Harshness and Image Instability in Aging ESL Hybrids: A Phase-Timing Hypothesis by ChatGPT

[Chris A]

The point of this topic was just to see whether, at least in this particular case, the robot’s summary of responses was even in principle in the ballpark. I wasn’t trying to actually diagnose or debug anything at that point.

There's your lead sentence needed at the top of post #1 in this thread. I would add that you also need to think about whether or not it's noteworthy for this forum.

There are a lot of folks here who are grasping for clues in a pitch-black room (i.e., those that ask questions). I think this forum's major purpose is to help answer those questions, when they can.

Telling people about what they've found relative to audio, etc. can also work on this type of forum (as I have done extensively in other audio forums*), but typically in a DIY or tutorial sense--like the practices that you see on diyAudio.

But again, it's courtesy to think about who might benefit, and make it clear to that target audience.

Chris

* for instance...

Survey results of forum member listening room sizes: https://community.klipsch.com/index...om-sizes/page/3/&tab=comments#comment-2038465

How to use Room EQ Wizard to set up DSP crossovers: https://community.klipsch.com/index...ew-to-find-parametric-equalizer-peq-settings/, and https://community.klipsch.com/index...rew-to-determine-time-delays-between-drivers/

DSP crossover FAQ: https://community.klipsch.com/index.php?topic/117543-active-bi-ampingtri-amping-faq/

Corner Horn Imaging FAQ: https://community.klipsch.com/index.php?/topic/131163-corner-horn-imaging-faq/

Why Horn-Loaded Sounds Better Than Direct Radiating: https://community.klipsch.com/index...aded-sounds-better-than-direct-radiating-faq/

K-402-based Multiple Entry Horn (MEH): https://community.klipsch.com/index.php?/topic/161404-a-k-402-based-full-range-multiple-entry-horn/

How to Demaster your Own Stereo Music Tracks Using Audacity: https://community.klipsch.com/index.php?/topic/155096-the-missing-octaves-audacity-remastering-to-restore-tracks/

The Historical Loudness War Effects on CDs and LPs: https://community.klipsch.com/index.php?/topic/143320-loudness-war-and-the-dynamic-range-dr-database-some-observations/

Subconscious Listening Effect of Linear Phase Loudspeakers Having Full Range Directivity (and How To Achieve Linear Phase Without FIR Filters): https://community.klipsch.com/index...y-effects-of-quasi-linear-phase-loudspeakers/

Understanding the effects of "minimum phase" loudspeakers in real listening rooms: https://community.klipsch.com/index.php?/topic/173932-chris-a-could-you-explain-minimum-phase-analysis-please/

Review and Discussion of Floyd Toole's 3rd Ed. book: https://community.klipsch.com/index.php?/topic/172272-review-and-discussion-of-tooles-book-third-ed/
_______________________________________________________________________________________________

The above list is not exhaustive of the tutorial threads I've posted, and in the above case just to one forum.

Chris

 

[antcollinet]

Still, with careful phrasing, you can usually steer around it.

You can't. When the training set consists of 90% bullshit - the LLM has no chance of generating anything useful.

As stated above - if it can't give verifiable references there is no point using it.

 

[tengiz]

You can't. When the training set consists of 90% bullshit - the LLM has no chance of generating anything useful.

As stated above - if it can't give verifiable references there is no point using it.

If you consistently ask it to provide verifiable sources, it will usually either do so or return a non-answer. I don’t see anything wrong with that - it just means what you’re looking for is beyond what this particular model can deliver.

Of course, there's also the separate issue of what qualifies as a “verifiable source” and why such a source should be trusted - even when it appears totally legit.

That said, I understand that some members of this community may find such an approach distasteful or vulgar. Fair enough - point taken.

 

[Sir Sanders Zingmore]

Don’t we have no “I asked ChatGPT….” threads policy

 

[Keith_W]

Anyway, since you asked about degradation / harshness of Martin Logan speakers as they age, I asked claude.ai the same question.

