Electrostatic amplifiers - power requirements

[John_M]

Hi all. This is my first post - I've come over from Head Fi where there has been a slightly frustrating discussion of the above (it quickly descends into a flame war). I was hoping some of the technical experts on this forum would be able to offer some insight (I don't have any real technical knowledge myself!)

In essence there is a view that in order to drive electrostatic headphones to their full potential, it is necessary to purchase a huge, super-powered amplifier. There appears to be no upper limit on this - electrostatic fans will have heard of the huge Stax T2 amp which both busted the company and was also prone to busting itself. Stax no longer make amps this powerful but enthusiasts have been making huge and enormously expensive DIY T2s.

Generally speaking the electrostat community appears to take the view that Stax's own amps are under-powered and can be bettered by expensive and powerful third party amps (Mjolnir Audio amps are the main example) but even these amps are bettered by the enormous and super-powerful/ expensive DIY T2 amps.

At the same time, there are counter-claims being made to the effect that sub-£1,000 Stax amps with (i) next to no distortion, and (ii) far less power than the Mjolnir amps, let alone the T2, are perfectly capable of driving the Stax SR-009 (my headphones) and would therefore be indistinguidable from the T2 save at loud listening volumes. Of course, if you search Head Fi you will easily find posts saying that the T2 will give the SR-009 "superpowers" on another level (and the word "superpowers" has been used in some of the more colourful audiophile descriptions).

A summary of what appears to be the technical basis for these claims can be found here:

https://www.head-fi.org/threads/the-stax-thread-iii.677809/page-1204

See in particular:
- Post #18,049
- Post #18,054
- Post #18,059
- Post #18,060

Can any of you good people help to resolve this issue? Some of the third party amps are large and expensive and I would be reluctant to spend money on them without clear evidence that they do actually make a difference.

Please assume that no owner of any of the expensive amps will be prepared to participate in a properly controlled/ double-blind test of their amps.

 

[solderdude]

You don't need superpowers to drive them.
You just need a high voltage and enough current to charge about 100pF at say max 100kHz.
At that frequency the impedance is about 16k Ohm.
As you can't go above the bias voltage = 580V = 36mA = 116dB (You can't go louder) requires 21W.
For 100kHz at max SPL !!!

Of course there is no need to want 100kHz at max amplitude because these frequencies are never present in music and certainly not at these levels,
A more practical approach would be 50kHz and max -10dB
This changes values drastically.
32k Ohm and 180V peak = 5.6mA peak = 1W while still reaching 116dB peak SPL where it matters, in the lower frequencies because charging 100pF with low frequencies requires very little power.
But... as the current is not in phase with the voltage there are other aspects to consider as well.
MOAR powerrr is not going to help

I didn't read the linked articles b.t.w. no idea what's in them.
As said it is just a very small capacitance that needs to be charged with audio frequencies. This doesn't require a powerhouse.

 

[JustAnandaDourEyedDude]

In essence there is a view that in order to drive electrostatic headphones to their full potential, it is necessary to purchase a huge, super-powered amplifier. There appears to be no upper limit on this - electrostatic fans will have heard of the huge Stax T2 amp which both busted the company and was also prone to busting itself. Stax no longer make amps this powerful but enthusiasts have been making huge and enormously expensive DIY T2s.

Generally speaking the electrostat community appears to take the view that Stax's own amps are under-powered and can be bettered by expensive and powerful third party amps (Mjolnir Audio amps are the main example) but even these amps are bettered by the enormous and super-powerful/ expensive DIY T2 amps.

I am not technically knowledgeable in this area, and certainly would not be able to better solderdude's science-based and precise answer even had I a better technical understanding. Instead, I have a question for you. I am not a head-fi member, so it is lucky for me that a head-fier with a knowledge of, and interest in, estats and their energizers has become an ASR member. Last year, I went through a lot of head-fi and head-case threads about estats, and kept coming across references to a Kevin Gilmore. So my puzzle which you may be able to clear up is: who is KG, what is his technical background, and what is his role in the energizers? I followed that Spritzer owns and runs Mjolnir Audio. Thanks.

