Stax Tube Amplifier Distortion vs Solid State

[Francis Vaughan]

That measurement has bad seal; the bass should extend flat on these headphones. The pads are shaped to fit a human head without having to clamp much, which makes sealing on an artificial head problematic.

I own a pair of Lambdas, so I am familiar with the fit. I agree, it is easy for a poor fit on the test system to mess up the measurements. All I was doing was looking for some confirmation that one could end up with test results as poor as Amir is getting.

I guess we need to see Amir's SR-303 tested on this system to put a baseline under the amplifier tests. If the HPs on the test rig are not providing a representative FR, it would be good to know. No matter what, the tests currently have the HPs 20dB down at 20Hz. After that both amplifiers are just doing the best they can, and behaving pretty much exactly as one would expect.

 

[Degru]

No matter what, the tests currently have the HPs 20dB down at 20Hz. After that both amplifiers are just doing the best they can, and behaving pretty much exactly as one would expect.

We don't know that unless he releases frequency response. And as I mentioned unsealed distortion from Lambdas is a LOT higher than what was shown here at those levels. It would be hitting 20% or higher; this looks like it's just the amplifier distorting since it's being run at its limits for those freqs.

I have no way to level calibrate this properly, but this was definitely nowhere close to 90db; I definitely don't have Amir volume tolerances and I could still listen to the test tone with the headphones on my head.
This was done with a 404LE and the PCB and mic capsule from an EARS put into a DIY flat plate rig (basically a foam yoga block with a hole cut into it). The bass distortion consistently hit as high as 20% when I raised the volume higher, which is a far cry from 5% @ 97db that Amir showed.

 

[Maki]

We don't know that unless he releases frequency response. And as I mentioned unsealed distortion from Lambdas is a LOT higher than what was shown here at those levels. It would be hitting 20% or higher; this looks like it's just the amplifier distorting since it's being run at its limits for those freqs.

View attachment 102318
View attachment 102319
I have no way to level calibrate this properly, but this was definitely nowhere close to 90db; I definitely don't have Amir volume tolerances and I could still listen to the test tone with the headphones on my head.
This was done with a 404LE and the PCB and mic capsule from an EARS put into a DIY flat plate rig (basically a foam yoga block with a hole cut into it). The bass distortion consistently hit as high as 20% when I raised the volume higher, which is a far cry from 5% @ 97db that Amir showed.

It isn't as simple as "sealed" vs "unsealed". It's a gradual scale, and the better the seal is, the more linear the bass frequencies are, with a perfect seal having linear bass and no bass hump at all.

 

[Francis Vaughan]

this looks like it's just the amplifier distorting since it's being run at its limits for those freqs.

I think we are in complete agreement. My point about the frequency response is that the absolute volume level at which clipping is seen is very significantly down. I don't think the HPs are distorting much at all. But the issue is that the amplifiers have both been run to clipping, and for whatever reason, the absolute volume levels measured at 20Hz are not as high as one might expect (or hope for) with these HPs when the amplifiers are flat out.
I don't think we learnt anything new about the ampifier performance compared to the 1kHz measurements; since the load at 20Hz is about the easiest the amplifiers will ever see, this isn't unexpected. Electrostats are a pretty easy load to charaterise. They give trouble at higher frequencies because they are basically a near perfect capacitor.

 

[pma]

As this thread is named
Stax Tube Amplifier Distortion vs Solid State
I am quite sure that in case of amplifier measurements we measure distortion against electrical values, namely voltage and frequency. Not with respect to the SPL, which is an acoustical value. Amplifier's output is voltage, not sound pressure. The SPL brings another variable which is dependent on the test setup. As presented in @amirm test review, the result at 20Hz is valid only and only for his unique measurement setup and cannot be generalized, so no conclusions on amplifiers clipping at 20Hz is possible to make. Now it throws bad light to the amplifiers under test and this should be IMO corrected and explained in the review. This would be the scientific approach.

 

[Degru]

I think we are in complete agreement. My point about the frequency response is that the absolute volume level at which clipping is seen is very significantly down. I don't think the HPs are distorting much at all. But the issue is that the amplifiers have both been run to clipping, and for whatever reason, the absolute volume levels measured at 20Hz are not as high as one might expect (or hope for) with these HPs when the amplifiers are flat out.
I don't think we learnt anything new about the ampifier performance compared to the 1kHz measurements; since the load at 20Hz is about the easiest the amplifiers will ever see, this isn't unexpected. Electrostats are a pretty easy load to charaterise. They give trouble at higher frequencies because they are basically a near perfect capacitor.