Q: What issues arise when electrostatic speakers age?

Diaphragm degradation is the most common problem. The ultra-thin plastic film (typically Mylar) that serves as the diaphragm can develop microscopic holes, tears, or lose its electrostatic charge over time. This leads to reduced output, distortion, or complete failure of sections of the panel.

Dust accumulation poses a significant challenge since the high voltage attracts particles to the stators and diaphragm. Even small amounts of dust can cause arcing, reduce efficiency, and create audible crackling or popping sounds. The dust is particularly difficult to remove without damaging the delicate diaphragm.

High voltage supply degradation affects the power supply components that maintain the several thousand volts needed for operation. Capacitors, transformers, and other HV components can fail over decades, leading to reduced bias voltage and diminished performance.

Environmental sensitivity increases with age. Humidity changes cause the diaphragm material to expand and contract differently than the metal stators, potentially causing sag or contact issues. Temperature fluctuations can also affect the plastic's properties and the precise spacing critical for proper operation.

Stator corrosion can occur over time, especially in humid environments. The perforated metal plates may develop oxidation that affects their electrical properties and acoustic transparency.

The good news is that many of these issues can be addressed through professional refurbishment, including diaphragm replacement and cleaning, though this requires specialized knowledge due to the high voltages involved.

To me, the answer seemed believable. About the only point I wasn't sure about was diaphragm degradation. So I asked it to cite references.

Note that I am not asking claude to think for me, I am asking claude as a tool to find resources which were written by humans. I then look at the links it generates, and decide if it's worth reading or not. Some of those links are quite credible.

Of course I am going to ignore the Stereophile and PS Audio pages it used to help generate the answer. But there are some academic publications in there, and a discussion from Hydrogen Audio (those guys are even more hardcore no-BS objectivists than we are!).

Q: Reference for diaphragm degradation please

electrostatic speaker diaphragm degradation aging mylar film

10 results

ElectroStatic Solutions - Elements

estatsolution.com

Electrostatic Loudspeaker Design by Neil S. McKean

audiodesignguide.com

best plastic film for ESL speakers | diyAudio

diyaudio.com

Electrostatic loudspeaker - Wikipedia

wikipedia.org

Quad ESL-989 electrostatic loudspeaker | Stereophile.com

stereophile.com

Electrostatic Speaker Diaphragm Assembly - Ken Seibert

kenseibert.com

Statclear and ESL Audio Speakers - Electroguard Anti-Static Paint

electroguardpaint.com

ESL Diaphragm material, Anyone tried different? | diyAudio

diyaudio.com

Electrostatic Loudspeakers: High End HiFi You Can Build Yourself | Hackaday

hackaday.com

(PDF) Vibration and sound radiation of an electrostatic speaker based on circular diaphragm

researchgate.net


electrostatic speaker mylar diaphragm aging failure research
10 results

ElectroStatic Solutions - Elements

estatsolution.com

Quad ESL-989 electrostatic loudspeaker | Stereophile.com

stereophile.com

Electrostatic Loudspeaker Design by Neil S. McKean

audiodesignguide.com

(PDF) Vibration and sound radiation of an electrostatic speaker based on circular diaphragm

researchgate.net

Electrostatic loudspeaker - Wikipedia

wikipedia.org

How electrostatic loudspeakers work – PS Audio

psaudio.com

The Pros & Cons of Electrostatic Loudspeakers | Stereophile.com

stereophile.com

Electrostatic speaker myths

hydrogenaud.io

Electrostatic Loudspeakers: High End HiFi You Can Build Yourself | Hackaday

hackaday.com

Electrostatic Graphene Loudspeaker

berkeley.edu
Based on the search results, here are several references that document diaphragm degradation in electrostatic speakers:

Professional repair service documentation:ElectroStatic Solutions, a professional ESL repair service, states that "Diaphragms may also fail due to rips caused by element arcing or ageing and fatigue of the Mylar" and notes that "speakers used in high humidity environments and left constantly energized suffer the greatest leakage damage." EstatsolutionPS Audio

Academic/Technical sources:Wikipedia's article on electrostatic loudspeakers notes that "Because of their tendency to attract dust, insects, conductive particles, and moisture, electrostatic speaker diaphragms will gradually deteriorate and need periodic replacement." Electrostatic loudspeaker - Wikipedia

Research literature:UC Berkeley research on electrostatic speakers mentions that "If conventional materials such as metalized mylar are made too thin, they invariably fatigue and break."

