Everyone agreed (except nVidia) that pushing 600W through 12VHPWR was a bad idea when the Molex connector wasn't made for it, but pushing 200+W through a much smaller USB-C connector with way smaller pins is somehow fine.
firewire-nvidia-12VHPWR.jpg

@phnt dell and lenowo still use large DC jacks for mobile workstations btw

@phnt nevermind they stopped in the latest ones...hooo boy I bet parts-people is gonna be rubbing his hands at all the new board repairs he'll do because the PD controller decided to find a pretty view.
https://www.dell.com/en-us/shop/dell-laptops/dell-pro-precision-7-series-16-laptop/spd/dell-pro-precision-pw716260-laptop/xcto_pw716260_usx

@phnt it mostly comes down to active vs passive power delivery, PD will cut out on failure modes, molex will start a fire and keep going anyway, that's why auxiliary safety things are starting to pop up too

@phnt Doesn't USB-C have some sort of overheat protection built in? From what I understand, 12VHPWR just dumps 200+W with no negotiation phase and any safety measures were tacked on by vendors after shit started catching fire.

@VD15 @i 12VHPWR just sends 12V and 600W max down the wire when all the sense pins are connected. Sense pins loosing connections will then throttle power draw or stop power completely depending on how it is implemented. (Number of sense pins connected also dictates max power draw.)

Still USB-C may have thermal protection, but you're still dumping large amount of amps on 20V right next to your data tx/rx pins which go right to your controller. Given enough power draw and inadequate connection which Type-C is known for and you can easily create an arc going to one of the data pins.

@phnt @i @VD15 also it's not uncommon to have laptops with blown out PD controllers

@phnt @VD15 yeah but on usb-c it's part of the spec, and on 12VHPWR it's a feature only implemented after multiple news rounds of house fires, at least gpu fires aren't as bad as lipo fires

@i @VD15 Molex Micro-Fit (the actual 12VHPWR connector) is only rated for 10.5A max current draw, which all GPUs using it practically exceed.

EDIT: It's Micro-Fit+, which is 13A max.

@phnt watts don't really matter, usb-c does it at 24v so that's uhhhhh, fuck that's still a lot of amps

@snacks The previous maximum was 20V, the new PD spec allows 48V which is just insane.
https://www.usb.org/usb-charger-pd

@phnt less than 5A sounds reasonable for usb-c connectors i think

@snacks It is reasonable amperage, if you can ensure that it won't spark on the nearby pins, which well, Type-C isn't known for the best mating tolerances and not being loose out of factory.

@phnt they'd need to be micrometers apart for arcing from what i can tell. You're just as likely to get a regular short

@phnt arcing on the power pins themselves while disconnecting is a bit concerning since that's gonna increase resistance over time

@snacks @phnt i’d trust the GCT type C connectors that i put on my diy keyboards if they weren’t USB 2.0 only but i’ve never seen anything close to that quality in consumer products

most type C connectors are pretty loose and wear out quickly

@snacks @phnt i think the pins are offset for that reason

@mia @phnt i'm still sure that you're more likely to get a regular short than a 2-3 micrometer gap (apart from the brief moment before the contacts touch)

@mia @phnt actually nvm. I forgot length units stop being 1/10th after milimeters. It's 0.3mm sorry

@mia @phnt and i just visualized that with calipers, bigger than i thought

@snacks @phnt what? no, 3 micrometers is 0.003 mm

@mia @phnt yes, that's what i meant. 48V should arc over 0.3mm, i just forgot that's not 3 micrometers

@mia @phnt disregard everything i've said. 48V over usb-c seems sketchy af

@snacks @phnt oh now i get it

yeah i would HOPE that a compliant (lol) device is at least protected from arcing across the data lines and will power down ASAP when it’s disconnected. i think the specification sets time limits for that

@mia @snacks Still no idea how someone can call this"good enough" design. 3 year old laptop and this is pretty good for what I've seen. The connector on my phone just falls out when I pick it up. I've cleaned it multiple times to no change.

VID_20260428191233.mp4

@phnt @mia even thunderbolt was prob much better than usb-c. This connector is so stupidly small and intricate

@snacks @mia Thunderbolt 2 was a miniDP connector.

@snacks @phnt zero wobble on the GCT connectors (the ones rated for 20k mating cycles anyway)

but yeah clearly they live in a fantasy world where people want to spend any amount of money on connectors

@mia @snacks Given how manufacturers cheaper out at first on two resistors making the orientation of the connector matter, I wouldn't trust them to implement proper protection.

@phnt @snacks yeah my sony mirrorless camera is like that

can’t use the sketchy chinese A-to-C charging cables with non-standard connector color

@phnt @mia god, im retarded today... meant lightning

@snacks @mia Some say that lightning also failed a lot, some say it's solid. It's the same as USB micro-B. I've never had a problem with it, but others say it easily weared out like type-C.

Also lightning is USB2.0 only, so "one connector for everything" people will hate it no matter what.

@phnt @snacks @mia that's the tightest usb-c hole I've ever seen

@phnt because it's not power that cause fires. It's current
USB PD never uses more than 5A, and scales voltage instead

@mo 5A@48V for their current 240W mode is insane.

@phnt much less insane than 50A@12V for 600W

Like, not even all outlets can't handle 50A

@mo The Micro-Fit+ connector 12VHPWR uses is rated for 12 A max, so it was always a recipe for disaster.

Still with 48V and relatively high current, you have a real potential for arcing between the pins and considering how flimsy almost all type-c connectors are, I think this is a ticking time bomb waiting to blow up someone's device. The power lines are right next to data lines.

@phnt USB starts with safe defaults, and only switches to higher voltage after both sides negotiated. At this point there should be no potential for arcing, pins are connected tightly

Also, If device is advertising 48V support, it's reasonable to assume it has necessary protection from that kind of voltage

@mo
>At this point there should be no potential for arcing, pins are connected tightly
Someone can slowly unplug the connector, trying to not destroy it and destroy their device because of it. Pins are never connected tightly, because almost all connectors have lots of wobble. Attached is my laptop with barely used connectors. My phone's connector is completely destroyed after 4 years of using it.

>it's reasonable to assume it has necessary protection from that kind of voltage
I don't think it's reasonable to assume that. It might be reasonable for the spec writers, to give it a "USB-compliant" stamp of approval. But considering how manufacturers cheaped out on just two resistors to make USB-C correctly work no matter the connector orientation and power supply, I wouldn't put much hope into manufacturers doing protection properly.

Remember, laptop manufacturers include such geniuses as Apple that put 50V backlight powerlines right next to iGPU data lines going directly to the CPU. Now USB-C PD is doing exactly the same thing.
VID_20260428191233.mp4