No, you're misunderstanding me. If the acoustic level was much lower relative to the mids at same voltage level due to bad seal we would see MUCH worse distortion than Amir measured coming from the headphones themselves. That he's only getting 1-5% indicates this is simply the amplifier distorting and clipping earlier at low frequencies than at 1khz, not that it is struggling to push very rolled off bass to 100db.

 

[Francis Vaughan]

clipping earlier at low frequencies than at 1khz,

Why would they both do this? I can't think of any mechanism common to the design of the amplifiers that could cause such behaviour.

If the acoustic level was much lower relative to the mids at same voltage level due to bad seal we would see MUCH worse distortion than Amir measured coming from the headphones themselves.

I'm not convinced. Yes there is a steep roll off that would accentuate the headphone distortion. But we have a 20Hz signal. The second harmonic is 40Hz. That is probably also suppressed by the poor seal, we might assume by about 10dB. The third harmonic is 60Hz and probably measured at full volume. There probably is not much distortion energy higher than this, but indeed, if it is there is will be over emphasised in the numbers by 20dB. If the distortion is 1% as measured, the worst case change would be to be over emphasising a 0.1% distortion. Given the 2nd harmonic is also likely suppressed, by say 10dB, we could get away with more like 0.3% distortion given the usual distortion profile. I don't know what we expect from these headphones at these frequencies, but it doesn't seem totally unreasonable.

Clearly both amplifiers are clipping. They are clipping at output relative output levels that is consistent with the output levels they clipped at 1kHz. Yet they are different internal designs. Somehow there would need to be a common design flaw that caused the voltage rails to fall to 10% of their normal voltage at 20Hz in exactly the same way. Given the amplifiers are delivering essentially no power, and have, for the purpose, grossly over-designed power supplies, I can't think of a way this could happen. A flaw like this would cause almost impossible to manage intermodulation distortion on real world signals. Every bass hit would cause the voltage rails to collapse, and everything would clip. That would be an impossible to ignore design fail.

 

[Degru]

Every bass hit would cause the voltage rails to collapse, and everything would clip.

Cuz that's what actually happens, at these volumes anyway.

Yes there is a steep roll off that would accentuate the headphone distortion.

No, the driver itself distorts much more when it's operated unsealed. I figured out how to do level calibration on my measurement microphone. Here is what an unsealed Lambda does when you try to make it produce 20hz at just below 80db:

I'm getting 93% 5th order distortion and 40% 3rd order, and this is at levels that are lower than what Amir started at. If Amir's headphone was operating unsealed, he would be seeing far worse and possibly damaging his stax. My previous 10-20% figures were assuming levels referenced to 1khz, but if you actually take a 20hz tone and crank it up (which is what Amir did) you get insane amounts of odd order distortion from the driver operating outside of its intended usage scenario.

I can't begin to explain why exactly these amps are distorting and clipping so much because I am not an electrical engineer, but I do know that it is in fact the amp behaving this way on its own and not as a result of the headphone being rolled off due to bad seal. If that was the case, we'd have over 100% distortion, not 5%.

 

[Francis Vaughan]

OK, numbers help a lot.
I think we have a bit of a conundrum here. I don't think it is likely that the amplifiers have a common design flaw. We have the schematics, there isn't any reason to imagine they are flawed like this. But the numbers you show are very interesting. What amplifier are you using?

What about a partial seal? How much can you crack the seal and get some intermediate changes in operation?

 

[Degru]

OK, numbers help a lot.
I think we have a bit of a conundrum here. I don't think it is likely that the amplifiers have a common design flaw. We have the schematics, there isn't any reason to imagine they are flawed like this. But the numbers you show are very interesting. What amplifier are you using?

What about a partial seal? How much can you crack the seal and get some intermediate changes in operation?

I'm using an adcom gfa-545ii through an SRD7. I can't test distortion on full or partial seal because my rig is a flat plate; there is a significant gap due to the contours of the pads. But with some experimentation using an in-ear mic on my head, messing with seal changed the location of the resonance peak, with the same sharp rolloff and presumably the same sharp rise in distortion below it. I could not measure the actual distortion because of some resonance with the earplug I was mounting the mic onto causing it to measure upwards of 300% distortion in the bass lol

It would be quite helpful if @amirm provided a frequency response measurement of the headphone in the same position he measured the 20hz distortion at, to clear all of this up

 

[Annamarie]

From these reviews it seems that Stax’s strength is in designing electrostatic headphones. Their circuit design skills are not at the same level and these ‘phones all need bespoke drive circuits. The current circuits don‘t do justice to them.

