cross-posted from: https://infosec.pub/post/51092505
I have a 3-pin Dell PSU spec’d as 19.5v 3.34A and DMM measures at 20v unloaded. The power reqs on the HP laptop are 19.5v 3.33A. AFAICT, that’s a match. Even the 3-pin barrel connector geometry matches. So what’s the problem?
The center pin is a “smart pin” aka “data pin”, as I came to learn. According to @over_clox@lemmy.world, this PIN basically passes brand info so HP can reject non-HP PSUs. But according to this article, the data pin is used to “regulate” the voltage – which I take to mean it negotiates the voltage for transmission as USB-PD does. If it’s merely negotiating power reqs, then the Dell should work.
It’s possible both sources are correct… perhaps HP has its own proprietary way of negotiating voltage and Dell has theirs.
Someone even claims that the data channel is used to communicate temp info so a PSU does not overheat. If that’s true, then perhaps it was a good design (noting that the USB-PD standard did not emerge yet).
But is it true about the communication? Considering there are hacks with a resistor or something to satisfy the data channel, it’s hard to believe that the communication is that sophisticated.
This page advises HP owners to always buy HP-branded PSUs. But it’s unclear if that’s just overly safe nonsense.
I’m getting the sense from your comments that you think that USB-PD will deliver a higher voltage than requested. This is not accurate. A device can request any of the voltages specified by PD, and the power supply will supply that voltage if it is capable of doing so. If not, it will provide 5V. It’s very possible to get a sippy that won’t power your device, but a working, PD-compliant power supply will not damage a working, PD-compliant device.
Obviously, if you get a barrel jack adapter that requests 20V and then plug it into a 5V device, you’re going to let the magic smoke out, but that’s because barrel jacks don’t do negotiation. The power supply correctly supplies 20V as requested, and then you plug that 20V into the wrong thing.
See the experiment I linked in the post that you replied to. If 19.5v is requested, 20v is supplied. If 12v is requested, 9v is supplied (not 5v). But also since 9v is optional, some PSUs would push 5v. So the voltage you get depends on what the PSU makers feel like sending.
Yet they will fail because compliance is subject to interpretation and the ambiguities are disputed. Try plugging a USB-C bicycle light into a Lenovo USB-C PSU. Even though they both handle 5v, it fails. The Lenovo device makers believe all voltages, even 5v, must be negotiated. The bicycle light makers believe a 5v “default” /means/ they should get 5v without negotiation. The Lenovo PSU makers believe “default” means 5v will be the result of a failed negotiation (but not absence of negotiation).
And other device makers believe USB PD requires supplying the closest voltage when negotiation fails, not 5v. Those people probably also believe 5v should be sent in the absence of negotiation.
You’re mixing up the different adapters. You can have a USB→barrel and you can have a barrel→USB. The interesting case is what I described, when the appliance and PSU are compatible in terms of voltage but still fail because of the protocol. Also consider that a 20v barrel PSU could feed a dumb usb-c adapter which then fries a 5v device. In principle, the adapter /could/ do a negotiation and refuse to supply 20v to a 5v device, or even just cut-off anything above 5v, but the shitshow that we have is a marketplace with dumb adapters. And in both situations (usb→barrel and barrel→usb), it’s not foolproof. Countless consumers cannot handle the voltage matching task so if the plugs fit, they will try it.
You’re misreading the blog post. That adapter has 19.5V written on it, but it requests 20V. 19.5V is not a PD voltage.
The latter is pretty much impossible to make to spec, because it can’t negotiate voltage. A barrel jack is single-voltage by necessity.
Respectfully, I’m going to have to disagree. It’s much more interesting when it catches fire, and I’d rather have a protocol that sometimes doesn’t operate in order to avoid the excitement of one that sometimes explodes
No matter what charging standard you use, some idiot could cut off the end, wire it to a stun gun, and send 100,000V into your phone. That doesn’t make the standard bad, it makes the adapter bad.
If you do it correctly, a USB power supply being adapted to a barrel jack won’t destroy any device that can tolerate 5V. It may not operate the device, but as long as nothing breaks, I’d consider it as foolproof as anything reasonably can be.
That’s just how electricity works. You can get 5V devices that use 5.5x2.5 barrel jacks and you can get 20V 5.5x2.5 power supplies. You can get C13 leads that connect to 240V wall outlets and you can get C14 devices that only accept 100V. If someone plugged things in without reading before USB, they broke things and started fires. PD simply doing nothing when incompatible devices are connected is a massive upgrade over that, even if it can be frustrating sometimes.
That is not what the blog post said. We know the /result/ is 20v, not what was requested. It’s a lie and abuse to print 19.5v on an adaptor that is actually just requesting 20v. It’s certainly /possible/ that is what is happening, which obviously proves my point nonetheless. It’s yet another demonstration of the USB PD shitshow. Either way in fact. Whether the negotiation is a lie, or whether the source is saying “close enough, will send 20v”, it’s a shitshow either way.
But now it’s worse because consumers who know how to pay attention to voltage are getting a product that’s not delivering what they signed up for.
Worth noting that the PPS spec of the PD3 standard allows for 20mV step adjustments to the fixed voltages. If the 19.5v adapter is not lying, it’s trying to down-step the 20v (which will only work if the source PSU is PD3 or higher).
First of all, the adapters exist. Not impossible. If you meant to say a smart negotiating one is impossible, I don’t believe that either. It could detect whatever the barrel passes and use that result to negotiate only for that available voltage. It could even have a voltage converter that supplies voltages that differ from the source. In fact I have a DC-DC adaptor. You probably have one and don’t know it. A USB charger that plugs into a car’s cigarette lighter converts 12v to 5v. I have one that has a variety of tips and a range of voltages. Detecting the source voltage may not be cheap, but not impossible. It would also be possible to have a cheap manual switch to specify input voltage.
If you find explosions interesting, you can go play with fireworks or watch war movies. It’s not academically interesting that tech illiterate consumers find countless ways to shoot themselves in the foot. When engineers — who /should/ know what they are doing, push a shitty standard that burns even consumers who have the basic knowledge to match devices, that’s academically interesting because it’s people we expect to be smart fucking up. When a whole industry fucks up, that’s academically interesting. I cannot charge a 5v bicycle light using a USB PD spec’d PSU because the industry fucked up on the meaning of “default”.
Dumb consumers are not in short supply. It used to be that only tech-illiterate consumers were getting burnt. Which created an incentive for consumers to gain knowledge. The nannying attempt at a remedy has enabled informed consumers to get burnt. At the same time, it dumbs down society. When expert consumers can get burnt, of course it reflects a poor standard.
And poor practices. A good standard anticipates what the market will do. They did not anticipate different product makers having a different interpretation of “default”.
It’s now clear why you have no problems with the standard – you have a low expectation. The Lenovo PSU that supports 5v /should/ be able to charge a 5v appliance. But they botched the negotiation and absence of it. When both sides of the negotiation are compatible in terms of physics but they cannot work together because of a miscommunication, it’s a shitty standard.
The whole point of a standard is to establish an expectation that different components /function/ together. Doing that without fires is a given but not the end game.
The blog shows situations where voltage is substantially less. It’s a general rule of thumb that supplying too little voltage usually does not damage things, but I do not think that is absolute.
I was describing how /people/ work.