hello, I have a lab bench supply that can do up to 30vdc at 10a and a battery bms that’s 12/48 volts dc. how do i use this to provide a voltage into my framework laptop? is complex switching with usb c pd the only solution to power my framework 16 mobo?
i fail to see why i’d want to do that when i could (theoretically) use buck boost converters or whatnot, which i allready HAVE in 250w variety instead of buying yet one more (expensive) cable
side note, can i provide power over usb c normally to the gpu or does it have to be 240w down a single cable to the motherboard
“why don’t u just use wall outlet”
i want to run my laptop off of a 12/48v solar system’s battery bank. switching up to wall outlet voltage and back down to the same voltage is… absurd and inefficient, and i don’t want to run 48->115->48 and loose 30% in conversion losses.
You discovered the main drawback of using PD for high-power electronic devices. For desktop computer, just use a DC-DC converter to 12V, replacing it’s PSU. For “traditional” gaming laptops, just convert to 19-20V and use a DC barrel jack or whatever the laptop is using.
The FL16 itself has a combined buck and buck boost. When 36-48V is used buck converts to 20V before going to the main VSYS rail while the buck-boost is used to charge the battery. if 9-20V is used buck boost is going to convert to 17.6V VBAT and VSYS is connected in parallel. Even if you know which specific pin is connected to VSYS, injecting 20V directly to the mainboard may cause a malfunction or worse.
However if you use PD, the main problem is negotiation. Putting a 48V source directly to the USB-C won’t work. You’ll need a PD chip. Unfortunately most PD “trigger” chip handles upstream PD signal, pretending to be a device that negotiate a specific voltage from the adapter. “downstream” PD negotiation that makes the device thinks of an adapter is very rare. I don’t think there’s any on the market for 48V. You can use a DC-DC converter but most of them are just adjustable bench power supplies. It’s very had to find one with built-in PD function AND 240W
Edit: Saw your “expansion-bay-shell” tag. If dGPU is not used, 100W is enough. There are plenty of DC-DC with 5-20V 15-100W PD output on the market like this
Edit 2: If you use GPU it’s connected to the mainboard and uses the same VSYS. You can use PD either from the USB-C on the dGPU or the one on the side of your laptop but not both i.e. no 2X100W input. For best gaming performance use 48V5A on a single socket
It depends how much work you wish to do.
The FW16 can take 48V input from the Expansion bay shell.
You would need to make a new circuit board for the expansion bay so that it linked up the 48V DC supply.
Look on page 10 of the FW16 mainboard schematic for some details.
The alternative is also to get a 48V DC to USB-PD adapter.
DC to DC converters are quite efficient.
The power loss mainly comes from converting DC to AC and back again, but using high spec converters even gets the DC to AC to DC conversion up to the 90% efficient, so the 30% you mention is unlikely to happen in practice.
I have not spotted any 48V DC to USB-PD adapters.
It is a little known fact that you can pass DC into a AC power adapter, you just need an AC power adapter that can either go as low at 48V, or put a DC-to-DC converter from 48V up to 110V DC, and then just plug that into a normal Main power adapter. It converts the whole process to essentially DC-to-DC conversion.
When I am repairing mains voltage switch mode PSUs, I normally feed them with 12V or 48V while testing, thus keeping all the voltages low during testing, but just make sure the load is small during testing also.
Until there are 48V DC to USB-PD adapter commonly on the market, maybe a 12V car charger unit will surfice. Just plug 12V DC from your solare to the car charger pins. You probably only get 100W from it though.
Wait, you can? Where’s the location of the buck converter that converts 48V to 20V? If it’s located in the expansion shell what would happen if you remove the expansion shell and connect the laptop to a PD input via one of the side ports
That may be true, but I’m able to do some shitty design work. Maybe I can inject power through the battery holder, but considering the framework is being treated like a SBC in my use case, I very much want to input the voltage through a sturdy jack that won’t pull out. USB c is “fine” but since I’m using the GPU for computer vision I’ll need the full wattage and then some so it doesn’t throttle. Power isn’t an issue, I have pouch cells the size of a car battery, it’s power delivery. My bms is 12 volts, but the laptop needs a ungodly 5+a 48 down one line. I really really wish I could use multiple lines of 36 volts or something to be easier on the converter,or maybe I can use the buck unit I have which is 250w, but I’d still have to negotiate it into the usb-c port, which I’m not sure is so easy.
That means our general issue here is USB c negotiation, or bypassing the circuit.
It’s probably harder as you’ll need to simulate a TI BQ40Z50-R3 BMS to trick the laptop into thinking it’s a battery. Another problem is using the battery only (real or simulated) makes the laptop throttle. The laptop will also throttle if the battery is less than 60% even if 240W adapter is connected because of the peak. It’s better to solder the 48V on the input side of the buck converter if you can find the location that @James3 described.
You can feed 48V DC into the VADP_GPU pins of the Interposer.
You would need to design an expansion bay PCB to do that and have the EC firmware control the on/off power to the VADP_GPU pins.
The DC-to-DC converters from 48V DC to 20V DC are on the main-board.
But saying that, I think a more useful device would be a 48V DC from Solar to USB-PD 240W usb-c power adapter. It could then be used for a multitude of other applications.
So, if you are making a PCB, make one that does this.
