I can think of several good reasons to use 1/8" TRS headphones (though as I point out in a lower comment, specifically for smartphones, space is at an extreme premium and I think that the majority of people probably don’t want to spend the space on an integrated headphones jack; it’d be better to use a small external adapter there):
But for the general case, not on smartphones, places where I have the space to stick a 1/8" TRS port, I am not very enthusiastic about using USB as an audio port.
1/8" TRS is a well-established standard. I mean, pretty much every device can handle it. USB for audio is in a number of places, but not even close to the level of 1/8" TRS.
1/8" TRS has been around forever. It’s electrically-compatible with 1/4" TRS, which has been around even longer.
The original 1⁄4 inch (6.35 mm) version descends from as early as 1877 in Boston when the first telephone switchboard was installed[9] or 1878, when an early switchboard was used for the first commercial manual telephone exchange[10][11] in New Haven created by George W. Coy.
USB is a young pup and already, physical USB-A ports are being phased out in favor of USB-C ports. I very much doubt that USB-C is going to be around ~150 years down the road the way that TRS has been. I can use a pair of headphones from the 1970s just fine with the latest device, and I can use an elderly radio from the 1970s with a new pair of headphones.
USB is a lot more complicated than 1/8" TRS. It’s got sleep states, trees, power consumption negotiation. That’s all room for things to break in interesting ways. I have, for example, a USB hygrometer/thermometer that sporadically triggers kernel errors on my computer when plugged in. I have a mouse that, for some reason, when plugged into a USB hub, uses a lower polling rate if plugged in when the system boots up (albeit not if unplugged and replugged). I have a USB audio DAC/ADC that decided to cut out the other day, for God knows what reason, until it was restarted. My last computer’s motherboard had a USB controller that supported a more-limited-than-required-by-protocol-USB tree size and had random devices not work if a sufficient number of devices were plugged in. None of this exists with 1/8" TRS.
Security. Same idea. I’ve got enough attack vectors into my devices as-is. People have definitely attacked bugs in USB stacks before; IIRC, that’s historically been part of how they attacked DRM on some consoles. 1/8" TRS is a dumb protocol, but that makes it safe. Same issue with USB for charging, though at least there you can get a “power-only” cable. You can’t have an “audio-only” cable.
USB sticks the DAC on the headphones. Why? Headphones don’t last that long; they’re disposable items. Put the non-disposable bits where they won’t die. A DAC can last pretty much forever. I have gone through many headphones over the years. I have never had a sound card or on-motherboard DAC or dedicated DAC die. The closest I came was once ripping the 1/8" TRS output on a DAC loose, which I could solder back into place. I have two USB-to-1/8"-TRS DAC/ADCs sitting on the shelf by my desk. They’ll probably be perfectly good twenty years from now.
Sampling rate issues. Can’t come up on TRS, because the DAC/ADC is on the device side. One of my USB DACs (this intended for professional audio) only supports a fixed sampling rate, the one at which it does internal processing; that makes sense, as a pro doesn’t want to have some device introducing resampling into their audio chain. Another, consumer one, can’t support a sampling rate as high as the professional one; it relies on the computer to figure out and do resampling if resampling has to happen above that rate. You can have software that doesn’t work with a given pair of USB headphones because it doesn’t like the headphone DAC’s supported sampling rates; I’ve seen that before. If I have a pair of 1/8" TRS headphones, they work everywhere. It doesn’t matter whether whether they’re new or old or intended for the professional market or consumer market. Plug 'em in, they work.
I have one wired audio-emitting device – a pair of elderly Logitech USB speakers, not headphones – that has an integrated DAC. For some reason, the engineers who did that appear to have decided to make the volume control on that linear in electrical power rather than in perceptual loudness, which means that the vast majority of the volume scale does very little and there’s a tiny range that has a large impact. I don’t want to deal with that kind of craziness on some cheap pair of headphones.
Latency. 1/8" TRS devices normally – unless you’re intentionally building something into the system – have zero latency, because the DAC on the device is directly electrically driving the membrane on the speaker. Every time one sticks higher-level protocols in, it’s an opportunity for some bright-eyed, bushy-tailed engineer to start cramming more shit into the pipeline that adds latency. TVs are a great example of this – they used to have no latency, and then someone figured out that they could show ads and do other processing on the feed and that that’d be easier if they had a buffer of some video frames, and so they started inducing latency, unlike a computer monitor. Now you have “gaming modes” on TVs that try to mitigate the problem which had never originally been an actual issue with dumb TVs.
