USB 3.0* Radio Frequency Interference on 2.4 GHz Devices (2012)
intel.com
intel.com
It's hard to believe that there is a standard and devices are widely deployed that mess so so much with their environment.
[0]: https://support.logi.com/hc/en-us/articles/360023414273-Wire...
Apart from that it’s just amazing, better (not as flat) keys than the magic make it perfect to type on, the usb on the side makes adding a wired mouse easy (and keeps the dongle close for wireless). I ordered several when I heard they were discontinued but still in stock at my old company just in case.
The answer, of course, is that you no longer have as many wires on your desk and in your bag. If you need very high reliability and fidelity, like if you’re a professional gamer or something, or you know you’re in noisy or otherwise troublesome environments, by all means use wired peripherals. I also use a wired keyboard and mouse, but it’s only because my favorite keyboard and mouse aren’t wireless and I haven’t found great wireless alternatives.
i love my bt keyboard and the freedom it gives me in the flat.
Logitech's wireless is spotty on a lot of products and their G hub application is awful but I love my Logitech headset.
When I was searching for solutions to my intermittent problems, I didn't believe that an extension would solve it. It just felt like "blowing into the NES cartridge" to me. But I kept seeing the same advice, so I gave it a shot and whaddyaknow? Zero issues afterward.
I'm certainly going to be looking at the problem through a new lens now that I'm aware of these wider issues.
I used to run HA in a VM via qemu/libvirt and it would always fail because when the device reconnects its on a different port and the USB pass through doesn’t work anymore.
Now I’m running the HA VM on proxmox which deals with disconnects reconnects better for pass through devices: it’s able to reconnect it to the vm without my intervention.
Looking at the logs, this happens between a couple times a week to 10+ times a week.
I tried an extension lead (which had no impact, I think this solves zigbee radio issues but not USB issues), using a powered USB hub, using a USB 2 port, a USB 3 port, three different machines... no dice. It really is a bug in the device itself. And Dresden electronics has piss poor support (still waiting for any answer from their email support six months later). The only avenue for “support” is via GitHub issues where other users answer but not actual Dresden electronics employees.
But same experience with Dresden/deCONZ. It's flaky.
Right now I have split networks while I lazily migrate devices away from deCONZ. So far every glitch I've encountered has been on the deCONZ network.
Which is a shame, because I really like the software UI. ZHA doesn't expose as much (or not as simply). But it works better, and that counts for more.
This is entirely a problem of our own making.
We give cellular providers hundreds of MHz of exclusive spectrum access, and then deprecate and auction off the old analog TV channels to give them hundreds more, but we expect everybody else to shove all of their traffic into a 100 MHz block at 2.4 GHz.
It's not that this standard is doing anything particularly nasty to the environment, it's just that it happens to be messing with the one tiny head-of-a-pin that we've decided everything needs to live on.
It's a computer controlled radio for crying out loud, give us 400 MHz and let the computer figure out how to hop around.
2.4GHz may be crowded, but it's nowhere near as bad as mobile where the demand for bandwidth has been doubling yearly for two decades.
There is utility in the general public being able to use more unlicensed spectrum at home and in the office.
We can argue about what the right breakdown might be, but I'll start by asserting that ~3 GHz in total between Verizon, T-Mobile, and AT&T, vs. only 250 MHz in total for the entirety of local, short-range wireless communication is absurd.
On Wi-Fi, everybody just shouts at each other. On mobile, providers buy huge swaths of spectrum, partially as a monopolistic strategy to make life harder for competitors, and partially because they're still using cell allocation strategies from the '80s. They maintain exclusive rights to blocks that they are not using in a region, because the towers two cells away are using them, and it's just easier to use a fixed checkerboard allocation.
Both Wi-Fi and 3GPP standards can and should be improved to make better use of temporarily unused spectrum.
A good start might be to prevent carriers from having exclusive rights to any spectrum. At least one layer of the cellular protocols should be standardized across carriers allowing towers and phones to dynamically request and then relinquish spectrum on an as-needed basis.
Today, if 500 people in a room use T-Mobile phones, but Verizon owns all the spectrum, then nobody gets to use anything. This is stupid. Verizon should have access to a fraction of spectrum proportional to their users in an area. More users, more spectrum, and vice versa.
The story now is "$80 billion in license fees" and "$20 billion in hardware" and that seems the wrong way around -- when most folks play poker the stakes are supposed go up, not down.
It's where there is the most money, not value.