 

[John_M]

Kevin Gilmore works with Spritzer on the amps. I believe he designs the circuits but could be wrong.

 

[maverickronin]

So my puzzle which you may be able to clear up is: who is KG, what is his technical background, and what is his role in the energizers?

He's an amateur amp designer (like @solderdude in that he doesn't sell his designs or hardware) who has published schematics. (Or at least he didn't use to sell them. IDK what his relation is with Mjolnir now though.)

I followed that Spritzer owns and runs Mjolnir Audio. Thanks.

Spritzer was resident collector and Stax history expert at head-fi/head-case before he started Mjolnir.

 

[JustAnandaDourEyedDude]

He's an amateur amp designer (like @solderdude in that he doesn't sell his designs or hardware) who has published schematics. (Or at least he didn't use to sell them. IDK what his relation is with Mjolnir now though.)

Spritzer was resident collector and Stax history expert at head-fi/head-case before he started Mjolnir.

Thanks for the info, which clears up my puzzlement! And thanks to John_M as well.

 

[John_M]

You don't need superpowers to drive them.
You just need a high voltage and enough current to charge about 100pF at say max 100kHz.
At that frequency the impedance is about 16k Ohm.
As you can't go above the bias voltage = 580V = 36mA = 116dB (You can't go louder) requires 21W.
For 100kHz at max SPL !!!

Of course there is no need to want 100kHz at max amplitude because these frequencies are never present in music and certainly not at these levels,
A more practical approach would be 50kHz and max -10dB
This changes values drastically.
32k Ohm and 180V peak = 5.6mA peak = 1W while still reaching 116dB peak SPL where it matters, in the lower frequencies because charging 100pF with low frequencies requires very little power.
But... as the current is not in phase with the voltage there are other aspects to consider as well.
MOAR powerrr is not going to help

I didn't read the linked articles b.t.w. no idea what's in them.
As said it is just a very small capacitance that needs to be charged with audio frequencies. This doesn't require a powerhouse.

Thanks for the reply but remember I have very little technical understanding.

In very simple terms would you say that the Stax SRM-353X has enough power to drive the SR-009 assuming the volume control never goes beyond 6/10?

353X measurements (scroll down): https://staxaudio.com/driver/stax-srm-353x
SR-009 (scroll down): https://staxaudio.com/earspeaker/sr-009

 

[John_M]

Also, it would be interesting to know if the site has any electrostatic amp tests in the pipeline...

 

[JustAnandaDourEyedDude]

Thanks for the reply but remember I have very little technical understanding.

In very simple terms would you say that the Stax SRM-353X has enough power to drive the SR-009 assuming the volume control never goes beyond 6/10?

353X measurements (scroll down): https://staxaudio.com/driver/stax-srm-353x
SR-009 (scroll down): https://staxaudio.com/earspeaker/sr-009

Well, solderdude may be able to give you a definitive answer, though the listed specs seem sparse, but just my guess at an answer is as follows. The Stax SRM-353X specs you linked to show it can output 400 Vrms at 1 kHz at its max volume setting. This is about 566 V peak or a little shy of the bias voltage of 580 V of the SR-009 (I would guess this is to avoid damaging the diaphragm and stators). According to the Stax SR-009 specs that you linked to, it has a sensitivity of 101 dB for a 100 Vrms input at 1 kHz. This implies that the 400 Vrms max output of the SRM-353X can drive the SR-009 to 113 dBrms at 1 kHz. And the sound would be higher SPL than 113 dB (but capped at 118 dB) for frequencies lower than that. Next, using my total ignorance of how volume controls attenuate in general and in particular on the SRM-353X as an excuse, I assume that the dB loudness is linear in the knob angle travel. If so, then the 6/10 position would correspond to about 68 dB at 1 kHz.

