One Hacker’s Battle To Slightly Improve A Sadistic Sony Headset

A collage of three images. On top is the main PCB of the headphones, with a charger IC and the main MCU. On the bottom left, it shows ACOK and EN signals going to pullup resistors near the MCU. On bottom right, it shows the charger chip with its pinout overlaying it, highlighting the pads to be probed and later cut.

One thing you won’t read in reviews of pricey Sony WH-1000XM3 headphones – if you choose them, you’re going to find yourself in a one-sided abusive relationship. A button press or low-battery notification makes the headphones scream at you, ignoring the actual sound volume of what you’re currently listening to. Once they’re discharged, they suddenly emit loud noises, lecture you about how the battery is low, then shut down. Oh, and you can’t use them as they’re charging – if your voicecall lasts longer than expected, you might find yourself being shouted at and forced to fumble around with wires, silently pleading for call participants to wait while you change over to a different headset.

On-PCB footprint for the IC, with two pads carefully cut in half as mentioned in the article[MisterHW] decided to dig in and figure out how to work around at least some of the shortcomings. Naturally, the “no charge while using” limit looked like low-hanging fruit, and a hefty usability improvement too – plus, he suspected the charge cutoff to be masking noise issues already abundant on these headphones. Some painful disassembly later, he was inspecting the charger IC , the MP2625, responsible for power management. Its signals were connected to the MCU using via-in-pad – and some pads had to be cut in half to disconnect the vias.

However, the laborious pad cutting and subsequent careful soldering didn’t turn out to be fruitful. Even with external control of the output inhibition pins, the shutoff still continued – something affected the circuit, whether it was VBUS detection, some other unnoticed via-connected pins on the charger, or sensing on D+/D-. About to run out of life force for this mod, [MisterHW] added a Qi charging circuit, powering a TP4056 wired in parallel with the MP2625. The mod, dubbed HW-1000XM3, made these headphones all that more reliable and less annoying – charger and MCU none the wiser.

Now, all that’s needed to charge these headphones is to slap a magnetic charger coil onto them, and it doesn’t interfere with voicecalls as much as the screaming and forced shutoffs do. Hopefully, Sony eventually learns to test their headphones by having humans use them – it’s far from the only gripes with this lineup, after all. We also hope that the voice notifications will be conquered eventually – this summer, we’ve seen a hacker firmware mod a Bluetooth speaker to make the sounds more pleasant. If your headphones are based on a relatively popular module, repurposing them might be even easier than that!

48 thoughts on “One Hacker’s Battle To Slightly Improve A Sadistic Sony Headset

  1. I use my pair for many years now in the office and on trains. Never had any problem with them. The spoken “battery low” message is by no means shouting at you lol.

    Yes, not being able to use the headphones while charging is stupid, I agree.

    But to be fair battery runtime is excellent on those even after years of constant use. If it gets to something like 30% (still an hour or 2 of usage left) it starts blinking red. Windows and Android/iOS show the battery left percentage…

    Yeah you can forget to charge them, but you pretty much know when to charge after a week of using them.

      1. That doesn’t make sense. The headphones look to be double-insulated, so even if you had 120 VAC in the headphone case, they’d be fine. If the USB charger isolation failed, you’d be at greater risk from touching the metal shell of the USB connector.

        1. I suppose the main issue might be sweat, highly conductive, reducing the clearance and “creepage” from the electronics to the user. Headphones charging on a table are inherently safer for the user. Just my guess!

        2. Never underestimate the corner cases or the litigious nature of many people. “My spouse was using the headphones in the bathtub with the charger plugged in. Now they’re dead and it’s the manufacturers fault.”

      2. I’d like to see a source for that. Besides, you can still plug in a USB cable and thereby electrically connect the inner circuitry (which is not connected to any metal part of the enclosure anyway). This means that electrically, there is nothing implemented (at least on the 1000XM3) to do what you propose.

        I should probably also mention that Samsung Level On Pro (EO-PN920) can be charged while active, but their ANC quality is sub-par.