Technical design documentation:The Audio Design Guide by Neil S. McKean discusses the structural requirements for Mylar diaphragms and the challenges of maintaining stability under both static and dynamic conditions, though it focuses more on design than aging issues.

These sources confirm that Mylar diaphragm degradation is a well-documented phenomenon in electrostatic speakers, with aging, fatigue, environmental factors, and electrical stress all contributing to gradual deterioration over time.

 

[tengiz]

Anyway, since you asked about degradation / harshness of Martin Logan speakers as they age, I asked claude.ai the same question.

Sorry for the long-winded testimonial below... but I can’t compress it further without losing clarity:

It began with a specific kind of vocal harshness that was highly sensitive to small changes in speaker placement and EQ, especially near the crossover. It didn’t seem related to panel wear per se - the panels were replaced four years ago, and that earlier issue had been obvious and fully resolved.

This new problem developed gradually after years of nominal performance. At first, I assumed it was just age-related hearing loss, but it became too pronounced to ignore - especially since it didn’t occur with other speakers or headphones.

It often sounded like harmonic distortion, similar to clipping or a torn baffle - instantly recognizable. Yet amp-speaker measurements showed no sign of excessive distortion, with THD under 0.5% in the vocal range and above.

Based on forum input, the likely cause was poor woofer-panel integration, panel radiation pattern in an overdamped room, or auto-calibration EQ.

After extensive trial and error, the issue clearly centered on the crossover region, where I also noticed image instability. Various forum-suggested tweaks helped temporarily, but the first small change like minimal toe-in variation or room mode re-calibration could easily bring the problem back.

Since the speaker isn’t biampable, there was no practical way to use DSP to address crossover issues, even if careful measurement confirmed them - and there wasn’t anything else I could do. So I gave up and retired the speakers.

The replacements worked flawlessly - no harshness, no imaging issues, no super-sensitivity to placement or EQ.

Out of curiosity, I gave the old pair one last listen in a different, untreated room. The vocal harshness was even more apparent across multiple recordings.

What still puzzles me - an amateur in the science of human audio perception - is this: if the root cause was woofer-panel integration (possibly combined with panel radiation pattern and/or calibration EQ), why did it sound like clipping?

 

[Keith_W]

Well, the answer is ... take some measurements and post it Everything else is speculation.

 

[tengiz]

Well, the answer is ... take some measurements and post it Everything else is speculation.

And I did. On this forum.

 

[Curvature]

Martin Logan Ethos speakers

Few measurements here: https://hometheaterhifi.com/reviews...tinlogan-ethos-hybrid-electrostatic-speakers/

Not high resolution, but you could use those curves to compare to your own if you have a microphone. Could help clear up what's actually happening.

 

[Curvature]

And I did. On this forum.

Could you please post the link here?

 

[tengiz]

Could you please post the link here?

Debugging an imaging issue - Martin Logan Ethos

II’ve been troubleshooting an audio issue in my setup, and after fixing one problem (thanks to help from the forum), I’ve run into something else. When I listen to a particular vocal track, one specific note — F4, around 349 Hz — sounds wrong. The voice gets blurred or smeared, and the image...

audiosciencereview.com

Rough pop/rock vocals

I like how my setup sounds, and I’ve enjoyed it for years. However, there are occasional — and very annoying — wrinkles. Specifically, some vocals in pop and rock recordings don’t sound very pleasing. Classical and jazz vocals, on the other hand, sound spectacular. I believe this may have to do...