So, yeah, fairly wimpy loudness at the 6/10 position of volume knob, but enough power at the 10/10 max volume to drive SR-009 to practically its limit with 113 dB rms at 1 kHz. Using solderdude's calculation method, each earpiece should ideally consume about 1.1W at 10 kHz and max knob position (assuming the same 400 Vrms as at 1 kHz). If we assume the energizer outputting the same max of 400 Vrms even at the upper limit 90 kHz of its stated bandwidth, then the worst case power required would be 10 W per channel for a total of 20 W, which the SRM-353X with its max input power of 30 W can handle. Yes, a more powerful and/or more efficient energizer could drive the SR-009 to its limit at the 6/10 knob position, but then you run the risk of damaging the SR-009 by carelessly cranking up the volume knob [EDIT: and additionally it would likely give you less useful volume knob adjustment travel than the SRM-353X; if linear, the latter gives you a useful adjustment range of roughly 6.5/10 (73dB rms, 76dB peak) to 10/10 (113dB rms, 116dB peak)]. Check out Harmful Noise Levels or similar internet info for examples of dB levels. If solderdude or someone else as knowledgeable replies to your question, I will read his answer to see how bad my misunderstanding of estat and energizer power and volume knobs is

EDIT: The above was edited to remove what were likely erroneous estimates of loudness at 10kHz. In reality, I am sure Stax designed the energizer so that the SR-009 earspeaker's frequency response would be flat in its normal operation for audio frequencies up to and above 20 kHz. The listed Stax specs do not show any sensitivity/efficiency figure for frequencies other than 1 kHz.

 

[John_M]

Thanks for the reply.... Just to add some more detail - I own the 353X and the 009 but might buy a more powerful amp. I listen at volume 5-6 depending on the track. Any higher is uncomfortable. I don't really find it 'wimpy' at those sound levels (and my wife listens on a 4 volume). I don't want to drive the 009s to their absolute limit - just interesting in testing the audiophile claims that even at low volume levels a more powerful amp will provide (for example) better bass. I asked this on head-fi and got the response that someone who was listening to a different Stax headphone on the 353X at volume 3/10 still saw a noticeable improvement on switching to a more powerful Mjolnir Audio amp.

 

[John_M]

Here is a typical example of what I'm talking about:

https://www.reddit.com/r/headphones/comments/56oeki
The 323s has the same power as the 353X, which I own. You can see that the guy has gone to a much more powerful amp and says that it has, amongst other things, improved the bass impact, mostly sub bass. He has concluded this based on a sighted test and having already been told by numerous people that this is what will happen. Other posters say that the 323S isn't bad but the more powerful Carbon is 'A LOT' better.

I'm not sure about the reliability of a sighted test on a very expensive amp (the Carbon) from someone who has already been given preconceptions about what to expect. However, I'm assuming no owner of a Carbon will subject it to blind testing under properly controlled conditions, hence asking for views on here...

 

[imrul]

So the thing about tube amplifiers is that tubes are added to purposely distort the sound, this makes any claims of superior sound be subjective. Although Carbon isn't exactly a tube amplifier, it uses SiC FETs, which are very close and thus would modify the sound in a way which may or may not sound good (I personally despise the tube sound, while others seem to like it).

So tubes aside, the measure of an objectively good amplifier is its signal-to-noise ratio, signal-to-distortion ratio, and power. Combining signal-to-noise and signal-to-distortion gives you signal/(noise+distortion) which is SINAD. A SINAD of 120dB generally means any noise or distortion is imperceptible and the sound is transparent (all you hear is the sound coming through without being colored by noise or distortion). While 120 is objectively transparent, smaller numbers are generally good enough -- most humans won't be able to detect noise or distortion with a SINAD of 100dB for example unless listening very loud.