There’s an entirely-separate world of audio software and hardware for professionals who want to do real-time audio processing (on Linux, JACK; I have a USB ADC and some audio cards that permit direct passthrough of input audio to output) to try to avoid all the points in the pipeline that various consumer audio devices and software have inserted latency.
That doesn’t matter for some uses, like an MP3 player. It’s not the end of the world for a phone call. But it’s really obnoxious for some uses. With 1/8" TRS, I have no latency. With USB, I have God-knows-what latency.
Durability. 1/8" TRS is more-rugged than USB-C. I’ve damaged both before by pulling on cables at right angles, but micro-USB, mini-USB, and USB-C are more-fragile. That being said, I will give USB this: the damage tends to be worse on the cable side, as the plug is flimsy and will tend to give out before the socket on the device, whereas with TRS you can more-readily mess up the device. I would be open to the idea that having a standard magnetic breakaway connector would be more sane than either 1/8" TRS or any existing USB standard.
There are only three decent reasons that I can see to use USB headphones for the general case (like, not the extreme-space-constraint situation that smartphones see):
It provides power. Some people want active noise cancellation on their headsets. If you want to do ANC, you’re gonna need power one way or another. 1/8" TRS doesn’t have a standard for that (with XLR, for condenser mics, there’s a 48 volt phantom power convention that was added, but TRS doesn’t have it). AFAICT, devices that do this with a 1/8" TRS interface either rely on a second USB wire for power or use batteries.
When initially plugging in a 1/8" TRS plug, one shorts connectors and it can make a loud noise on the speaker membrane. Not an issue with USB, because the speaker membrane isn’t in that pipeline.
1/8" TRS doesn’t specify a single impedance everywhere. You can get very-high-impedance headphones that a DAC with limited output power can’t drive at a reasonable volume, even with the volume all the way up. That isn’t usually an issue for most people, but USB avoids the issue.
EDIT: Apparently I lied on the phantom power argument for using USB; according to WP, there are 1/8" TRS devices that do take phantom power (or something comparable; sounds like it’s not, strictly-speaking, “phantom power”):
Plug-in-power (PiP) is the low-current 3–5 V supply provided at the microphone jack of some consumer equipment, such as portable recorders and computer sound cards. It is also defined in IEC 61938.[16] It is unlike phantom power since it is an unbalanced interface with a low voltage (around +5 volts) connected to the signal conductor with return through the sleeve; the DC power is in common with the audio signal from the microphone. A capacitor is used to block the DC from subsequent audio frequency circuits. It is often used for powering electret microphones, which will not function without power. It is suitable only for powering microphones specifically designed for use with this type of power supply. Damage may result if these microphones are connected to true (48 V) phantom power through a 3.5 mm to XLR adapter that connects the XLR shield to the 3.5 mm sleeve.[17] Plug-in-power is covered by Japanese standard CP-1203A:2007.[18]
Also, regarding the power argument – USB power can be a source of noise leaking into what you hear.
USB power can be incredibly, mind-bogglingly dirty. I couldn’t believe it the first time I watched some video of some guy with an oscilloscope showing it. I guess it makes sense – I mean, keeps USB controllers and hub prices cheap – but there’s all kinds of electrical devices that have to deal with it. Anyway, point is, it’s the responsibility of the USB device containing the DAC to have a power supply that cleans that up sufficiently before feeding the DAC. It turns out that…they don’t necessarily do that. I have one USB-powered (not using a USB audio interface, or switching away from my computer’s USB bus wouldn’t be an option) mixer with 1/4" TRS output where using the USB power bus off my computer for power resulted in perceptible audio artifacts, humming and such.
This appears to be something of a not-uncommon problem, as I see various references to it online for other devices:
Some of you guys may be aware of my posts and other’s in the Topping D10 review thread. It seems that this DAC, like many audio devices that get their [power from the USB Bus, suffer from some noise coming from the USB port itself.
From my own experience, plugging the DAC into a Raspberry Pi 3B (+5v PSU and Ethernet connected) dropped the noise considerably compared to any port on the PC.
And if I can hear it, then I guarantee that there are USB audio devices that are inserting all kinds of garbage into the signal going out the output that are maybe less-egregious.
