• 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

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
You on the other hand, didn't bother to do a 1 minute search on the impedance of the load and ran off with super faulty assumption. Learn what is important, and what is not.
We know that for all intents the headphones are a capacitive load. At 10kHz they might have 150kOhm impedance. That simply means that at 1kHz we expect 1.5MOhm impedance. At 100Hz 15MOhm.
you measured the amps at 1kHz with an impedance of 100kOhm. That’s is 15 times out of spec by your own metric of using a known headphone. You cannot extrapolate the impedance at 10kHz to the rest of the audio band. It is just plain wrong. It isn’t resistive.
The designer assumed a capacitive load. You have measured into a resistive load delivering a wildly different impedance to the designer’s intent across the majority of the audio band. I don’t think this is the designer’s fault.
 

Newman

Major Contributor
Joined
Jan 6, 2017
Messages
3,448
Likes
4,209
Correct. They don't specify anything else so you can't make the assumptions about half a meg load and such.
Hi Amir, do you have the opportunity to measure the impedance of your personal Stax headphones, across the audible frequency range? It would be interesting.

cheers
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Hi Amir, do you have the opportunity to measure the impedance of your personal Stax headphones, across the audible frequency range? It would be interesting.

cheers
I just don't have the time or interest to keep messing with this, especially given the reception. More Stax amps are coming in the future so I will take another bite out of it then. For now, I need to go and wash the bad taste out of my mouth with some pear juice......
 

bobbooo

Major Contributor
Joined
Aug 30, 2019
Messages
1,479
Likes
2,079
the koss amp is a good example of something that tests great and sounds awful. If you look at the schematic there are lots of hints as to why.
ac coupling between the opamp and the output section causes significant phase shift that varies over the frequency range. minimal bias current in the output stage does not help. poorly matched p and n channel mosfets cause additional issues. etc.

Can you provide objective measurements to explicitly show these supposed audible deficiencies that the Koss amp has (and in comparison with Stax amp measurements)? And have you done blind, level-matched AB listening tests comparing the Koss with a 'better' amp that have led you to the subjective conclusion that the former 'sounds awful'?
 
Last edited:

Veri

Master Contributor
Joined
Feb 6, 2018
Messages
9,596
Likes
12,036
I just don't have the time or interest to keep messing with this, especially given the reception. More Stax amps are coming in the future so I will take another bite out of it then. For now, I need to go and wash the bad taste out of my mouth with some pear juice......
Awww. This thread has been very interesting to peruse through, though.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Indeed. For a forum citing the scientific method as its principle reason for existence, this seems to be a rather undignified and un-scientific conclusion to what could have been an enlightening thread. I'll say no more.
Please... reviews are not science. They Are engineering tests manufacturers should be running but aren't. So please don't use these debating tactics with me. Heaven knows the others had done enough to dissuade me from doing more and didn't need more help from you.

Maybe I should charge to see these reviews so this sense of entitlement goes away. :(
 

mnemonix

Member
Joined
Jul 4, 2020
Messages
86
Likes
118
Location
London
Apologies, I deleted my post thinking it contributed nothing useful before I read your reply, but it was meant simply to express my disappointment that a potentially enlightening thread with clearly knowledgable participants appeared to be cut short due to to an apparent unwillingness to further engage where there still appeared to be valid questions.

There is no sense of entitlement other than a belief that information intentionally placed in to the public domain, particularly when presented as more than simple opnion can and should be challenged if necessary.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
.
Also someone measure the amps with headphones.

Amir already did:
1604307960887.png


At 1kHz I see at least 60dB distance and some of the noise there may well be surround noises picked up by the mic.
I assume Amir used the SRM-007t for this so better than 0.1%.
A.F.A.I.K. the above test was done at 100dB SPL = 100V is close to the 82V (-1.7dB) from the amp test.
Amp test shows 35dB (1.8%) at 1kHz so there is quite some discrepancy here.

1604309891498.png


The AP would be a nice device to measure the amps on but my suggestion would be to use a front-end with no resisitive load (or > 10M) and only a capacitive load similar to actual stats.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
Amir already did:
View attachment 90892

At 1kHz I see at least 60dB distance and some of the noise there may well be surround noises picked up by the mic.
I assume Amir used the SRM-007t for this so 0.1%.
A.F.A.I.K. the above test was done at 100dB SPL = 100V is close to the 82V from the amp test.
Amp test shows 35dB (1.8%) so there is quite a discrepancy here.

