Differences Between the SB and non-SB Stax Transformers

Discussion in 'Portable and Other Gear Measurements' started by takato14, Feb 26, 2023.

  1. takato14

    takato14 God of Ruin

    Pyrate
    Joined:
    Sep 26, 2015
    Likes Received:
    654
    Trophy Points:
    93
    Location:
    Michigan, USA
    Due to the electrical problems that tend to arise when trying to direct-drive an electrostatic headphone -- an open circuit with almost no actual current draw -- some people elect to forego using a dedicated Stax amplifier at all in favor of a transformer-based solution with a traditional speaker amp in front of it.

    In fact, many normal headphone amps are capable of driving a Stax transformer -- not just speaker amps -- and this kind of setup can outperform all but the very best discrete energizer solutions if you have an amp with enough stability and balls to handle a highly reactive 16 ohm load. Depending on who you ask, people who use this type of setup will claim that most if not all of Stax's own "SRM" lineup of components fail to get anywhere close to a proper transformer setup, and some even claim that they rival the quality of 3rd party solutions like the Kevin Gilmore offerings and HeadAmp Blue Hawaii SE.

    However, when building a transformer setup, these people are also adamant on one key detail: avoid the "SB" lineup transformers. Instead of blindly accepting this as fact, I had to find out for myself.

    I got my hands on both an SRD-7mk1 and an SRD-7SB, two units that share the exact same featureset and look largely the same superficially but have nearly 16 years of time between them, and the best normal bias headphone available: the original SR-Lambda.

    [​IMG]

    The big difference between these units is that the all-black "SB" unit is self-biasing -- that is to say, it rectifies the AC input from the amplifier input signal to generate a DC bias charge, as opposed to generating it from the AC wall outlet like the original SRD-7 does.

    I, guess the idea behind this unit was that it would save you one outlet on the wall...? If we're being serious, this sort of design makes perfect sense for a portable or transportable adapter... but not so much for a home unit that's roughly the same size and weight of the original SRD-7. Generating DC from a clean 60Hz wall input is going to be much more consistent and simple than a signal that fluctuates a lot, and doesn't really provide any tangible benefit at home where the unit just sits next to a wall outlet anyways. Unless, of course, there is no performance loss as a result.

    Huh? A sticker on the SRD-7mk1? U-uh, I have no idea what you're talking about. Moving on.

    For the following measurements, I left the same channel of the SR-Lambda in the exact same position on the measurement rig, and drove the SRDs using my TEAC UD-501 and Schiit Mjolnir v1, an extremely powerful and speedy SS amp that manages to get stax headphones to earsplitting levels before even using up half of its volume control.

    [​IMG]

    Immediately, we see that the sensitivity on the SB is much... wait. Higher?? The f**k?

    Yep, it turns out the SRD-7SB manages about 2.6dB better sensitivity at the same position on the volume wheel, an almost 25% improvement in perceived loudness over the SRD-7 according to the measurements. Due to the inevitable variations in the bias charge voltage, I was expecting the opposite.

    The way I see it, the only possible explanation for the increased sensitivity is that the SRD-7SB uses different, higher-ratio transformers than the original SRD-7 -- which is in line with what I have read online. However, a stubborn old f**k I am, and I aim to be certain. Off come the top covers for both and...

    [​IMG]

    Yep. These are different parts entirely.

    "Tak those trafos look exactly the same." No, no they do not! They are obviously from the same manufacturer yes, but if we look closely It's only the same core and galvanized steel frame. The part we're looking at -- the coils in the center -- are entirely different.

    [​IMG]

    See? The physical dimensions don't quite line up. The SB transformers have larger coils.

