Sonoma Acoustics Model One Electrostatic Headphone and Amplifier System Page 2


Sonoma Model One HPEL Driver
In the exploded view above, the top part is the very fine protective mesh between your ear and the drive and isn't really a part of the driver assembly itself but rather mounted on the baffle plate in front of the driver.

Under it and again at the bottom are the two rigid polycarbonate halves of the casette that holds the diaphragm in position. The two thin parts just inside each cassette half are die-cut adhesive film that firmly attaches the diaphragm and stainless steel screen to the cassette and ensures mechanical separation of the cells.

The green chevrons are small pieces of circuit board to provide termination for the drive and bias signals on the diaphragm and ground for the mesh screen. The three central parts are the HPEL driver itself. The upper part is a 0.1mm thick stainless steel mesh; the lower part is the 15μm thick metalized laminate diaphragm; the central part is a 0.7mm thick open cell Formex™ insulating spacer that has adhesives on each side. These central three parts are manufactured with a semi-automated, roll-fed process. Warwick expects this process to become fully automated as volumes increase.

As you can see, this is a single-ended driver design; all electrostatic forces occur between the diaphragm film and the stainless steel mesh on its rear side (away from the ear)—quite unusual for an electrostatic driver that usually has a stator both in front and behind the diaphragm. Sonoma claims the advantage here is that there is very little (just the acoustically transparent protective screen) between the diaphragm and your ears.

The down-side of a single-ended drive system is that it's prone to higher levels of distortion, primarily 2nd and even order harmonic byproducts. The good news is that 2nd harmonic products are considered euphonic, which may make headphone a little tube-like in character.

Another unusual characteristic is that the design essentially creates eight separate 'drum-skin' diaphragms from one piece of film due to the mechanical separation of the hexagonal cells. This is essentially a parallel driven, multi-driver headphone! A great deal of finite element modeling went into optimizing this design. Here's how Sonoma describes it:

Thanks to a proprietary Finite-Element Analysis software package, WAT is able to fine tune the characteristics of the 'drum-skins' such that they have different resonant frequencies. Each cell is acoustically independent, but driven in parallel. As a result, the sound from each cell combines in acoustic space, but the independent resonances average out, avoiding any large resonant peak in the audio band (as can happen with a single driver area).

I was really curious about 2nd order distortion and whether Warwick embraced a single-ended for its euphony. They answered:

Although the single-sided topology increases levels of even-order harmonic distortion, primarily second, this can be compensated for through the use of digital signal processing. Indeed, the combination of FEA (finite element analysis) with sophisticated DSP allowed WATL to counteract this particular aspect, and WATL feels the many positives of the HPEL design far out weigh this one negative (which is mitigated via DSP).

No, there was not a conscious decision to go with a single-sided solution because of any desire for 'euphonic second-harmonic distortion'. The design goal throughout has been to minimize all forms of distortion to maximize the dynamic range and resolution of the transducer (and systems using HPELs like Sonoma's). Extensive analysis of the distortion performance of the HPEL has been undertaken using the COMSOL simulation tool. As a result, distortion has been reduced, and we have found a good correlation between simulation, measurement and listening tests.

One of the jobs the DSP is busy doing is the calculations to compensate for driver non-linearities. I decided to have a look at the distortion products with the FFT function on my Audio Precision tester at 500Hz at various volume settings to see how well it behaved. (Note: scale to left is a relative scale where 90dBspl=-40dBra. All measurements taken through the high level analog inputs; input voltage 1Vrms for 100dB.)






If you look at the level difference between the fundamental and 2nd harmonic (at 500Hz and 1kHz) you can see that each time the level increases by 5dB, the difference between the two peaks decreases by about 5dB. At 90dBspl, which is a solid level, the 2nd harmonic is 45dB below the fundamental tone; by the time you get to 105dBspl, quite quite loud, the two are 30dB apart. This seems to me like a very well controlled behaviour...we're seeing the DSP do its job.

I'll note that at 90dB, nasty sounding 3rd order distortion products are -70dB down, which I calculate to be 0.05%. (Correct me if I'm wrong.) That's a dandy number. At 105dB these numbers increase to -52dB and 0.22%, not bad. You can also see that the fundamental and first three harmonics have good proportion with each one ever smaller than the last—this is thought to be a positive indicator for sound quality.

The thing you don't normally see is all those high order distortion products popping up between 2kHz and 10kHz. I reckon those are by-products of all the processing. However, also note that these products remain under -70dB down until the very end at 110dB...where things go haywire.

At 110dB the input voltage would be about 3Vrms, which is outside spec for the M1AMP; we're likely seeing the analog input section clipping somewhere...maybe at the ADC inputs. But, it could also be uncontrollable non-linearities in the driver. Now 110dB is pretty loud, so I think they've actually done a pretty good job of balancing things out until the system begins to reach its limits.

But in listening using the digital input—which would bypass any chance of overdriving the ADC or DSP at fairly loud levels—louder than I would normally but not at what I would consider unsafe levels for a burst of fun, I could often hear sub-bass distortion, especially with EDM drops. I took a look at low frequency distortion spectra.

50Hz at 90dBspl

50Hz at 100dBspl

100Hz at 90dBspl

100Hz at 100dBspl

Here you can see that the 50Hz signal is fine at 90dB but is significantly breaking up at 100dB; the input voltage here was about 1Vrms—well within spec. However, at 100Hz the system handles the change in volume well. I think we're just hitting the excursion limit of the transducer at very low frequencies. In email exchanges I was told:

"...the system is dimensioned to meet, but not exceed, EN-50332 limits for maximum volume, etc."