What you want is an amp that can deliver all the power you need without losing ground on SINAD. This is why some very low powered electrostatic amps start to sound distorted or harsh at higher volumes -- good engineering is maintaining a very high SINAD while providing all the power a user will need.

So with the case of the Carbon, since we don't have actual measurements, you may be able to get a ballpark sense by looking at the circuit layout. To achieve a very high SINAD and low distortion you would likely need a lot of feedback, but this reduces the gain, meaning you may need more gain stages which creates their own challenges. You would need very high bias currents (to lower distortion) which also generates a lot of heat and is power hungry. It is possible that the Carbon is doing this but you'd have to look at the circuit schematics or the topography, but alas its kind of a moot point because its trying to distort the sound on purpose with it's SiC FETs.

If I had to guess, the objectively best amp from mjolnir is probably the KGSSHV ( the non-carbon version).

Another route you could take is to get the Benchmark AHB2 power amplifier, which is the best measuring power amp that's been tested here, and then feed that to an energizer such as the SRD-7 or SRD-7 Lundahl. But you would still need to determine how much noise and distortion those transformer boxes add as well.

 

[John_M]

That is an interesting reply... I saw about the SiC FETs but didn't really understand what they were. I don't want any part of an amplifier which is designed to add distortion which someone subjectively considers to be euphonic. Maybe the designers also think the SR-009 is too bright and are trying to modify its sound. As regards the KGSSHV - I might get their KGSSHV mini but does the suggestion of using the Benchmark amp + SRD-7 suggest that you think there might be something in the "moooooore powerrrrrr" theory?

 

[imrul]

So the reason I mentioned the benchmark amp was that it has the highest signal-to-noise and signal-to-distortion ratios out of any power amp we've measured here. It certainly has power, but I'm not suggesting it for its raw power -- but rather for its ability to deliver that power with minimal distortion and noise. Of course, it wouldn't matter if the SRD-7 itself has weak SINAD measurements.

 

[JustAnandaDourEyedDude]

Thanks for the reply.... Just to add some more detail - I own the 353X and the 009 but might buy a more powerful amp. I listen at volume 5-6 depending on the track. Any higher is uncomfortable. I don't really find it 'wimpy' at those sound levels (and my wife listens on a 4 volume). I don't want to drive the 009s to their absolute limit - just interesting in testing the audiophile claims that even at low volume levels a more powerful amp will provide (for example) better bass. I asked this on head-fi and got the response that someone who was listening to a different Stax headphone on the 353X at volume 3/10 still saw a noticeable improvement on switching to a more powerful Mjolnir Audio amp.

Quite welcome. Interesting to know that your wife and you find the range 4 to 6 volume to be comfortable. The action of the volume knob was probably designed to be a little nonlinear, to increase the useful travel.

Here is a typical example of what I'm talking about:

The 323s has the same power as the 353X, which I own. You can see that the guy has gone to a much more powerful amp and says that it has, amongst other things, improved the bass impact, mostly sub bass. He has concluded this based on a sighted test and having already been told by numerous people that this is what will happen. Other posters say that the 323S isn't bad but the more powerful Carbon is 'A LOT' better.

I'm not sure about the reliability of a sighted test on a very expensive amp (the Carbon) from someone who has already been given preconceptions about what to expect. However, I'm assuming no owner of a Carbon will subject it to blind testing under properly controlled conditions, hence asking for views on here...

The question of sound quality at low volumes is the counterpart to what member imrul describes in the quote below (the high SINAD high power amp). The SRM-353X specs say that the max harmonic distortion is 0.01% of a 100V signal fed to a SR-L500, which is a signal-to-harmonic-distortion ratio (SDR) of 80dB. The SDR is somewhat low by the standards of today's best headphone amps, which reach SDRs of 120 (though of course a headphone amp and an energizer are different beasts). But Stax do not say what the signal-to-noise ratio (SNR) is. Generally, if you dial down the volume, the SNR drops because the broadband noise level stays the same but the signal (music) level drops, and the noise may become audible at lower music volumes due to less masking. Distortion on the other hand usually scales with the signal, so as member imrul points out, you would have to listen to music at loud volumes to notice small distortion levels. I think that designing a good energizer is more challenging than designing a good headphone amp of the normal variety, because the energizer amplifies the input voltage by up to a factor of 1000.

