I share the author’s concerns for off grid power use. 3 watts per drive (I assume a wall wart for a 3.5” drive) works out to 72watt hours per day. About 1/10th of a $200 solar panel. Double or triple that to cover batteries and electronics. Then you run into discontinuities in system sizing where you can’t easily add just one more panel. Conserving energy suddenly makes a lot of sense.
USB hubs with per port power control are the flying unicorn of hardware. Once in a while some vendor slips up and makes one, but by the time people figure it out the vendor has ended the model or replaced it with a different model under the same model number. It’s especially frustrating because many of the USB hub chipsets support the feature, but only if the manufacturer puts a switching transistor on each port. It literally would only cost pennies per port to have the feature.
Yepkit [1] makes these! They have 3 port hubs for USB 2 and 3, as well as single port "hubs" -- all controlled through a simple command-line interface or Python API. They also have other good stuff like USB controlled relay boards and upstream USB hubs to switch devices between hosts.
I've used their products in a variety of automation projects for clients and at home, and have only good things to say.
uhubctl's home page has a list of supported hubs. Apparently Amazon's house brand usb hub ("Amazonbasics") currently supports it, so it seems currently easy to find one.
I happened to have earlier purchased an industrial quality hub that's powered by 12-24V DC, and it turned out to have per-port power control.
Power factor is another concern BTW; I was measuring something like 0.15 power factor when the drive was powered down, and I'm unclear how such a low PF affects inverter efficiency (it probably depends on how good the inverter is). Powering the drives from 12V DC would probably improve their offgrid efficiency.
It's not that they don't want to, it's just a lot of extra cost for a product their bulk userbase won't be all that interested in. Think about it, before today (or even after), would you have spent $20+ more for a USB hub with per-port power control?
Not that much extra cost, I don't think. MOSFETs capable of switching USB loads are in the low single-digit cents each. It's probably just that everyone wants to squeeze out the last few cents of manufacturing cost.
Sorry to be critical, but I think that MOSFET is undersized for an application trying to save power. There are two issues, it is N-channel so will require a charge pump in order to use on the high side. It also has a rather high R_DS of 0.3 ohms. This means when switching (just) 1A, it will cause a voltage drop of 0.3V and will dissipate I*2.R_DS = 0.3W. 6% power loss when delivering 5W.
I'm in the middle of building an enclosure for a cluster of Raspberry PI like boards (~2.4A draw) and want to have software power controls. So I've been looking to do this...
There are specialized devices for this sort of thing from many manufacturers. I'm currently looking at devices from TI - TPS22958. It has a built in charge pump and an R_DS on of 0.014 ohms. So dissipation will be a lot less 0.081W at 12W power delivery, a 0.7% power loss. This of course comes at a cost -- even in the 10,000 of units it will cost ~40 cents a unit.
Tangentially, I'm curious on how OP monitors health of USB drives.
Getting SMART data from USB drives has been a pain on linux as the kernel rejects SAT ATA pass-through commands & so doesn't work with smartmontools[1].
I have tested across different USB 3.0 drives such as Seagate Backup Plus, MYDIGITALSSD|OTG, SAMSUNG Portable SSD T5 all of which use 'uas' driver under linux & if used with USB 3.0 hub like Tplink USB 3.0 port hub; it bridges using 'hub/4p' driver.
After trying out several other methods in vain, the only tool which I found to be helpful in checking for USB drive health on linux is discscan[2]; which tests for unreadable sectors & latency without using SMART values.
It's cool but I don't know if I would want additional latency (especially in the case where they are currently going to sleep) plus possibility of turning them off at the wrong moment to save $0.2 per month.
Why do you think this is about cost? The article says ”I've not had any network attached storage at home, because it's offgrid”, so I think it’s about preventing the house from powering down.
I can't find a concise page to link, but he has a house that's off-the-grid, using solar panels and batteries. So he pays attention to power consumption.
Thank you! The following quote resonates heavily with my mind:
> I'm fairly addicted to that point in development of a project where it's all about exploring a vast solution space, and making countless little choices that will hopefully add up to something coherent and well thought out and useful. Or might fail gloriously.
USB hubs with per port power control are the flying unicorn of hardware. Once in a while some vendor slips up and makes one, but by the time people figure it out the vendor has ended the model or replaced it with a different model under the same model number. It’s especially frustrating because many of the USB hub chipsets support the feature, but only if the manufacturer puts a switching transistor on each port. It literally would only cost pennies per port to have the feature.