      3. Makes no sense… You could still wear them while they charge for the passive isolation they provide.
        Got my Jabra Bluetooth Evolve 75 which work while charging and even provide USB audio connection while charging the battery. Same for Poly/Plantronics set I have hanging around.
        This Logic would make any USB Headset a liability if you connect them to a USB charger/power source.
        Just a cheap way to hide design issue or to save a few cents :-(

      4. My guess is the reason is more performance than technical, as in efficient charging with a switching charger ic introduces noticeable noise into the audio if charged while in use (that is without introducing even more complication and cost into the design). So to prevent a bunch of consumers from complaining about the higher noise floor during charging they just remove that capability completely. Of course they could just have gone with a linear charger ic like this mod uses but then there are plenty of downsides to that as well (namely less efficient at higher charge currents/lower battery voltage and losing the quick temporary charge feature without overengineering the circuitry).

        Overall though, I can respect that the op hacked his pair to fit his needs, but it’s a bit of a hyperbole to make the xm3 sound like they were horribly designed just because it’s core use case didn’t match his needs (there are plenty of use cases that don’t require features he’s added as absolutely necessary and plenty that require ones that are lost because of said alterations). Different strokes for different folks.

        1. > charging with a switching charger IC introduces noticeable noise into the audio if charged while in use

          That is what needs to be explored.
          Sure, I could make a breakout with the original circuitry, but I’d like to get the MP2625 to work in-circuit to have a representative assessment (expecting primarily conducted EMI). The argument falls short a bit because the main signal processor core voltages are – by the looks of it – also provided via buck converters(1).

          Audible noise or auditory-range interference can originate from having multiple switch-mode converters running off the same rail with the same but not synchronized operating frequencies. MP2625B has an internal 1.6 MHz oscillator which cannot be synced, but that frequency is already so high that not even sub-harmonics would fall in the auditory range.
          Another potential source of perceptible noise can originate from burst-mode operation under light load. With the processor using less than 50 mA (1000 mAh battery, > 20 h operating time) and charge termination at 10% (5-15% as per MP2625 datasheet), we’re still looking at ~ 100 mA. I doubt there would be intermittent operation at this level.

          Finally, when taking a look at the USB-C connector PCB, it seems like there are unpopulated footprints for input filtering components. As 1000XM3 was the first pair of USB-C headphones in that series, it would be interesting to find out why they were omitted. Certainly less filtering would be needed for scenarios that are not susceptible.

          > but it’s a bit of a hyperbole to make the xm3 sound like they were horribly designed

          There’s the actual hardware, and then there is the user experience (which one could also call part of the design). It is my conviction that one needs to speak up against what I feel is an ever-increasing corporate hunger for user compliance. “Be happy with what you got or get lost”. That voice assistant certainly sounds to me more like a sales pitch that visitors at an exhibition might find flashy, but certainly not an everyday experience any person that is not tone-deaf might actively choose.

          If you take away nothing else from this journey, at least consider this: there have been better times for relationships between businesses and customers. Only in recent times has there been an increasing push to get more compliance out of users, which will ultimately reduce everyone’s freedoms and rights. Companies do not exist outside a legal framework, but where they have lobbyists, consumers can only be inspired by showing them alternative imaginations of what a product should look like and how it should function to serve its owner.


          1. Addendum:

            There are simply things which I find inexcusable:
            turning off during conference calls with no way to recover, no way to prevent shutdown by plugging in at the beginning of a call, and muting my audio to blare oh-so-important voice messages.
            As these are all software “features”, it would be possible to allow the owner to customize the behavior with full granularity. Even under Windows you can (de-)select sounds on an individual event type level – and that is an OS that’s become more and more known for manipulating users to put up with all kinds of things from accounts to putting up with ads.
            With how much time Sony apparently spent on integrating assistant support features I personally have no use for, it’s hard to dismiss the bottom line here: they wanted it to be exactly as limited and bossy as it is.