But that is not what the guy who owns the rig in the photo and those folks who own the Carbon are talking about it fixing. [The Mjolnir in the photo certainly dwarfs the Stax.] They are talking about an improvement in bass quality, which on the surface sounds a little dubious to me. From what solderdude posted, the lower frequencies of electrostats require less power to drive at a fixed stator voltage amplitude than higher frequencies do due to the capacitive nature of the load. Stax themselves list 400 Hz as being where the max SPL of 118 dB occurs. Tones below that require even less power than a 400 Hz tone to reproduce, and I imagine the main reason why the SPL drops off at increasingly lower frequencies is because the diaphragm excursion is constrained to a fixed distance (that between the stators), and so the diaphragm is unable to move the increasing volume of air needed at decreasing frequency to achieve a fixed SPL. So the SRM-353X should have no problem dealing with bass relative to mids or treble, at any volume knob setting. How would more power help with the fixed excursion problem at sub-bass and low bass frequencies? The SRM-353X already has an extended treble response up to 90 kHz, so transient response speed should not be an issue at any audible frequency including bass and sub-bass. How would more power reduce any distortion or noise in the bass, since it seems the SRM-353X has adequate power (and should not reach a clipping state) to drive the SR-009 to its limits?

However, as I said, I know little about estats and particularly their physics, how the electric field force drops off with distance and charge, etc. in a real geometry. The earspeaker and energizer need to be matched more than with a normal headphone and amp, because in the former case the earspeaker impedance varies much more widely with frequency than does a headphone's impedance. Perhaps there is some factor I am unaware of that causes the more powerful amps to improve the bass response at low volumes. Folks like Spritzer and Kevin Gilmore have a pro level understanding of estats and energizers. And then also, the clever people at Stax thought it worth designing, building and selling the T2 in search of giving their customers better sound out of the SR-009. As you point out, controlled measurements would tell, and sighted tests performed by someone who has been fed an expectation bias are not reliable when it comes to subtle effects at the fringes of our hearing ability.

What you want is an amp that can deliver all the power you need without losing ground on SINAD. This is why some very low powered electrostatic amps start to sound distorted or harsh at higher volumes -- good engineering is maintaining a very high SINAD while providing all the power a user will need.

Another route you could take is to get the Benchmark AHB2 power amplifier, which is the best measuring power amp that's been tested here, and then feed that to an energizer such as the SRD-7 or SRD-7 Lundahl. But you would still need to determine how much noise and distortion those transformer boxes add as well.

SRD-7 etc. are an attractive option, but as you say, unknown in terms of noise and distortion. And there is a similar transformer-based product from Woo Audio. Does anyone have experience with the iFi Pro iESL? It is also a transformer-based product, and has the advantage of multiple bias-voltage settings. I read Spritzer's unflattering take on it, and iFi sorta responded on their own website ridiculing Spritzer's claims. I was considering the iESL last year and beginning of this year, in conjunction with a Kaldas Research RR1 Conquest estat HP (the latter something I could afford at $500, unlike a SR-009). However, some iFi products have tested poorly here on ASR, and that makes me wonder about the build quality and noise/distortion of the iESL in spite of iFi's impressive marketing claims for it.