That would require a fairly radical departure from the infrastructure of the existing cellular networks, wouldn't it? Right now, each provider has a monopoly on a portion of the spectrum within a defined geographical area, and provides the base station and backhaul infrastructure to support their network.
It's not like "Verizon owns all the spectrum in a room"; it's that Verizon has better base station coverage of that room than T-Mobile does.
The providers compete on, amongst other things, coverage and network performance. Sharing spectrum would effectively require mutualization of base station infrastructure. You would effectively have a single monopoly with the networks operating as virtual operators. It's very far from clear that would be a good outcome, to me, at least.
I think it can be done at the legal layer - just require roaming agreements between providers. They can already roam across borders so there's no technical barrier.
That’s the spin.
Yes, the FCC is the steward. But it’s also a bureaucratic, underfunded, somewhat backwards agency.
SIGINT is pretty cool, and FCC has some cool things and people that interact with them, I’m sure. But the outward face of FCC and how they appear to operate seems to have hardly changed in past years. I respect that typically, but it’s like an old bowling alley kind of respect; it’s fun, but it’s old and dirty, and the nachos aren’t bad but they aren’t good. The Lysol sprayed in those shoes is for psychological comfort. You don’t need to know the real reason.
All licenseholders should add up their transmitted joules of energy at the end of the year and pay the FCC. It will be in everyone's best interests to make most efficient use of the spectrum because then they can get their information to its destination with fewer joules.
Devices like home wifi routers should be able to buy upfront a license for X number of joules/year, and the cost of that is included in the purchase price of your router/laptop.
If market rate were justified for everything then we wouldn't even have space reserved for HAMs.
> There's only one industry with enough demand and revenue to justify that purchase price, which is mobile.
There's plenty of industries with enough demand. There's just no way they can compete with the money that mobile offers. That's a damn shame given that money shouldn't be the only, nor even the main, driver.
I will also add that 60ghz works great, but it will not penetrate walls. I have found it penetrates my doors and thus makes a great replacement for 2.4ghz/5ghz mesh networking in my apartment, but not much else.
This is actually incredibly annoying, especially in a residential setting where the 2.4GHz band is incredibly noisy (5GHz doesn't penetrate as well so it ends up being less noisy).
We have that, it's called 5GHz.
And also 6GHz now, that allocation is 1200Mhz wide!
In real life one of the drivers behind having so many wireless devices in the first place is avoiding the effort and cost of laying cables everywhere, 5GHz often doesn't solve that problem well.
For now I'm using powerline devices to connect rooms with their own set of problems, nothing beats having a well thought through wiring in the house though.
Most people I know don't have cables going to different rooms, there's a router where the signal comes in and wifi from there. Often there's not even the choice which room or part of the room that is.
Unless your building is relatively new and someone thought about setting up the cables, it can be very expensive to do so later.
To lay a cat6 cable and have it snake over the ceiling in a cable gutter? The cable I grant you isn't dirt cheap but it's by far the most expensive part of that arrangement.
In fact the lack of range works so well that I can use almost the entire spectrum for various networks and not feel guilty
Of course, the cat6 cable I used subsequently got recalled, and so the manufacturer had to pay for a contractor to rerun it. They said that it was the type of job they wouldn't have even quoted for any price originally because it was so gnarly.
I have 2.4Ghz and 5Ghz from two APs on either end of the house. I turned off support for pre-n speeds on the 2.4Ghz to hopefully save some bandwidth on the beacon frames. I have Ethernet over Power to my garage, where I have a third AP for our inlaw unit.
The Ethernet over Power seems to be pretty good, but I had to find the right brand of equipment for it to not be flaky. WiFi still sucks, but my desktop and TV are hardwired, and it's good enough for mobile devices. I can go for about a week without losing my work VPN connection, through wifi through Ethernet over Power to PDSL.
The only reason the things aren't outright banned is the frequencies tend to be sub 100mhz so it only has a significant impact on HF. Still, nasty things and many have been found over the legal noise limits.
If you can I recommend point to point 2.4ghz wifi links. Since wifi is regulated by ERP they have a high gain antenna that usually punches through things better than an omni AP. I've done them through multiple walls from 300-400ft away.
Before I ran cat6, I indeed used WDS wifi. It wasn't reliable, though, and the particular hardware I had suffered from a bottleneck somewhere that seemed to limit throughout to 6Mbps or so. I could have spent another $500 on less crappy equipment and maybe made it work, but moving to wires everywhere I could was highly effective.