            Oh, and as far as EMC goes – 1000XM3 mics are analog, and all cables are routed from the right to the left can through an unshielded? bundle of cables. When taking some buses, one can hear a high-pitched noise that correlates with the diesel engine RPM. This is probably due to interference caused by piezo injectors. They also do funny things when you get near an induction stove… nothing is perfect – but there is absolutely some leeway for charging. As I wrote above, EO-PN920 do that, too.

          2. If i had to guess without knowing more about the architecture of the xm3’s the supply noise on the analog amplification stage could have a greater noticeable effect than on the dsp stage. The way things like supply ripple and hf emissions affect digital vs analog circuitry is different. But like I said without knowing more info I really cant say, it’s still interesting to think about the problem though.

            And I absolutely agree that everyone’s opinion is valid for their use/setup, but this works both ways and cant be used to discredit a differing opinion just because their situation doesn’t align with another person’s. We all have different needs and what is ok for one person may be unacceptable for another and that’s perfectly fine.

          3. I thought I remembered seeing the mics were of the mems variety from an ifixit teardown I read ages ago, though the output could be either analog or digital in that case.

          4. sjm4306
            > thought I remembered seeing the mics were of the mems variety

            See for yourself – while there may be MEMS mics around in earlier iterations, it’s only WH-1000XM5 that seems to fully make the switch. 1000XM4 uses analog mics particularly for ANC. It seems to me that initially, MEMS mics were used exclusively for the bluetooth hands-free profile mic synthesized from multiple microphones for directionality / ambient noise suppression. In the latter case it doesn’t matter how bad the sound is, as it’s already limited to 64kbps thanks to codec limitations.

            WH-1000XM3: (one of the best teardowns out there)

            As far as MEMS mic arrays go, it’s a popular topic on which papers are published and eval boards are made (say hi to Big Home Assistant)


        1. There are a few things that could be found to counter the assumption.

          – Other devices don’t do that (imagine your smartphone did – in reality, you can charge while doing phone calls).
          – Are they really pushing the charging rate limits in headphones? (1000 -12000 mAh cells and 0.5 or 1 A charging rates -> <= 1C)
          – 1000XM3 are superficially equivalent to other products that charge while in use.
          – All 1000X models I've seen have an NTC sensor on the battery connected to the charging IC to ensure safe charging temperatures.
          – For decades, Sony Energy Devices has been manufacturing LiIon battery cells, with the battery business only sold off to muRata in 2016. It's reasonable to assume that in-depth knowledge of battery technology and charging requirements still exists in other parts of the company.

          It will – for now – remain an unsolved mystery – until we can experince the sound performance during native charging.

    1. Sony isn’t the only one, I have a HyperX Cloud MIX headset and it shuts off bluetooth if I attach the charger to it. It does let me use the wired analog jack while charging, as it appears to be a completely passive system. Doesn’t scream at me when the battery is low, instead a led flashes about once every 5 seconds. Overall I’m very happy with them as they fit my budget and fit my head. ;-)

      1. My Skullcandy headphones disable the wired input as well while charging. It’s weird, as the input can operate in 2 modes: pass through where the device powers the drivers directly and amplified mode where the input goes through it’s internal amplifier. It doesn’t need energy for the pass though mode, but it can disable it.

    2. Even worse than a loud beep at every button press or low battery warning: a synthetic woman voice with a chinese accent screeching “volume mæx” or “bætöry lõw”

    3. Another possible reason the phones switch off when charging is that Sony didn’t want to pay for validating/fixing the EMC/RED emissions with the charging circuit coupling into the bluetooth (or simply with the USB cable changing the otherwise minuscule ground plane and thus the antenna radiation pattern)

    4. Long time owner of the MX1000 line, with USB C on the newer ones all it takes is like 15 minutes of charging to get over an hour of mid-volume listening out of them. It’s an inconvenience, but it’s not a deal breaker.