    (I had to pull the calipers off the trafos to get a good picture of the reading since the housing was in the way, so the gaps you see are just me being bad at lining up the photo. lol)

    [​IMG]

    Digging further into the measurements, I noticed something. Due to the way the bias is built from the signal input, you have to drive the transformer for a short period before the diaphragms will reach full charge. You can see here that it takes about 3 full length (~20 second) sine sweeps for the FR to stop increasing in sensitivity. I didn't notice anywhere near this amount of variability listening to actual music, but I also didn't spend much time with it... for reasons I'll get into in just a moment.

    [​IMG]

    After adjusting the output level from the amp manually, so that both units were outputting exactly the same SPL, we can see clearly that the SRD-7SB is slightly bandwidth-limited by comparison, losing a small amount of bass extension and a much more significant amount of high treble.

    [​IMG]

    At the same SPL, with the SRD-7SB fully charged up, the THD in the midband actually swings lower than the SRD-7... while bass distortion more than doubles. Yikes. The distortion behaviour on the SRD-7SB is also somewhat erratic, with random fluctuations between different orders. SRD-7mk1 has the usual "flat line of 2nd order" for the majority of the response, like we look for from good headphones.

    [​IMG]

    Looking at raw dB numbers instead of percentage, we can clearly see that the THD on the SRD-7MK1 is "mirroring" the FR of the headphones, with the upper midrange peak corresponding with a large dip in harmonic distortion. The SRD-7SB on the other hand does the opposite, with THD actually rising as you approach that peak. Subjectively, this will emphasize the feature and make it sound more prominent, but by how much remains to be seen.

    [​IMG]

    No major differences in spectrogram. The SRD-7SB looks very slightly cleaner, but not by an amount more than I'm used to seeing with variance in the room noise level. Additionally, with pure dipole headphones like the SR-Lambda and some planars, spectro can be somewhat unreliable because it's so easy for reflected sound waves to make it back into the ear cavity after bouncing around the room during the measurement. The only reliable way to counteract this would be to build an ancechoic chamber like Tyll Hertsens did for his setup -- something I am not currently able to do within my 90sq. foot living space.

    [​IMG]

    The impulse response suffers a bit both in overall shape and adherence to the min phase estimate, with the SRD-7SB being the worse performer of the two.

    Subjectively, the sound on the SB is both hazier, thinner, and less impactful sounding. Immediate, noticeable drop in quality. Vocals stop "popping" and lack air, while the bass becomes weak and limp. The legendary lambda headstage collapses in on itself, and overall it sounds much more "dead" -- almost like running a low impedance headphone off high output impedance. Simultaneously, it sounds a little brighter in the upper mids. Oof.

    This should not come as a surprise. Since the SRD-7SB measures with more sensitivity, then the transformers it uses must have a higher wind ratio than the original SRD-7. As you increase the wind ratio, the voltage on the output coil increases for the same input voltage -- but at the cost of both transient response and frequency response linearity. The extra power wasted on rectifying the AC to DC for the bias signal is likely why they chose to make this compromise... or they were simply unable to acquire/afford the original part they used in the SRD-7mk1.

    Unfortunately, I only have one true normal bias headphone I can test. I would expect these issues to get *much* worse with something less sensitive, but as far as I know most normal bias Stax headphones are around the same sensitivity, and it would be unfair (and potentially unsafe) to try using either to drive something it wasn't designed to when you're working with such high voltages. If only I had found an SR-5NB at some point...

    Despite being much worse than the SRD-7, the SRD-7SB is still better than most SRM-type energizers. It has many of the same problems as they do subjectively, but isn't nearly as dead sounding. If you have no other option, it will certainly do the job -- just not to any degree of excellence. Tons of performance is left on the table, putting it lightly. Do not take the relatively small differences on the measurements to mean I am exaggerating -- the fact that I'm able to measure any difference at all with a rig as simple as mine should be an immediate red flag.

    ~T.
     
    • Epic Epic x 10
    • Like Like x 4
    • List
    Last edited: Feb 26, 2023
  2. peef

    peef Friend

    Pyrate
    Joined:
    Oct 5, 2015
    Likes Received:
    311
    Trophy Points:
    63
    Location:
    Montreal
    This is fantastic. Thanks for sharing your measurements and analysis.

