Stax SR-L700 Mk2 Electrostatic Earspeakers Measurements, Analyses, and Reviews

Discussion in 'Headphone Measurements' started by Vtory, Oct 23, 2021.

  1. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    IMG_0306.jpg

    If I recall correctly, Stax released the SR-L700 Mark 2 (hereafter called L700.2) a couple of years ago. I kinda completely ignored any electrostatic product news that time -- as my estat game seemed over in 2015 and I thought I was done with Stax.

    However, I should have not ignored the L-series lambda lineup (at least its first revision, mark 2) which I originally thought was just a renaming or gimmick.

    After being thoroughly filled with wonders and wows for four straight days by the loaner pair, I’m almost about to pull the trigger on L700.2 for myself.

    These things are good.. Very, very good in almost every single way I can think of. They scratch nearly all Stax itches I had previously.

    Before dropping measurements, let me first state a few broad level highlights of subjective evaluation..
    • Metal york and improved joint structure eliminated many of the buggy plastic noises. Also clamping force noticeably increased (note prior lambdas’ clamping was infamously weak), which resulted in stable wearing experience AND solid seals.
    • Drivers angled much nicely against the ears. Ear holes got bigger so that my ears don’t touch the inside any longer.
    • Detachable cables greatly increase the peace of mind as I don’t have to worry about cable shorts or awkward situations too much.
    • MLER stators seem to bring semi-009 technicalities to L700.2 without losing the lambda presentation. I could hear technical and spatial cues much better and easier with L700.2 than any other non-estats I had at the moment.
    • L700.2 has fairly agreeable tonal balance. Leagues better than any rectangular types (i.e., lambdas) or circular types (i.e., 007/009) Stax ever manufactured. More on this shortly. The only exception may be the X9000 which I have zero idea about.
    All the measurements and evaluations done with the following configurations
    • Chain: Soekris 2541 (black filter; usb in; 2x oversampled to 88.2/96khz) - Stax SRM-353X - Stax L700 Mk2
    • Other associated gears: KGSSHV (loaner), Nectar Hive
    • Measurement: MiniDSP EARS with SBAF compensation curve v1.0
    • Measurements calibrated at 95db SPL @ 300hz
     
    Last edited: Oct 24, 2021
  2. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    Measurements

    stax_sr_l700mk2_lr_freq.jpg

    Frequency response is pretty decent. Indeed, surprisingly linear and smooth for Stax. Specific comments below.
    • A big departure from prior lambdas. For the reference, pre-L lambdas measure more or less similar to each other let alone non-FR technicalities.
    • Also any prior lambda can’t measure this good in bass below 100hz “on the EARS” without pushing both cups toward the inside, which I didn’t do this time around.
    • The gradual 4db peak at 1.3khz is often seen in many estat products. Maybe it’s associated with physical behaviors of thin films with certain surface area. But I digress.
    • 2khz upward remains reasonably neutral. Heck, is this really a Stax product? One caveat is the 8.7khz peak is more audible to my ears in certain tracks than the graph suggests. But a simple notch filter did magic. L700.2 responded to parametric equalizers well, too.

    stax_sr_l700mk2_dist_l.jpg stax_sr_l700mk2_dist_r.jpg

    The harmonics don’t look bad in any frequency range. While the graphs above are only for 95db spl fundamental, I could see 90 and 100db spl measurements remained the same in pattern with shifting up (90db) and down (100db).


    stax_sr_l700mk2_csd_l2.jpg
    stax_sr_l700mk2_csd_r2.jpg
    (EDIT: Color scheme and Y/Z axis ranges changed for better readability)

    CSD is mostly clean considering my room's environmental noise is around 50-60 db spl. Sustained energy shown at 1.2-1.3khz looks interesting. I’m curious if they’re associated with the sweet and romanticized vocal presentation I heard. ES1a had this pattern and also presented romantic coloration.
     
    Last edited: Oct 24, 2021
  3. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    Here are comparative plots against other estats I measured before.

    stax_sr_l700mk2_vs_sr_207.jpg

    I'd make an assumption that SR-207 serves as a representative fr for all prior lambdas. Regardless of diaphragm thinness or material difference, previous dummy-head measurements for lambdas from various generations were more similar than different, which convinces me hypothesize pre-L lambdas would measure similar (to each other) on EARS as well. Note that flat couplers are a totally different story and not very compatible with mine.