 

[kevin gilmore]

so much wrong here

the headphone is a capacitor, the voltage and current are 90 degrees out of phase. That pesky little "jw" term.

the carbon has 20db of feedback at 1khz, 6db of feedback at 20khz. and less than .01% thd at full power 20hz to 20khz into 1 standard load.
for reference the magni3+ has 30db of feedback at 1khz. and better thd numbers as a result.

the carbon has 56db of gain (balanced input). (some people change this a bit)

could i raise the open loop gain to allow for 30 or 40db of feedback and reduce the distortion to .001%, unlikely
and neither can anyone else including stax or any of the transformer things. Besides which testing to these levels
at 1kv voltage swings is seriously problematic.

to drive the headphones at 0db reference 1khz at 20khz and maintaining 6db of feedback requires...

at 450v power supplies, 18ma
at 400v power supplies, 16ma
at 350v power supplies, 14ma (most stax amps run at this voltage, but none of them come close at 5 to 6ma typical)

Even the T2 at 500v supplies is only biased at 15ma

tubes do not add distortion, not in this case, tubes love high voltage, silicon does not. The emission labs 20b is THE most linear and low distortion
output device for electrostatic amps (at 400v power supplies). The silicon carbide j-fet is very close in performance. Far better than the 2sc4686a

 

[imrul]

tubes do not add distortion, not in this case, tubes love high voltage, silicon does not. The emission labs 20b is THE most linear and low distortion
output device for electrostatic amps (at 400v power supplies). The silicon carbide j-fet is very close in performance. Far better than the 2sc4686a

Interesting. What about this case makes it so that tubes don't add distortion? And in which case would it add distortion? What are the differences between the two?

Also, going off the description in Spritzer's Carbon amp, it says "The original design concept for this amp was born to tame the rough top end of the SR-009". What does this mean exactly? Isn't purposefully trying to change the sound distorting it?

Thanks

 

[kevin gilmore]

big difference between triode running class A and a tetrode or pentode wired to a transformer running class B into a low impedance load.
remember that all tube tektronix scopes were high accuracy dc coupled designs well into the 100's of mhz region. tubes don't have to produce
distortion. atmasphere amps are one really good example of current production very low distortion fast tube amps. No transformers and only one
really good cap per channel.

The sic-fet runs at more than 3 times the output current of previous solid state designs. So faster slew rate and lower distortion at high frequencies.
You can look at this as the equivalent of a higher damping factor so resonances in the diaphram are better controlled. I design the best amp i can
with the parts that are available. I don't deliberately mess with the frequency response or anything like that.

 

[raindance]

I'm wondering when more folks will start realizing that where a volume control is pointing has no relationship to power output. It's arbitrary and tied to the taper of the pot (if it's a pot) plus the gain of the amp plus the output voltage of the source plus the recording / mastering level of the source.

I've seen many people "upgrade" their amp to one that sounds exactly the same and outputs similar power but has more gain because they "feel" it's more powerful when needing to turn up the volume control less.

There, my rant for today

 

[DonH56]

Much as I love old Tek (etc.) tube scopes, high accuracy relative to the kinds of numbers we are talking today is not one of their strengths, and they tended to drift so you had to recal them fairly often. Most DSOs today use 8-bit converters, with some having 12-16 bits at generally lower speeds, and are much more stable.

Pure class B operation generates a goodly amount of crossover distortion (tube or SS).

Tube circuits tend to be single-ended and thus have higher even-order distortion (though this may be preferred sound-wise). Thermionic devices also tend to have higher noise in general, and of course in differential operation precisely matching tubes is tough.

Interesting factoid: A perfect tube's distortion expansion is factorial while for a bipolar transistor it is exponential so a tube actually has intrinsically lower distortion. An ideal MOSFET has second harmonic distortion and that's it, no higher terms. In the real world, circuit design (including feedback) and non-ideal devices make that fairly meaningless.

Edit: I missed that this is about headphone amps, not preamp or (speaker-level) power amps. Tubes still add noise and distortion, everything does, and what I said I believe to be true, but less relevant for headphones. IME hiss (noise) is a bigger problem with tubes than for SS headphone amps.