audiosciencereview.com

Rough pop/rock vocals

If old Abba vocals sounds displeasing to you, I would suggest you have frequency irregularities due to either speakers, room or a combination of both. If it is the speakers, it may not be easily fixed by EQ either. With regards to Kef, did you listen to the Reference 5 or The R5 (Yes, the...

audiosciencereview.com

Rough pop/rock vocals

yes, the new graph you posted is nowhere near my suggested scale though. 20-20,000hz 45-105dB ... "Should I laugh or should I cry" has heavily processed vocals, that doesn't sound all that great. But they're not harsh or very forward in the mix either. But it's always hard to know what a person...

audiosciencereview.com

Rough pop/rock vocals

yes, the new graph you posted is nowhere near my suggested scale though. 20-20,000hz 45-105dB ... "Should I laugh or should I cry" has heavily processed vocals, that doesn't sound all that great. But they're not harsh or very forward in the mix either. But it's always hard to know what a person...

audiosciencereview.com

Rough pop/rock vocals

Your tinnitus might be contributing as well here then. To me all these tracks or close to harshness but not becoming harsh on a balanced system. You would probably benefit from a low Q reduction of the 1-4khz area, if you have access to some manual EQ, try putting a 0.7Q -2dB EQ at 2.5 or 3khz...

audiosciencereview.com

 

[Keith_W]

Debugging an imaging issue - Martin Logan Ethos

II’ve been troubleshooting an audio issue in my setup, and after fixing one problem (thanks to help from the forum), I’ve run into something else. When I listen to a particular vocal track, one specific note — F4, around 349 Hz — sounds wrong. The voice gets blurred or smeared, and the image...

audiosciencereview.com

Rough pop/rock vocals

I like how my setup sounds, and I’ve enjoyed it for years. However, there are occasional — and very annoying — wrinkles. Specifically, some vocals in pop and rock recordings don’t sound very pleasing. Classical and jazz vocals, on the other hand, sound spectacular. I believe this may have to do...

audiosciencereview.com

Rough pop/rock vocals

If old Abba vocals sounds displeasing to you, I would suggest you have frequency irregularities due to either speakers, room or a combination of both. If it is the speakers, it may not be easily fixed by EQ either. With regards to Kef, did you listen to the Reference 5 or The R5 (Yes, the...

audiosciencereview.com

Rough pop/rock vocals

yes, the new graph you posted is nowhere near my suggested scale though. 20-20,000hz 45-105dB ... "Should I laugh or should I cry" has heavily processed vocals, that doesn't sound all that great. But they're not harsh or very forward in the mix either. But it's always hard to know what a person...

audiosciencereview.com

Rough pop/rock vocals

yes, the new graph you posted is nowhere near my suggested scale though. 20-20,000hz 45-105dB ... "Should I laugh or should I cry" has heavily processed vocals, that doesn't sound all that great. But they're not harsh or very forward in the mix either. But it's always hard to know what a person...

audiosciencereview.com

Rough pop/rock vocals

Your tinnitus might be contributing as well here then. To me all these tracks or close to harshness but not becoming harsh on a balanced system. You would probably benefit from a low Q reduction of the 1-4khz area, if you have access to some manual EQ, try putting a 0.7Q -2dB EQ at 2.5 or 3khz...

audiosciencereview.com

My apologies for missing those threads. I'll go read the threads and respond.

 

[tengiz]

My apologies for missing those threads. I'll go read the threads and respond.

C’mon, no need to apologize - with so much going on here, it’s hard to catch everything.

 

[Curvature]

C’mon, no need to apologize - with so much going on here, it’s hard to catch everything.

Take it as a collective apology We kind of piled in earlier in the thread.

 

[tengiz]

Take it as a collective apology We kind of piled in earlier in the thread.

Well, I stepped into it by violating the rule against suggesting that robots can have opinions

If it’s hangin’ time, I’ll bring the rope. If it’s a duel at dawn, I’ll polish my boots and write my will. Just do me one kindness: tell my horse I went down swingin’.