In the M1AMP manual in the troubleshooting section for the problem "Distortion is present on some source material" it says:

Check the volume control setting. It is possible to overdrive the electrostatic panels with excessively bass-heavy source material at high volume settings. When listening to these types of source material it may be necessary to reduce the volume setting slightly.

I think the HPEL driver in the Model One has been brilliantly pushed to the very edge of its performance envelope. Bottom line: this is not a system for people who want a system that's capable of playing at elevated levels cleanly, or for EDM-like music or bassheads in general—you especially will be running into these limits.

Modified, Pseudo-Diffuse Field Target Response
The other thing the DSP is busily crunch numbers about is equalizing the headphones to match a 'modified, pseudo-diffuse field response'. Basically, these headphones are EQed to something akin to the Harman Target Response. Here's the frequency response measurements.


The raw plot gray lines are quite close to the Harman target; in the compensated plot above the raw responses, you can see the plot is nearly flat to 1.5kHz and then tilts downwards in a nice even fall. Other than being a bit rough, the response is free from large peaks or vallies until about 8kHz where it would be unheard of to not have a somewhat ragged plot. Again, this seems to be a very wisely tuned and well controlled headphone.

So, how do they sound?

Sonoma Acoustics

Tyll Hertsens's picture
Thanks, Suuup! got a little thump back into estats...long as you don't crank it.
Suuup's picture

Another interesting comparison is This is harman-compensated.

Tyll Hertsens's picture
...damn...that's pretty good.
JimL's picture

Actually, I believe that -70 dB is 0.03% rather than 0.05%. Both are very good.

JimL's picture

Hm, -30dB at 105 dB is 3% 2nd harmonic distortion.

-40 dB = 1%
-50 dB = 0.3%
-60 dB = 0.1%.
-70 dB = 0.03%
-80 dB = 0.01%
and so on - actually, each 10 dB represents Sq rt of 10 decrease = 0.3162x

MRC01's picture

Correct. In case anyone wants to compute it for himself, the formula is: DB volts = 20*log(ratio). The log is base 10, so it's easy to see every factor of 10 is 20 dB. For power it's DB power = 10*log(ratio), so every factor of 10 is 10 dB.

The reason the formula is different for V versus P is for consistency. This way, an X dB difference in voltage creates an X difference in power. If the formulas were the same, an X dB difference in voltage would make a 2X dB difference in power. That's because doubling the voltage doubles the current (since V=IR), which quadruples the power (since P=VI).

GearMe's picture

Your description of the sound reminds me how I felt a long time ago with the Quad ESL-57s compared to the other speakers I owned. For certain types of acoustic music they just had "it"...sounding more real to my ears than any other speakers I've owned. That said, their dynamic range sucked enough that I eventually got rid of them. In hindsight, a dumb decision. ;-)

ab_ba's picture

"Take your average $4000 headphone..." Love it.

Magoo's picture

WOW! $5K for a HP with limited Bass and grainy HF? The DAC has the infamous glare producing ESS Sabre?

Not for me at least...Thanks for the review!

Tyll Hertsens's picture
Seriously, I wouldn't say it had any glare or grain, just a bit hazy. Limited bass level though, for sure.
detlev24's picture

"[...] if you don't throw at least another $4000 in amp and DAC at it you're probably wasting money with underperforming very expensive gear."

That sounds like a commercial for unreasonable expensive gear. Just one example: Take the Focal Utopia, add 50% of its price (~ $2000) and you have it attached to a state-of-the-art amp/DAC, the RME ADI-2 Pro. It will absolutely not be underperforming on this gear and you could even save some bucks by choosing a different combo; if you are not in need of the built-in ADC and EQ.

Of course, you can spend as much (more) money as you want if you subjectively prefer, e.g., the euphonic distortion of tube amps (which could be added to any neutral gear by effects, btw.) or like to have a "design object" to also please your visual stimuli. Again, that is an option but certainly not a necessity to fully power your headphones!

elmura's picture

Whilst $5k ain't cheap, it's within reach of high-end systems, and certainly cheaper than LCD-4 / Utopia etc systems. The fact that the system is tuned to work optimally together using DSP EQ means buyer doesn't need to worry about system synergy, trial and error of different amps / dacs etc. And, I'd bet they will be sending DSP updates in future. I'd like the ability to tune the DSP EQ to suit different hearing (or taste).

I'm interested in this product's future.

MRC01's picture

It seems more like a science project than a product. I love Tyll's open minded optimism & enthusiasm, and as an engineer I respect what they did, but why they chose this kind of driver mystifies me. It seems like they picked an inherently poor driver and did some real brain sweat and engineering to make it sound less bad, ending up with a very expensive product with mediocre performance. I can't help but think that if they applied that same engineering effort to a conventional or ortho driver it would be truly astounding.

Magoo's picture


When will you report on these latest Grado HP's? $2700 and yet built like $30 Hp's?


solrage's picture

I had high hopes for these. $5k for an all-in-one system synergistically designed to work well with each other isn't really outrageous given the price of high-end headphones/amps/dacs out there these days. The frequency response is definitely on point, but those distortion figures make this a no-go for me. On days where I don't plan to listen for very long I like listening >90dB, and even on days where I keep my average level <85dB, a lot of music I listen to--especially orchestral usic--has dynamic peaks over 100dB, so being able to play clean loudly is a prerequisite. Oh well, guess I'll just have to see how the LCD-4/Utopia fiasco turns out... or maybe just wait until some company finally puts out a great headphone that ticks every box. Until then, I'll remain relatively happy with my HD800s, HE-6, and LCD-X combo.

heymamahey1990's picture

People need to make use of what they have. - Steven C Wyer

harry56's picture