    Contrast in bass indicates how good L700.2 is in seals. You can also easily see healthy differences in entire trebles.


    stax_sr_l700mk2_vs_hive.jpg

    Next up is vs Hive. Hive is more linear and neutral up to 1khz. Lowest notes have better presence and heft in Hive, too. Hive also had less upper-mid coloration. L700.2 is smoother in highs. All those were very consistent with my subjective listening.


    stax_sr_l700mk2_vs_es1a.jpg

    Comparison with ES-1a shows fairly similar response up to 2khz. While 4.2khz peak on ES1a followed by a large “scoop” wasn’t as bothersome as you may guess from the graph, I definitely do prefer how highs are articulated and properly etched in L700.2.


    stax_sr_l700mk2_vs_esp95x.jpg

    Except for leagues of difference in technicalities and bass performance (caveat: I ran 95x with Koss energizer atm), 95x has more pleasing and darker tones. Maybe arguably better timbre, too. Interesting to see a very similar uprising pattern up to 1.3khz. I don’t think I heard it much in 95x when I had them, which can be explained by gradual downward tonality on upper midrange and highs.


    stax_sr_l700mk2_vs_hd650k.jpg

    Finally, above is the comparative plot with HD650 (Kiss-mod) as a reference point.

    That’s pretty much it for now. I’ll report back about my subjective evaluation later once I collect enough thoughts. Meanwhile, questions or thoughts are all welcome.
     
    Last edited: Oct 24, 2021
  4. nishan99

    nishan99 Friend

    Pyrate
    Joined:
    Dec 1, 2019
    Likes Received:
    1,617
    Trophy Points:
    93
    Location:
    Layla
    Almost all of the Focals, Audezes, Abyss and some of the Hifimans have some ~1khz-1.5khz emphasis. idk if it's an indication of an abused diaphragm or is it intentional to enhance something. My guess it's the former.
     
  5. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    This is great work @Vtory

    Interesting is the mirrored distortion profiles for 2nd & 4th order harmonics near 5khz.

    Overall even with the peak it's still <1%. Stax definitely knows what they are doing.

    Are you ever going to get a chance to review the Audeze CRBN? I would trust you with measurements and subjective impressions
     
  6. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    Good catch. There must be something going on either driver surface or space around ears. On the other hand, FWIW typically that frequency range is tunable via different pads, different damping, etc. But I speculate all such come with some kind of non-trivial tradeoffs. So, they might decide to leave it as is. Just guess tho.


    I really want to do so. Currently there aren't many CRBNs in the wild. Pandemic and new manufacturing process seem to give audeze (and crbn buyers) hard time. I'm also hoping audeze would kindly offer a loaner tour to sbaf.. but given low interest on estats (compared to dynamic hps and orthos -- audeze themselves proved it by presenting crbn and lcd5 together in canjam this year. I heard way more people showed interest on lcd5 than crbn.), I'm quite unsure how likely it would happen.

    From a little different angle, I'm debating to get crbn for myself next year or so when they pop up in the used market.. 300g of weight is insane and very attractive. But a couple of things are making me unconfident: (1) pads equipped via adhesive tape; and (2) cable non-detachable. Plus, their MTO (Make to Order) nature.. Well, we will see.


    It should be taken with enough caution. I believe measurement artifacts play an important role there. Let me explain potential caveats.

    harmonics_analysis_01.jpg

    The plot above present fundamental harmonics (y axis is now dbu rather than %). Note the fundamental shown above is uncompensated.

    When creating distortion plots, REW decomposes original signal into fundamental, harmonics, and noise. Fundamental will be no problem as it's much louder than the rest. But I'm under the impression that REW can be inaccurate to identify harmonics when they're at much lower level than noise floor. In this respect, I doubt if H3 and H4 are really accurate around 4.5khz.

    Keeping that in mind, let's change y axis to percentage again. This implies all the data points on harmonics will be divided by the fundamental (or alternatively subtracted in log scale such as db).

    harmonics_analysis_03.jpg

    Therefore, mirror patterns are almost always unavoidable. It's also effected by the noise level at the time of measurement.

    Nonetheless, my measurement experience says how and in which strength patterns are presented vary across different products. There must be some information indicating product quality. But to make stronger assertions we may need much better environment and/or system (like Tyll once built a dedicated noise-isolating station for measurement).
     
  7. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    That's too bad about low interest. I want to say it's simply because most people don't have the required amplification. Audeze coming out with CRBN is the most excited I've been about a non-STAX estat. It would have been #1 tryout at canjam for me

    SBAF tour would be very low number unless it was packaged with an amp. Maybe something like the iFi ESL could work.