  2. The only reason I bought Sony noise canceling phones was for those long plane rides we used to do. But they disconnect and turn off after 20 minutes if not connected to something and that cannot be changed. Sony is constantly doing stupid stuff like this, ignoring a primary reason someone might choose their products.

    1. You use them minus the Bluetooth connection to anything to block out sound?

      What if I told you there was a product whose primary purpose actually is exactly this, even more effectively, for a much cheaper price, that never requires charging, never turns off after 20 minutes, and takes up a lot less space? Mind blown? Check this out!

        1. They have passive noise cancellation instead, which is effective at all frequencies, for all sources of noise (not just periodic noise, as ANC cannot handle random or even aperiodic noise), and does not have a battery life limitation. If you specifically want an over-the-ear form factor, a good pair of Optime IIIs will be hard to beat.

    2. I was using them on long flights connected by cable to the plane’s entertainment system. No automatic shut off.

      It might detect the audio signal and stay on, but maybe only the presence of the jack plug is detected and enough to keep them on. If that’s the case, a plug alone (without cable) might do.

      1. I have WH-1000XM4s and they detect the jack alone, no signal required. If I am switching from BT to wired for whatever reason, the BT shuts off as soon as I plug the cable into the headphones, long before I’ve gotten around to finding the other end in the dark and connecting it to my PC’s almost invisible black-on-black headphone connector.

        1. Just tested it: (WH-1000XM3, Bluetooth not connected)
          No plug in the 3.5mm jack: Switches off after 5 minutes.
          Just a 3.5mm plug (no cable, no signal): Stays on.

          So, just plug in any cable or cut off a plug of some crappy headset and use this. (One might even call the a ‘Hack’)

    3. I can’t check right now, but if I recall correctly you can disable the turning off or change the timeout via the gadgetbridge app (available via F-droid). I do remember that you can change a few other things that I don’t think you can configure in the official app: For example, I sometimes disable the touch interface if it is raining outside and I have my raincoat hood on since the water on the outside of the raincoat will trigger the touch interface in that case.

      You can also disable the spoken voice when you press the button.

  3. Do not buy Sony products!
    Sony is anti-repair, anti-standards, and anti-consumer. They constantly try to lock suckers customers into their proprietary trash.

    I stopped buying their products

    1. This is generally true, but those headphones are just top of the class stuff since their first version.

      Noise cancelling is a field in which absolutely and only the top companies shine. Sony, Bose, Apple. Sennheiser and a few others trying to catch up. China-NC is trash, listened to many cheap NC headphones.

      I googled the repairability of the XM3 before I bought them used. You can change the LiPo easily on them.

      Of course DON’T install any firmware update on them if you’re happy with how they perform. Just don’t. Sony is notorious for making things worse all the time :(

    2. +1

      I’d learned this with an earlier Sony product, forgot, and was reminded again while troubleshooting cooling fan issues in my VAIO laptop. Sony MAY allow visibility into cooling fan operation, but offered no way to deal with fan problems (small wonder why the CPU was at >95C – the fan wasn’t running because it got stuck on motor cogging.)

      (Soln: place a 3-way switch in the fan power circuit to toggle between the existing (automatic) control, or direct to 5V.)

      Sony does make good stuff, but often, you eventually get to a situation where you need to toggle a configuration switch to correct a problem, but Sony didn’t make it available.

    3. I totally agree. I will occasionally put up with Sony’s proprietary connectors and paternalistic ways if I’m given a piece of their kit, but I will NEVER give them any of my money, nor pay for used Sony gear.

  4. THIS VEHICLE HAS LIMITED ON-STAR CAPABILITIES. Yup GM not Sony. Always has had limited capabilities and now after 3 years has even less. But the phrase keeps going on and on and on for the life of the vehicle. And its LOUD. Much louder than my Sony Headset. Looked at disconnecting On-Star but the PCB is plugged into the radio and it appears that the radio will not run without it. Would have to re-engineer the On-Star PCB to keep the radio and loose everything else.