I wound up avoiding the problem with my mixer (well, at least to the point where I couldn’t hear it) by sticking the mixer onto an isolated USB charger, not on my PC’s USB tree. Now, yes, you can make a fancy power supply that avoids that, and it’s fair to say that the guys that engineered the mixer should have used a better power supply if they were gonna use USB power. But if you’ve got some guys engineering headphones and are under pressure to try to make the things as cheap as possible, because headphones are a disposable item, not to mention as light as possible because they’re gonna sit on your head, I’m not sure I’d bet on how much expense and weight they’re gonna put into the power supply feeding the DAC.
I haven’t tried quantifying how the power supplies on various USB DACs perform, though I would suggest that in a world where people are using USB audio rather than 1/8" TRS, given that you have headphone reviewers that cover things like frequency response, it’d be interesting to have a device that intentionally screws with the USB input power voltage and then have an oscilloscope or something attached to the leads coming off the magnet driving the speaker’s membrane and see just exactly how much glop from USB power is leaking through to the membrane at various dick-with-the-voltage patterns.
EDIT: Oh, and I forgot one other point. Cable length. 1/8" TRS cares very little about cable length. If you want a 200 foot cable, sure, go for it. USB, especially newer and faster forms, is pretty restrictive on cable length. I decided, a few years back, to move my PC to the other side of the room to reduce noise at my chair and had fun discovering that a number of current PC cable standards are not incredibly friendly to long runs. USB couldn’t communicate without repeaters or an optical bridge, DisplayPort had visible artifacts and occasionally saw the screen go black and need to re-handshake, etc.
The maximum recommended cable length for USB 2.0, is five meters, or around 16 feet. That’s actually the longest maximum length of any standard, passive USB cable specification, with USB 1.0 cables restricted to just three meters.
You may find some USB 2.0 cables that run longer than 5m, but they’ll need to be made with a thick wire gauge to ward off signal loss and interference at anything much past that maximum. Alternatively, you can run longer USB 2.0 connections by bridging two USB 2.0 cables with a powered USB hub.
How long can a USB 3.0/3.1 Cable be?
USB 3.0 and 3.1 Gen 1 cables don’t have an official maximum length, but their recommended maximum is around 2-3 meters in length or around nine feet. Like USB 2.0, you can extend this with a powered USB hub, potentially linking a few together to extend your run, but there are far better solutions for longer USB 3.0 cable runs.
The maximum length for USB 3.0 and 3.1 was maintained into the final USB 3.0 Type-A specification, known as USB 3.2 Gen 2. It was also mirrored in the higher-performing USB 3.2 Gen 2x2 Type-C connections, which also had a nine-foot maximum cable length.
However, USB4 cables, which leverage the USB Type-C connector but can transmit data at up to 40 Gbps, only have a maximum recommended cable length of 0.8 meters or around 2.6 feet. That goes for both the existing 40 Gbps cables and the USB4 80Gbps spec cables which will see greater use in the coming years.
These cables can transmit data at a much greater bandwidth than their USB 3.2 Gen 2x2, or earlier counterparts, which means the cables need to be made of a higher quality to ensure the data is fully transmitted. Unfortunately, that also means signal attenuation is a bigger problem, hence the shorter maximum cable runs for USB4 cables.
You forgot that 3.5mm is a stupid connector that makes you pass charged metal pieces over the connector to plug it in. You can’t power an anc chip or a dsp with it because it can’t do power delivery. That’s how you get headphones sounding different based on whether they are turned on
I can think of several good reasons to use 1/8" TRS headphones (though as I point out in a lower comment, specifically for smartphones, space is at an extreme premium and I think that the majority of people probably don’t want to spend the space on an integrated headphones jack; it’d be better to use a small external adapter there):
But for the general case, not on smartphones, places where I have the space to stick a 1/8" TRS port, I am not very enthusiastic about using USB as an audio port.
1/8" TRS is a well-established standard. I mean, pretty much every device can handle it. USB for audio is in a number of places, but not even close to the level of 1/8" TRS.
1/8" TRS has been around forever. It’s electrically-compatible with 1/4" TRS, which has been around even longer.
https://en.wikipedia.org/wiki/Phone_connector_(audio)
USB is a young pup and already, physical USB-A ports are being phased out in favor of USB-C ports. I very much doubt that USB-C is going to be around ~150 years down the road the way that TRS has been. I can use a pair of headphones from the 1970s just fine with the latest device, and I can use an elderly radio from the 1970s with a new pair of headphones.