    I have a suspicion that the differences that you are measuring and hearing are not necessarily due to the transformers so much as to the bias supply itself. Both the bias supply voltage and the supply impedance can impact the drivers' sensitivity, so it's worth measuring.

    Pulling up a couple of schematics suggests that Stax did a few revisions, and that at least a few of them would have used a 2.2M bias resistor rather than 5M. Some models also place a bypass cap right after the ballast resistor, which shift the drivers from constant charge operation to constant voltage operation. The difference is audible.

    35187-599801a4ccf7eb3cfdeb3b6712968db6.jpg

    There's a Japanese site where the author traces the SRD7 and finds it doesn't match the schematic.

    With that in mind, it might make sense to measure the transformers directly to check the ratio. The one I measured was 1:25+25, but with a slight asymmetry on the two waves (which would result in even order distortion).

    Edit: there are also some models that shipped with thermistors in series with the input, which hugely impact distortion. Not Stax's finest moment... https://www.linearz.com/stuff/transforming-the-stax-srd-7/
     
    • Like Like x 2
    • Agreed, ditto, +1 Agreed, ditto, +1 x 1
    • List
    Last edited: Feb 28, 2023
  3. Bowmoreman

    Bowmoreman Facebook Friend

    Contributor
    Joined:
    Jun 28, 2023
    Likes Received:
    186
    Trophy Points:
    43
    Location:
    Temple, NH
    Apologies for restart, but seemed the best possible place after much thread searching. I’ve JUST resurrected my old (as in bought them USED in 1979 whilst in University) Stax SRD-6/SR-5 headphone setup. The SRD-6 is the NON SB version (it has the AC cord into its in chassis power supply).

    I cleaned em up, and connected them to my MASSIVE Classe DR-9 Amp to test em out and see if they sounded as good as I remember “back in the day” (when they were driven by my Onkyo Receiver. They did! They work just fine. Treated the leather ear cups and they’re “good as new” (well, everything has a patina of 45+ years at this point TBH.)

    But, there is NO WAY I can use them with the Classe, since it’s job in life is to throw massive current at my Apogees, and those dinky AF input wires (and speaker out binding clip thingies) can handle anywhere near the power/current the speakers need). Literally, the ONLY thing I COULD do would be to buy some very long-ass speaker wire, run it from the Amp at one end of the 20+ foot room over to the SRD-6 near my listening position. And then manually disconnect the Apogees every time I wanted to listen - an obvious PITA that I’d like to avoid (though it’d be FREE for all intents and purposes)

    So, I am thinking a dedicated Amp for them, and just run a set of interconnects across the room to it.

    In listening (again) a bit, I can confirm the lack of bass below around 60Hz, and the top octave is a bit soft; but that’s ok by me - the mids are glorious in all their ‘stat-ness…

    Which got me to thinking… could I use a dedicated HP amp (maybe something seriously TUBEY - grin - to really revel in a Fat/luxurious AF midrange experience for folk, Jazz, small ensemble classical, etc. And also tripping at night to prog/space music)…

    I know I could snag a decent regular speaker amp to drive them (the Stax Manual says 8-30 watts) - that would be easy-peasy (The Schiit Aegir is on the consideration list if I were to go that route), but… I was wondering if I could snag a dedicated HP amp and wire em up that way? I know there’s impedance deltas to ponder, but might a Schiit Lyr pull it off?

    Any collective thoughts on this here? I read in some of the threads that some particularly good HP amps can drive the Stax SRD-5/6/7 non SB’s well despite their being spec’s for 16 or 32 ohms, etc?

    Anyone got anything other than - just try the freaking thing(s)? (I do have an ask in with the Schitt team… but there’s a much broader and deeper set of expertise here).

    TIA SBAF’ers
     

Share This Page