    I don’t sell my vehicles. I scrap them out. So there is no other user that’s going to need to know that On-Star is not being paid for. Nor will anyone need to know that it has air-bags and you should wear your seat belts when sitting in front of an air bag that can go off. But hey, everyone loves looking at big yellow stickers on their $60,000 trucks.

  5. Some Sony products are reasonably good. Others — such as the WH-1000XM3 — exhibit typical Japanese group decisions:

    1. Products are tested for specific use cases only.
    2. If issues are noted, they are either fixed or noted in the manual.
    3. If it’s noted in the manual, it’s not a bug but a feature.
    4. If it has a feature that is sufficiently annoying to a company employee who is actually eating his own dog food, it will be noted for a fix in a future product.
    5. No fixes will be made if doing so could possibly embarrass some higher-up, or rock the boat in any manner.

    The one good thing I can say for that flow is that it’s vastly better than the Chinese decision making process that skips every step noted above.

  6. The spoken low battery warning that interrupts audio playback and can’t be disabled is definitely fairly loud and very annoying, but calling it “shouting” is way too much drama. The fact that you can’t use them as wired USB headphones or use them wirelessly while charging is just bad, though, and this was rectified in the XM4, at least. Another really annoying thing about them is that you must pair them to an Android or iOS device to change settings; you’re only begrudgingly allowed to use them with a PC apparently. I think this is true of all ANC headphone brands.

    (you can still use the XM3 with a 3.5 mm wire while they’re charging, though ANC doesn’t work in that case and I assume the mic won’t work either. OP probably didn’t realize this because they were using Apple products)

    Still a better experience than I had with the Sennheiser PXC 550 though: the thing never charged properly, often taking a full day to charge or not charging at all. I returned them under warranty, but had them sent back to me after two weeks, clearly untouched, with the note that I must not be using the right charger (the headphones didn’t include one, and every charger I did use was a name brand one that managed to charge everything else I tried just fine). I’m certainly never buying a Sennheiser product that has a battery in it.

    1. >The fact that you can’t use them as wired USB headphones or use them wirelessly while charging is just bad, though, and this was rectified in the XM4, at least.

      Did they add this via firmware update? The help guide claims the opposite:

      WH-1000XM4: “When the headset is connected to an AC outlet or computer, all operations such as turning on the headset, registering or connecting to Bluetooth devices, and music playback cannot be performed. ”

      WH-1000XM5: “When the headset is connected to an AC outlet or computer, all operations such as turning on the headset, registering or connecting to Bluetooth devices, and music playback cannot be performed.”

      So that’s either a copy-paste mishap, or all iterations up to XM5 cannot be used when charging. I only have 1000MX2 and 1000XM3 to experiment with, perhaps an owner of the others can chime in.

  7. Sony is not the only one with excessively loud aural status indication. I have a pair of Bose NC 700 noise cancelling headphones. The noise cancellation works really well! However:
    – Turning the volume all the way down by accident gives a very loud “thunk” indication.
    – Turning them on or off gives a loud orchestra sound, even at low volume.
    – They are bluetooth, but don’t support newer codecs such as low latency.
    – They have a wired audio input, which doesn’t support use of the headset’s inbuilt microphone.
    – The microphone only works with the “calling” mode which reduces audio quality for the soeakers part. (To be fair, the noise cancelling on the microphone also works wonderfully except in excessively loud environments.)
    – They cannot be charged while in use.
    – They don’t support NFC pairing.
    – The “assistant” button is easy to accidentally activate but hard to cancel, and can’t be remapped except to choose vs google assistant or Alexa (or possibly Siri on an apple device).

    Unfortunately (kinda) the noise cancelling (which I bought them for) is so good that I put up with all the shorcomings (mostly) happily.

    However if anyone from Bose reads this: I will literally consult with you on the design process for free, test prototypes and buy the resulting headphones if you’re willing to go to bat on these issues.

    Discord: DustLight #6309

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.