USB is a lot more complicated than 1/8" TRS. It’s got sleep states, trees, power consumption negotiation. That’s all room for things to break in interesting ways. I have, for example, a USB hygrometer/thermometer that sporadically triggers kernel errors on my computer when plugged in. I have a mouse that, for some reason, when plugged into a USB hub, uses a lower polling rate if plugged in when the system boots up (albeit not if unplugged and replugged). I have a USB audio DAC/ADC that decided to cut out the other day, for God knows what reason, until it was restarted. My last computer’s motherboard had a USB controller that supported a more-limited-than-required-by-protocol-USB tree size and had random devices not work if a sufficient number of devices were plugged in. None of this exists with 1/8" TRS.
Security. Same idea. I’ve got enough attack vectors into my devices as-is. People have definitely attacked bugs in USB stacks before; IIRC, that’s historically been part of how they attacked DRM on some consoles. 1/8" TRS is a dumb protocol, but that makes it safe. Same issue with USB for charging, though at least there you can get a “power-only” cable. You can’t have an “audio-only” cable.
USB sticks the DAC on the headphones. Why? Headphones don’t last that long; they’re disposable items. Put the non-disposable bits where they won’t die. A DAC can last pretty much forever. I have gone through many headphones over the years. I have never had a sound card or on-motherboard DAC or dedicated DAC die. The closest I came was once ripping the 1/8" TRS output on a DAC loose, which I could solder back into place. I have two USB-to-1/8"-TRS DAC/ADCs sitting on the shelf by my desk. They’ll probably be perfectly good twenty years from now.
Sampling rate issues. Can’t come up on TRS, because the DAC/ADC is on the device side. One of my USB DACs (this intended for professional audio) only supports a fixed sampling rate, the one at which it does internal processing; that makes sense, as a pro doesn’t want to have some device introducing resampling into their audio chain. Another, consumer one, can’t support a sampling rate as high as the professional one; it relies on the computer to figure out and do resampling if resampling has to happen above that rate. You can have software that doesn’t work with a given pair of USB headphones because it doesn’t like the headphone DAC’s supported sampling rates; I’ve seen that before. If I have a pair of 1/8" TRS headphones, they work everywhere. It doesn’t matter whether whether they’re new or old or intended for the professional market or consumer market. Plug 'em in, they work.
I have one wired audio-emitting device – a pair of elderly Logitech USB speakers, not headphones – that has an integrated DAC. For some reason, the engineers who did that appear to have decided to make the volume control on that linear in electrical power rather than in perceptual loudness, which means that the vast majority of the volume scale does very little and there’s a tiny range that has a large impact. I don’t want to deal with that kind of craziness on some cheap pair of headphones.
Latency. 1/8" TRS devices normally – unless you’re intentionally building something into the system – have zero latency, because the DAC on the device is directly electrically driving the membrane on the speaker. Every time one sticks higher-level protocols in, it’s an opportunity for some bright-eyed, bushy-tailed engineer to start cramming more shit into the pipeline that adds latency. TVs are a great example of this – they used to have no latency, and then someone figured out that they could show ads and do other processing on the feed and that that’d be easier if they had a buffer of some video frames, and so they started inducing latency, unlike a computer monitor. Now you have “gaming modes” on TVs that try to mitigate the problem which had never originally been an actual issue with dumb TVs.
There’s an entirely-separate world of audio software and hardware for professionals who want to do real-time audio processing (on Linux, JACK; I have a USB ADC and some audio cards that permit direct passthrough of input audio to output) to try to avoid all the points in the pipeline that various consumer audio devices and software have inserted latency.
That doesn’t matter for some uses, like an MP3 player. It’s not the end of the world for a phone call. But it’s really obnoxious for some uses. With 1/8" TRS, I have no latency. With USB, I have God-knows-what latency.
Durability. 1/8" TRS is more-rugged than USB-C. I’ve damaged both before by pulling on cables at right angles, but micro-USB, mini-USB, and USB-C are more-fragile. That being said, I will give USB this: the damage tends to be worse on the cable side, as the plug is flimsy and will tend to give out before the socket on the device, whereas with TRS you can more-readily mess up the device. I would be open to the idea that having a standard magnetic breakaway connector would be more sane than either 1/8" TRS or any existing USB standard.