Perhaps they should strengthen their analog electronics capability?

Query, i am going to but L700mk2, i am feeling that Ifi Iesl and ican is a far better choice for energiser

 

[Annamarie]

Query, i am going to buy L700mk2, i am feeling that Ifi Iesl and ican is a far better choice for energiser

 

[milosz]

Buy a KGSSHV you will not regret it. You'll have to look around for one - they do become available used or made by small Chinese builders on eBay from time to time, or you can buy one from Mjölnir-Audio

The KGSSHV Carbon is a step or two better, but more expensive.

Many Stax users are happy with a KGSSHV or Mini KGSSHV. I think the KGSSHV sounds better than the IFi Lesl

You can also build a KGSSHV yourself, if you have DIY electronics skills.

 

[amadeuswus]

> I assume the SR-303 is plugged into a "pro" high bias voltage socket (580v) , and not the low (230v) socket.

If it was plugged into normal bias, Amir would have a dead lambda on his hands right now You can really only run it at extreme low (barely audible) volumes on normal bias without too much risk of permanent damage. I know someone who got impatient waiting for a piece of gear to arrive and plugged a Lambda Signature into normal bias, and ended up damaging it after a few minutes of low volume listening. Squealing, imbalance, bass completely messed up sounding, basically totally broken.

In a parallel Stax amp thread, I asked whether I would be sacrificing anything other than ultimate output level by driving a Stax Lambda Pro with the original SRD-7 adaptor (non-Pro). The experience you relate is alarming! I do not listen loudly and I have done this in the past without noticing a problem. May I ask why using the lower bias voltage would tend to damage the Pro headphones? Thanks very much.

 

[lewdish]

Seems like a shame that there isn't any known high performing EST amps in regards to Sinad, which seems like it would be a shame given the performance oriented nature of electrostatic drivers in general.

 

[Ra1zel]

Seems like a shame that there isn't any known high performing EST amps in regards to Sinad, which seems like it would be a shame given the performance oriented nature of electrostatic drivers in general.

It's not as bad as you color it, amps like BHSE, T2 or Carbon are all relatively low distortion, and more importantly compared to Stax amps, remain low distortion even at high voltage (T2 measures 0.002% or better at 500V). You need to understand that the nature of this circuits simply doesn't allow for ultra low distortion.

On the other hand stax amps are mostly terrible from my experience, the best from current production would be SRM-700S but even this sounds a little anemic.

Alternative route and probably best performance you can achieve is very low distortion power amplifier like AHB2 hooked up to iFi iESL.

 

[milosz]

It's not as bad as you color it, amps like BHSE, T2 or Carbon are all relatively low distortion, and more importantly compared to Stax amps, remain low distortion even at high voltage (T2 measures 0.002% or better at 500V). You need to understand that the nature of this circuits simply doesn't allow for ultra low distortion.

On the other hand stax amps are mostly terrible from my experience, the best from current production would be SRM-700S but even this sounds a little anemic.

Alternative route and probably best performance you can achieve is very low distortion power amplifier like AHB2 hooked up to iFi iESL.

I found the difference between using my Stax SR-007 mk 1's on my KGSSHV amp and the Stax SRM-T1 I used to own to be really AMAZING. The Stax amps are really not very good.

 

[Zoide]

I wonder how the KGSS Klassik stacks up against some of the other amps mentioned here.

 

[rocksteady]

Add eq to the mix and things will get from bad to unbearable real quick. I love electrostatic headphones but never found an amp that totally cuts it with this problem. It’s the reason to buy a SRD7 mk2 transformer and use with a very low distortion and powerful speaker amp.

Or better still, buy a Woo Audio Wee and be transported to Audio Nirvana. I feed mine with a humble, but very good NAD 2100 Monitor and the combination is unbeatable for Resolution, Dynamics and Speed. Not to mention Powerful, Taught and extended Bass! In my experience, far better than the admittedly short range of Direct Stax Drivers that I have tried so far. A recently loaned SRM-T1S proved to be particularly awful, to be brutally frank…

 

[Tom C]

Necro