On average that sounds pretty reasonable for a “normal workload”, get file, go to sleep for long time. With caching you aren’t touching the disk a lot on most cases.
It’s plausible but it’s not really a useful number, then again it’s a marketing flyer.
Fair enough, maybe not as absurd as I thought.
Since sleep doesn’t work with most usb enclosures I used those numbers are likely of limited use to you.
Just did a quick test with a USB power meter and at half it’s rated speed (about 800MB/s) it pulls around 3w of power. So can confirm the 75mw is indeed a pipe dream for actual disk access… who woulda thunk it 
They never claimed that, it literally says AVERAGE power over the whole run, so if it used 7W for like a second and then slept at pretty much 0 for the rest of the time you get 75mw average or something.
Yea I know. It would just be a very strange ‘benchmark’ if that is how it’s set up. I’m well aware that manufacturers fluff numbers and make things look better than they are, but I wouldn’t expect a benchmark test to allow a drive to just sleep through the entire test.
Obviously it looks like that’s pretty much exactly what happened, it’s just one of the most grievous abuses of such a thing for masketing fluff I have ever seen. It puts Apples charts with no values to shame!
Massaged numbers in a marketing flyer?!!!
Best massage I have ever seen.
Back on the topic -
Looks like USB port powering is a no-go. The USB ports remain powered after the motherboard is shut down, and the PCIE card draws power as soon as 12v is available even without the PCIE port/slot being active… From a quick search it doesn’t look like its possible to change the behaviour of the USB ports so this pretty much counts them out.
Going to have a bit of a look into pulling power from the main bus today, looks like it’s my only realistic option now.
Edit…
This is actually going to be the case for any of the discussed sources of power.
I can find 5v near the laptop VRM’s but it’s impossible for me to know if pulling another 10+w from these points is going to be ok, and they are also powered when the laptop is shut down.
At the moment I can’t see another option for this… I need the card t power on as the laptop is booting and at the moment that doesn’t seem to be possible.
… The Nvme slot has 9 +3.3v pins, each one should be 0.5a but stepping up 3.3v to 12v is inefficient in the regulators I have seen and even if it was perfect any voltage spikes would either cause the card to cut out or damage the port.
It’s getting a little more involved than I was hoping… But I can hypothetically create a circuit that uses the 5v from the keyboard breakout board I am already using (which is off when the laptop is off) to switch on a circuit drawing power from the USB.
You don’t what those, you want the 17-20ish V on the input to the vrm, I can pretty much guarantee you those aren’t on when the board is off.
Hmmm. I thought there must be a higher voltage but I didn’t find it, I’ll have another looksie.
it is probably easiest to grab off the input caps, there are probably a bunch of caps then some inductors and then caps again, what you want is probably the furthest from the cpu. Be careful probing though.
Edit: then again when looking at pictures online, I can’t even tell where the hell the vrms are, everything is covered in tape XD
So I can find 14v near the CPU and 20v near the battery… but they are all active when the laptop is off. So long as the battery is plugged in I can read voltage from all of these points. Not in sleep, fully shut down.
But There are some nice exposed caps and inductors right next to the battery with 20v. So I’m beginning to think just tapping this and adding relay or similar to control the PCIE card when the voltage is on/off may be my best approach.
What are you measuring between? A lot of stuff is low side switching so if you measure the high side vs ground it may look like more is on than actually is.
Ground and the input pin of the caps. I don’t really know what you mean when it comes to high side/low side.
You should probably measure both sides of the caps instead. Low side switching turns off the way to gnd which can be a bit counter intuitive but n-channel mosfets are better, cheaper and easier to drive so low side switching is used a lot.
Ok thanks yea just looking at a digram of high/low side switching. So if I were actually wiring this up after successfully finding a source I would solder to each side of the cap rather than positive and ground?
It’s obviously a little safer probing with only one probe being moved, but I’ll have an extra careful look.
That is a good point, if power to the vrm input is turned off via low side switching that would not work since the pcie card has it’s own ground and would bypass it. It is possible they just disable the vrm and don’t cut it’s power.
You may be kinda boned there but is it even a problem that the card is powered before connection? One of my usb4 to pcie adabters does that and it doesn’t seem to be a problem with all the devices I tested.
So I tested all of the caps that I can get a 14v voltage reading from, and they all show voltage across the cap as well. There may well be something I am missing given my lack of knowledge in this area. (for example I’m not potentially just measuring power stored in the caps am I?)
The 20v I found is actually only active when the laptop is plugged in, so I assume that is USBC C negotiating 20v for the battery charging.
There is a continuous 14v on a cap near the battery connector which is the easiest/safest source of voltage to tap. I can’t really see another way around it at the moment other than tapping this as a source of power, and controlling the on/off state of this with something I know turns off with the motherboard.
Or I ditch the internal battery and just use an external power source and tap that which is obviously way easier to manage.
edit:
The PCIE card is fine being powered on first, but if it remains powered it will be a phantom power drain, and if the enclosure is ‘switched off’ it is going to continue to drain the battery. The card definitely continues to draw power even when the PCIE connection is not attached.
Don’t know where else 20V would come from, the battery is way below that.
That does sound like by far the most reasonable solution
This would be quite a bit easier if we had the damn schematic XD.