There are only three decent reasons that I can see to use USB headphones for the general case (like, not the extreme-space-constraint situation that smartphones see):
It provides power. Some people want active noise cancellation on their headsets. If you want to do ANC, you’re gonna need power one way or another. 1/8" TRS doesn’t have a standard for that (with XLR, for condenser mics, there’s a 48 volt phantom power convention that was added, but TRS doesn’t have it). AFAICT, devices that do this with a 1/8" TRS interface either rely on a second USB wire for power or use batteries.
When initially plugging in a 1/8" TRS plug, one shorts connectors and it can make a loud noise on the speaker membrane. Not an issue with USB, because the speaker membrane isn’t in that pipeline.
1/8" TRS doesn’t specify a single impedance everywhere. You can get very-high-impedance headphones that a DAC with limited output power can’t drive at a reasonable volume, even with the volume all the way up. That isn’t usually an issue for most people, but USB avoids the issue.
EDIT: Apparently I lied on the phantom power argument for using USB; according to WP, there are 1/8" TRS devices that do take phantom power (or something comparable; sounds like it’s not, strictly-speaking, “phantom power”):
https://en.wikipedia.org/wiki/Phantom_power
Also, regarding the power argument – USB power can be a source of noise leaking into what you hear.
USB power can be incredibly, mind-bogglingly dirty. I couldn’t believe it the first time I watched some video of some guy with an oscilloscope showing it. I guess it makes sense – I mean, keeps USB controllers and hub prices cheap – but there’s all kinds of electrical devices that have to deal with it. Anyway, point is, it’s the responsibility of the USB device containing the DAC to have a power supply that cleans that up sufficiently before feeding the DAC. It turns out that…they don’t necessarily do that. I have one USB-powered (not using a USB audio interface, or switching away from my computer’s USB bus wouldn’t be an option) mixer with 1/4" TRS output where using the USB power bus off my computer for power resulted in perceptible audio artifacts, humming and such.
This appears to be something of a not-uncommon problem, as I see various references to it online for other devices:
https://www.audiosciencereview.com/forum/index.php?threads/cleaning-usb-for-bus-powered-audio-devices-discuss.5899/
And if I can hear it, then I guarantee that there are USB audio devices that are inserting all kinds of garbage into the signal going out the output that are maybe less-egregious.
I wound up avoiding the problem with my mixer (well, at least to the point where I couldn’t hear it) by sticking the mixer onto an isolated USB charger, not on my PC’s USB tree. Now, yes, you can make a fancy power supply that avoids that, and it’s fair to say that the guys that engineered the mixer should have used a better power supply if they were gonna use USB power. But if you’ve got some guys engineering headphones and are under pressure to try to make the things as cheap as possible, because headphones are a disposable item, not to mention as light as possible because they’re gonna sit on your head, I’m not sure I’d bet on how much expense and weight they’re gonna put into the power supply feeding the DAC.
I haven’t tried quantifying how the power supplies on various USB DACs perform, though I would suggest that in a world where people are using USB audio rather than 1/8" TRS, given that you have headphone reviewers that cover things like frequency response, it’d be interesting to have a device that intentionally screws with the USB input power voltage and then have an oscilloscope or something attached to the leads coming off the magnet driving the speaker’s membrane and see just exactly how much glop from USB power is leaking through to the membrane at various dick-with-the-voltage patterns.
EDIT: Oh, and I forgot one other point. Cable length. 1/8" TRS cares very little about cable length. If you want a 200 foot cable, sure, go for it. USB, especially newer and faster forms, is pretty restrictive on cable length. I decided, a few years back, to move my PC to the other side of the room to reduce noise at my chair and had fun discovering that a number of current PC cable standards are not incredibly friendly to long runs. USB couldn’t communicate without repeaters or an optical bridge, DisplayPort had visible artifacts and occasionally saw the screen go black and need to re-handshake, etc.
https://www.cablematters.com/Blog/USB-C/usb-cable-max-length
Alas I have but one upvote to give.
You forgot that 3.5mm is a stupid connector that makes you pass charged metal pieces over the connector to plug it in. You can’t power an anc chip or a dsp with it because it can’t do power delivery. That’s how you get headphones sounding different based on whether they are turned on
I have two sections in my above comment talking about power delivery over 1/8" TRS.