This is the using a hackrf and I'm sending a file vs. not sending a file locally (i.e. to roughly max out the throughput) using powerline ethernet
The trouble is that protocols like ZigBee and ZWave and Bluetooth and ANT+ are stuck in the 2.4 GHz band and practically you cannot turn off your 2.4 WiFi access points and be happy.
Thus I have to go to the woods to pair my fitness band with the heart rate sensor because at my house who knows what is going on with the hue light that is on the wrong side of my monitor from the hue hub or the smoke detector that posts the battery status to SmartThings, etc.
I don't think this is a bandwidth issue. USB is a wired standard. Using shielding is not that hard, but doing it inexpensively, while maintaining cable flexibility and low attenuation at high frequencies is very hard.
I wonder how do manufacturers of those devices manage to not get caught by the fcc in us and equivalent agencies in EU. Perhaps they sell directly to customers overseas?
Seeing as there is a wide range of quality for USB hubs, part of me wonders if this is just poorly designed ones worsening potential issues
The only thing that has worked for me is to plug the Logitech receivers into a USB 2.0 only hub.
eg the badly designed PWM radiates RF crud from the sharp edges.
It also interferes with GPS and Iridium. USB3 is a broadband jammer.
Previous thread on the subject: https://news.ycombinator.com/item?id=24707479
I’ve been wondering whether the clock spreading in the USB standard could be manipulated to notch out certain frequencies, like in noise shaping for ADCs.
https://github.com/osqzss/gps-sdr-sim
(Seriously: Don't do this unless you're taking extreme care not to radiate outside your lab.)
There's no way a manufacturer could be expected to EMI test a product in every conceivable way a customer might wish to use it. For many products there's simply too many combinations of use-cases or features. So the rules generally require to test in a typical customer use or by following the instructions for use which come with the product.
As soon as you start using cables which didn't come with the product, it's on you to ensure that the emissions of the new cable and the product don't combine to cause problems.
Some cables (not just USB cables) are utter crap for interference.
If it does require a particular cable, then the cable should not be removable.
Shonky manufactures wouldn't be able to gain advantage by cutting corners.
The standard EMC tests should be specified using generic cables, connectors, etc.
Copper tape is better, thick copper even better, and thick copper with welded seams better still.
On more than one occasion I've had external USB drives lose their MFT (Master File Table) data as a write update was corrupted by noise. Essentially, the external noise killed the data on the disk.
Often, when one pulls these USB cables apart one can see the shielding coverage is 30% or less.
There's not the time or space to address your points in sufficient detail except to say that years ago, I used to work for a well-known electronics company that made radio and television broadcasting equipment and I was employed in its prototype laboratory. My work covered a diverse range of technologies including audio, VHF and microwave. All those points you've mentioned were day-to-day bread and butter and we had to address them in detail as the equipment had to pass/comply with broadcast standards. For example, we had to design audio equipment so that the 600Ω balanced lines joining them had to reach a specified common mode rejection ratio, this involved multiple techniques, shielding, good earthing, using 'technical' earths and testing EMR leakage to and from interconnecting cables (for instance, Belden Beldfoil cables)—all of which were especially important in very low distortion designs that didn't rely on transformers to achieve balanced inputs and outputs.
Similar issues were relevant in RF and microwave equipment, minimizing stray capacitance and inductance in chassis/component layouts, etc. Rather than give you an example I refer you to one of my treasured books on the subject, Microwave Receivers edited by S.N. Van Voorhis, 1948, specifically pages 261-266, Section 10.6 (although other nearby sections are also relevant). This book is ancient but the theory and rules still apply. It's part (vol 23) of the 28 volume Rad Lab† series produced after WWII and this set is still one of the greatest tomes ever written on any subject in engineering. It can be downloaded from the Internet Archive from here: https://archive.org/details/MITRadiationLaboratorySeries23Mi.... (Incidentally, I still have my softcover Dover edition from when I was a teenager, it's not the kind of book you throw out.)
† List of Rad Lab series: http://web.mit.edu/klund/www/books/radlab.html
We have generally had good experiences when sticking to these patterns, but are always looking for possible improvements.
There are several issues here: (a) this topic will time out in a few days and I'll not be able to contact you about the matter, that's likely a problem; (b) the paper address both general shielding issues and some very specific ones that relate to specialist research equipment; and (c), whilst I am happy to provide some general comment on the paper here at HN, there are some in-depth issues that I would prefer not to comment thereon given the paper's origin and the nature of my intended comment.