    Speaking of Tyll I wonder what happened to his special edition KGSSHV with variable bias made to review any possible estat
     
  8. rhythmdevils

    rhythmdevils MOT: rhythmdevils audio

    Pyrate
    Joined:
    Apr 15, 2020
    Likes Received:
    12,238
    Trophy Points:
    113
    Location:
    Bay Area, CA
    Home Page:
    It's easy to re-attach them using cutouts from sheets of double sided 3M tape. I'd even send you a sheet if you ever need it.
     
  9. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    I updated CSD plots by changing display options a bit. Hope these might be more readable and comprehensive.

    Also I pulled out other estat csds with the same option. Comparison was more interesting than I thought.

    Potential caveats
    • I don't think CSD is solely affected by driver movements. It's also highly impacted by trapped energy around ears. Front/back damping also plays a meaningful role.
    • Different mechanisms (regarding influx and outflux of sound waves) may introduce different biases.
    • CSD/waterfall via EARS is experimental at best. Disqualified for robust evidence for assertions. Let's not take the results too seriously.
    That being said, below were what happened.

    stax_sr_l700mk2_csd_r2.jpg

    stax_sr207_csd_r.jpg

    nectar_hive_csd_r.jpg

    eslab_es1a_csd_r.jpg

    koss_esp95x_csd_r.jpg

    PS. For the records, each estat above driven by the following amplifiers if I recall correctly.
    • L700.2 : SRM-353x (as described in the OP)
    • 207: SRM-1/mk2
    • Hive, ES-1a: Kgss Carbon
    • 95x: Koss E90/x
     
    • Like Like x 2
    • Epic Epic x 2
    • List
    Last edited: Oct 24, 2021
  10. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    That will be lovely. I briefly experienced it when I rolled the pads for LCD2C. And it was much easier and simpler than rolling pads for legacy Stax .. removing glue was super painful. haha.

    I may apply some velcro for even easier pads rolling. I wasn't a big fan of consuming 3m tape every time. Maybe it's just me tho.
     
  11. rhythmdevils

    rhythmdevils MOT: rhythmdevils audio

    Pyrate
    Joined:
    Apr 15, 2020
    Likes Received:
    12,238
    Trophy Points:
    113
    Location:
    Bay Area, CA
    Home Page:
    That will change the sound dramatically unfortunately because it will break the seal between pad and baffle. Audeze has a felt ring around the earpad to create a very specific damped break in the seal between pad and baffle, increasing this with velcro would likely lead to a huge drop in bass response. Same with the CRBN, though I don't know if they come with the same felt ring but I'm guessing they do.
     
    Last edited: Oct 24, 2021
  12. rhythmdevils

    rhythmdevils MOT: rhythmdevils audio

    Pyrate
    Joined:
    Apr 15, 2020
    Likes Received:
    12,238
    Trophy Points:
    113
    Location:
    Bay Area, CA
    Home Page:
    Mind throwing my RP-18 in there for fun? And to show ortho superiority? :p
     
  13. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    Of course not. But I found orthos using a little more absorptive material (than estats) so it may not be an appropriate "apple to apple". Enjoy them with enough caution. Not saying to RD because he already knows it well .. but warning to general thread readers.

    nad_rp18_rd_csd_r.jpg
    audeze_lcd_x_csd_r.jpg
     
  14. rhythmdevils

    rhythmdevils MOT: rhythmdevils audio

    Pyrate
    Joined:
    Apr 15, 2020
    Likes Received:
    12,238
    Trophy Points:
    113
    Location:
    Bay Area, CA
    Home Page:
    Yeah I am always kidding about my ortho superiority complex. I mean it’s what I believe but others have different valid opinions. Electrostatic headphones do their thing well, I enjoy them and I agree the CSD’s probably don’t paint the whole picture . Also those RP18’s were tuned so horribly. Treble much?. :oops:

    Interesting to compare though.
     
  15. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    Eyeballing estat csd graphs I posted hours ago, there seem largely two clusters. If we group both staxes together, then the rest might be categorized to the other group. The former group characterized as fast initial decay, low level sustain, and very late release. The latter looks more linear in z-axis (time).

    As per Sajeev, Hive loosens diaphragm tension to get good bass performance and heft. Based on the similar pattern shown in ES-1a, I am suspecting ES might take similar approach. I didn't observe any sticking driver issue during my time with ES-1a though. Stax is known to take the opposite, which may be consistent with the decay character observed in the plots.

    And extending this speculation, I'm curious if Koss loosens the tension compared to Stax but not so much as Nectar/ES.

    Such observed pattern's association with perception of speed, heft, or artificiality is still debatable but very interesting to me.
     

Share This Page