View attachment 90893

The AP would be a nice device to measure the amps on but my suggestion would be to use a front-end with no resisitive load (or > 10M) and only a capacitive load similar to actual stats.
But not versus power. That way you'll be able to see the clipping point whether or not is different in both cases.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
What the measurements show is that the amp alone appears to have much, much higher THD than the headphone measurements which clearly shows the measurement method should be looked at again (special front end is needed).
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
yes, that's why I did not quote the clipping point.
Clipping point is related to the bias voltage and max. swing of the amp which will most likely be somewhat lower for higher frequencies because of the load it has and topology of the amp.
Using transformers (and power amp) you can go beyond the bias voltage but will distort and may damage the driver.
Yes. These are the "knowledge" that we know. It's better to settle the dust with a few graphs of measurements.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Yes. These are the "knowledge" that we know. It's better to settle the dust with a few graphs of measurements.

That would require the AP to measure 1000Vpp for which you need a front end that is suited for the task and should have a very high input R and perhaps even have a coupling cap of a few tens nF at least.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
That would require the AP to measure 1000Vpp for which you need a front end that is suited for the task and should have a very high input R and perhaps even have a coupling cap of a few tens nF at least.
It only requires 200V to demonstrate the point. And that's why I asked "someone". Because I see a few people (i know Kevin knows his thing but) out here seemingly know everything. I also would love to see the results.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Why would it be merely 200V ?
When the bias is 580V then one can swing +/-500V thus 1000V differential. (700V rms)
The AP is spec'ed at 230Vp which you really don't want to reach. That won't be a cheap repair.

I assume Kevin has measurements but as he hasn't shared them to serve as evidence I suppose he has his reasons for not posting or linking to measurements. It appears to me that when you design e-stat amps you would have some measurements lying around but maybe not distortion measurements going into the 0.00something numbers.
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
When the bias is 580V then one can swing +/-500V thus 1000V differential. (700V rms)
The bias is a different question to the output swing - it comes from a different source - and changed from generation to generation of headphone despite the amplifiers swinging about the same voltage. The amps pretty universally have +/- 350 volt rails with differential outputs. However the nature of the output devices and circuit limits the swing to rather less than the rails. I would shoot for more like a bit over half what you think.
 

kevin gilmore

Member
Joined
Feb 29, 2020
Messages
54
Likes
259
so the real part of the impedance of the headphones is 170kohm at 10khz. But the current is out of phase with the voltage. You will not get the same test response if you replace the 120 pf cap with a 170kohm resistor at 10khz.

i found my koss box and the power brick. I have not located the koss test connectors yet. i am looking for the 2.5kv rated 100x scope probe so that i
can measure 1200vpp safely. then i will post a picture.

the input circut of the ap is an actively biased lundhall sourced transformer, with an extra winding and feedback that eliminates all parasitic interaction with ground. Very slick. Also patented. (likely the patent has run out by now)
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
The AP would be a nice device to measure the amps on but my suggestion would be to use a front-end with no resisitive load (or > 10M) and only a capacitive load similar to actual stats.

Loading by the AP is the core problem. Looking at the specs it is actually 100kΩ to ground on each differential leg. So it isn't quite as bad as 100kΩ between legs. Still way too low. It also has 190pF in parallel with this load, so it's impedance at 10kHz is already under 50kΩ.

AP's load is very different to the roughly 100pF that the headphones actually provide to the amp.
 

kevin gilmore

Member
Joined
Feb 29, 2020
Messages
54
Likes
259
So since the kgsshv-carbon (and before) i have been using the keithely on the design bench. Unlike any other device out there it can measure thd on signals with significant DC components. So measuring power supplies for example. Or measuring the current sources of the vas stage. But until recently it was not a user friendly device unless you REALLY liked labview. As in piles of labview code to make usable graphs etc. The james m king software changes all of that and turns it into something resembling the graphs people have seen from the ap, dscope etc.

But the fact is for electrostatic amps, once i hit .005% thd i'm done. The gain of these amplifiers is 60db. johnson noise on the input stage at room temperature is the equivalent of about 2 microvolts. So unless you dump the thing in liquid nitrogen, you are never going to get any better numbers out of it. And of course the tube input stage of the t8000 is much worse.

Similarly with dynamic amps, i know how to crank the open loop gain, add bunches of feedback and get results in the .0000x range. I deliberately
choose not to do so. Recently a fair number of people are liking my zero feedback amp which is .006%. The super symmetry version is .003%.
people seem to prefer the zero feedback.

yes i have listened to the koss vs kgsshv-carbon on the same headphones. many times. No i can't even think of a way of actually measuring the differences because the koss clearly tests better. If someone offered me $100k in any test gear i wanted, it would still be doubtful that i could corelate sound quality to any publishable numbers.

striking out so far on finding the hv probe. best i can do is measure 400vpp without blowing stuff up. next step the basement.
 
Last edited:

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686
entertainment

That's the goal I guess... The general pattern seems to be

1) Amir posts some controversial measurements.
2) knowledgeable members object.
3) Amir goes into "charge of the light brigade" mode.
4) objections are refined and discussed.
5) the dust (usually) settles.

The whole process, while turbulent, is beneficial for the rest of us as it gives us the opportunity to learn quite a few things about the matter being fought over.
 
Top Bottom