The reason for my point (c) is that I used to work at an establishment whose work concerned itself with hardware that is somewhat related to that mentioned in the paper and thus its instrumentation incorporated magnetic and EMR shielding of a similar standard to that [which would likely be] employed on equipment mentioned in the paper.
This raises two issues and it depends on your reason for asking in the first instance. If you want general information about shielding then I have no trouble giving that; alternatively, if you work with technology similar to that which is discussed on page 4 of the paper then it's a somewhat different matter. Thus, I would have to be much more specific as the shielding requirements are not only anything but trivial but also they are often much more demanding to implement. It's here that I have some issues with the paper and it would be unfair to air them in public without informing the author first (and as he's not sought feedback the matter is moot).
If you are involved in similar technologies that demand similar standards of shielding then I'd be glad to discuss the matter further in private but that seems difficult given there's no way on HN to exchange private messages/ensure privacy.
Anyway, please provide me with more specifics about your application/requirement and if I can contribute anything useful then I'll try to help.
Related: https://annoying.technology/posts/08834ce6ea3edc5a/
I also, personally, have a strong preference for USB Type A connectors over the Type C and Micro variants, cable stability and port wear is noticeably lessened with the larger cable seating.
USB-C ports are supposed to be 6X more durable tested to 10,000 connections vs 1,500 for USB-A.
https://en.wikipedia.org/wiki/USB_hardware https://www.anandtech.com/show/8377/usb-typec-connector-spec...
I've personally noticed the USB-C is much more durable. Something about the cage of the USB-A catching and bending more easily.
This was with ~5-10 plug ins a day, 5 days a week, for ~2.5 years. So around 3-6k connections.
Only port I've broken on my laptop is HDMI. Fortunately it has a thunderbolt port as well for video, but now I need a dongle.
Now that the port is becoming a charging port, it's used a lot more than data-only USB, and in ways that torque the port worse. You plug your usb phone into a usb brick and it largely isn't going anywhere, but on a laptop, you are plugging and unplugging the device all the time. You might be leaning at angles with it on your lap and adding pressure on the port (something I inadvertently notice myself doing once a week). On top of that, the go bad part, the flimsy inner pin, is on the computer side rather than the cheap cable side. After a year of use, my usb-c port went from snick snick to wobbly, both on my macbook and my nintendo switch.
In contrast, I've never had this happen with lightning port on my iPhone despite all the abuse and lint I give it, because the design is inverted with the flimsy pin on the cable being inserted into a simple slot in the phone. The old macbook magsafe plug was just a magnet holding contacts firm against each other, not even inserted into anything, so if it got torqued or abused it would just pop off and there would be no harm to the computer or the plug really, and you spent no force or effort jamming it into the computer since the magnet did the alignment work and made the connection for you (Typically, I would just grab my magsafe cable a foot up from the end and loosely slap it against the port basically and it would seat itself).
With the shortcomings of the design of USB-C, in comparison to the older standards, you put on a lot more wear and tear on the port.
One point in favor of USB-C when it comes to wear is that the moving parts (springy tongues for metal contact and clips to hold the plug in place) are on the cable, whereas on Lightning they are in the plug. So in regular use with Lightning you are wearing out the port, whereas with USB-C you are wearing out the cable.
Anecdotally, I've worn out the Lightning port on my Apple smart charging case, but I don't have any USB-C phone to compare to (I have USB-C on my laptop but the use case is too different to draw any conclusions)
Also anecdotally, it seems due to required tolerances, it's much more difficult for third parties to manufacture lightning cables - if you buy cheap lightning cables on Amazon, they tend to have poor fit and get loose quickly. Whereas I've never had that problem with the cheap USB-C cables that get packed in with aliexpress junk.
The issue is a compounded one - most laptops come with two or maybe three USB-C ports with the expectation that you'll dongle the crap out of them - they may be more stable than USB-A sockets but they suffer a lot more wear from the need from dongles which comes about due to the lessening of ports available. So USB-C might be a lot more reliable if manufacturers didn't fetishize trying to provide the absolute smallest number of ports possible but as things are the UX today is far worse than the UX five or ten years ago when I'd have a litany of ports to plug in whatever I need.
