is needed to make the power supply perform better would improve these measurements?
 
Top Bottom