I hate dongles and they work absolutely terribly with USB-C, give me a mix of USB-A and USB-C connectors so I can plug in everything without a dongle and I'd be happy as a clam. I thought I'd write this out just due to the fact that mistook a bad UX as a purely port based technical issue - I can't say I've had too many problems specifically with the USB-C connectors but every time I've had to use USB-C connectors it's been a terrible experience - if you follow what I'm saying? I think you may be quite correct about reliability but also missed the compounding factor.
1. Which rests on a laptop stand to angle the device to a better reading level
2. Every time the internet goes out the dongle fails to automatically reconnect for some reason I can't be arsed to figure out
I’ve been using a major OEM laptop and dock at work for about 3 months. The cable is already jiggly.
Kinda odd that you even have to think about basic features like that these days.
It seems mine can only be connected to two devices, and gets a bit weird when the secondary device wants to make sound but the primary device is already making sounds.
I have missed calls because of that.
I'll throw them on to listen to a podcast while I'm cooking or something. But then they'll connect to my desktop as well. Depending on what it's doing that day, sometimes it connects in headset mode instead of headphone mode which means it has exclusive control of the headphones and I can't get my podcast to play for hell or high water.
So instead I have to trek down to the basement, unlock my computer, disconnect the headphones from there, then get back to it.
And I found my Rpi using 50% cpu being stuck connecting by bluetooth to the digital piano.. for no reason. :)
Too many features, no standardized labeling for what cables support, not truly reversible connectors, dongles and hubs that barely work unless you drop hundreds, etc.
I'm hoping to god we learn for USB4.
I agree that that will be the most likely outcome, but I wish it wasn't going to be that way.
What do you mean by that? USB-C connectors are not truly reversible? USB3 can also exist in USB-A and B variants that are indeed non reversible.
While you are taking the rectangle and turning it into a circle, go ahead and change the design...
Support for these features is still mostly optional. The only real upgrade is that -- as far as I can tell -- USB4 requires host devices to support DisplayPort. It's still a USB-C connector, so I don't see the cabling situation improving.
I've only ever noticed Thunderbolt (which I've always seen denoted by a lightning bolt) and USB-C. And never really had a compatibility issue outside of that.
One issue I have noticed is that there are dozens of 5 and 7 port USB A hubs out there, but there are basically no 5 or 7 port USB-C hubs. Lots of multi-purpose hubs with SD slot, multiple USB-A ports, and maybe 2 or 3 usb-c ports. And those USB-C ports will often have higher output via power delivery protocol, so obviously there's a thermal limit to how many of those you can pack onto a small hub. But why are there no non-power delivery simple usb-c hubs that would be a drop-in replacement for USB A hubs? It seems like this is kind of an issue for more simple peripheries to switch over to USB-C.
Oh boy. When it comes to data, USB-C cables can be either limited to USB 2 or fully featured USB 3.2. But that's just one dimension - for high power PD your cable has to be e-marked, and for longer cables (1m) they have to be active in order to support Thunderbolt at full speed.
If one looks at how well Bluetooth finally got straightened out after version 5, I think we have good reason to hope!
It's not surprising but still insane that so many USB cables aren't properly shielded, thus making all of the FCC's efforts regulating devices effectively useless as your USB cable turns into an antenna to transmit garbage.
[1] https://twitter.com/newhouseb/status/1352796299700162560 (note this is running at 6ghz, but also works at 5ghz w/ more noise)
the bitrates is also much lower for SPDIF. According to wikipedia it only supports uncompressed 48khz 20 bit PCM audio, which translates to a bandwidth of 120KB/s. I'm not sure when it was introduced, but wikipedia says USB 1.x was introduced in 1996 and had a bandwidth of up to 1.5 MB/s.
I'm back to plugging headphones directly into the case, as the output from the jack on the screen was noticeably worse
There isn't a good solution here. People want neato fast stuff and they want it cheap, small and everything else that precludes good RF hygiene, and no one wants regulators interfering. It's intractable.
Optical stuff has a curve too, but until it can compete with a bit of stamped metal inside a molding on price, it will be too expensive for USB. The market simply will not allow it.
[1] https://www.bestbuy.com/site/pny-elite-x-fit-128gb-usb-3-1-f...
Some problems are intractable.
Of course USB wants to also support dirt cheap devices that don't need all that transfer speed. So you probably need to put the transceivers in the cable, to allow cheap devices to ship low-speed copper cables. Then you have problems with bulky cables, in addition to the bending radius challenges of fiber.
It's crazy that the issue was reported as back as 2011, but Apple didn't do anything about it.
I'm pretty sure I've also used that arrangement with both 2.4Ghz and 5Ghz Wifi connections.
I also have an M1 MM that uses a 2.4ghz dongle plugged into a USB-C hub from Anker to connect to a Logitech mouse and it’s barely usable.
The big takeaway is that at 5GHz signals often "leap off the conductors" at the slightest provocation. And clock skewing and other attempts at breaking data/signal correlation have limited ability to counter this.
For a long time I had a USB 2.0 cable extender with an RF choke on it, that I would connect to USB 3.x hubs and then plug the Logitech transceiver into that.
2.4ghz is like the duct tape of the electromagnetic spectrum...everything runs there.
Basically microwave ovens came before the ISM band. 2.4GHz is the sweet spot for heating water with microwaves - there's a few frequencies that work, but too high is expensive and difficult to shield, too low requires more power to get the same results.
So this band was effectively "written off" by the ITU/FCC as being too noisy to commercialise. The result is a band where you can do almost anything you want, as long as you emit less power than a microwave oven does. This makes it ideal for local-range applications that don't want to deal with the licensing requirements of 'real' bands - as long as they don't mind sharing it with microwaves. It's the typical story of a lightly-regulated free-for-all.
2.4GHz isn't a mess because everything uses it - everything uses it because it's a mess.
Sorry, but no.
2.4GHz being a sweet spot for water molecules is a myth.
https://wtamu.edu/~cbaird/sq/2014/10/15/why-are-the-microwav...
It was always a junk ISM band, eg used for industrial applications that radiate interference.
Companies can use it on the condition that they accept any interference.
We then used a spectrum analyzer to better understand the extent of the interference, tried shielding as described in the whitepaper, as well as separating the components physically as much as 30cm -- all without success. The only solution that worked was to replace the USB3 cables with USB2 cables and acquiring images at a lower frame rate.
I don't even want to know how many people have been affected by these issues over the years. USB3 devices should come with a warning sticker on the box.
And I thought the motherboard was defective...
Things can go as extreme as covering the entire USB 3.0 lane on the PCB with a solid RF shield from the chip, to the connector.
Things such as above preclude any chance at USB 3.0 getting into cheaper product niches.
I once looked for a good USB 3.0 testbench laptop to test devices with, but found out that laptops themselves have terrible USB 3.0 RF isolation.
I went through many laptops of reputable brands, but the only laptop I seen where USB 3.0, and WiFi were working flawlessly was a very old Sony laptop from 2010.
I had a similar problem, and it turned out, I think, to be Bluetooth congestion.
I was in an office that was partially a call center, so lots of people on Bluetooth headsets. Plus all their mice and trackpads. Plus their keyboards. Plus regular headphones. Plus plus plus.
My desk faced out the window onto a public street. Very often, when a group of people would walk by my window, presumably each with his own Bluetooth devices my trackpad would disconnect from the MacBook.
The solution was to keep the trackpad plugged into the machine, since I never took either anywhere anyway.
But I'm reasonably sure I've also experienced the Bluetooth congestion issue in a different context. I have an old, mediocre Bluetooth head unit in my car, and it disconnects from my phone ~90% of the time I drive in traffic. Works flawlessly on back roads and the like.
http://www.smallnetbuilder.com/wireless/wireless-reviews/322...
For the first time I bought myself nice headphones. And for some reason, unlike previous headphones, I can now hear very clear buzzing patterns any time my phone gets a Signal message.
I don't get it. It's not SMS, it's Signal messages over LTE. And the phone isn't connected to my headphones in any way.
And it just really diminishes my enjoyment of these things.
If your headphones are wired into the back of a desktop, bad power shielding in your desktop itself can cause similar problems. An external dac can isolate your headphones from your desktop.
As a result I probably changed how my headphone cable behaves as an antenna compared to the shorter cable that went straight from laptop to my ears.
Rather than meet an appropriate level of RF Immunity, the Manufactures/Importers lobbied the FCC to permit Lax standards and instead rely on a FCC Part 15 Compliance sticker which reads "...this device must accept any interference received, including interference that may cause undesired operation".
I don't actually use wifi, but I suppose it's got something to do with wifi and bluetooth being handled by the same mPCIe card.
https://hackaday.com/2019/11/28/raspberry-pi-4-hdmi-is-jammi...
Things like good shielding, appropriate PCB design, differential cable drivers, and common-mode chokes.
If it were a good design, it would work satisfactorily with generic cables and connectors. And if it does require specific connectors (etc) it should not have removable cables.