# Running 3.5" Hard Drive off USB 3 (900mA / 5V)

I was having a discussion with someone about why 3.5" hard drive adapters don't exist that run solely off USB. Please forgive me but my electrical engineering knowledge is minimal and been a long time since I've used it for anything practical.

I know typical USB 3 can provide 5V / 900mA (4.5W) DC.

12V conversion end up with 5v/12v (0.416667) x 900mA = 12V 375mA

So effectively you should be able to get 9W out of each USB port. 5v 900mA or 12V 375mA (or somewhere in between, right?) Hard drives require 12V and 5V DC power. Would it be possible to run a hard drive off USB power, and how many?

Looking at the Seagate Exos Product Page as an example (pg 12): https://www.seagate.com/www-content/product-content/enterprise-hdd-fam/exos-x-16/en-us/docs/100845789f.pdf

Peak spinup current 12V: 2.04A

Peak spinup current 5V: 0.90A

Maximum Operating Current 12V: 0.65A

Maximum Operating Current 5V: 0.43A

Question 1: Assuming spinup current could use a capacitor? How to calculate time required to charge capacitor and/or what capacitor(s) to use to charge from 5V/900mA (or multiple lines) to deliver 12V/2A + 5V/0.9A?

12V / 0.65 A = 5V / 1.56A

So 1.56A + 0.43A ~ 2A @ 5V

Question 2: So is it correct to assume 3 USB port should be able to accomplish this?

If I can understand the math I'm almost tempted to try to build such a circuit and see if I can get it to work. Maybe I'm way off base here?

Thanks for any responses.

• You can't run it off of USB alone. You could use a battery pack though. Commented Dec 24, 2021 at 17:54
• Your math is waay off. You need to supply the 24W (12V 2A) for the HDD to be able to spin up. That is about 5-6 USB-A ports worth of power. USB-C kind of fixes that problem. Commented Dec 25, 2021 at 10:10

If I'm not mistaken, a USB 3.1 port can even deliver 2A (nb : it might require power negotiation first).

But basicaly, it's just a question of power : each USB port gives you 4.5W (if 900mA) or 10W (if 2A).

Your hard drive, you need : 5*0.43 + 12 * 0.65 * 1.3 = 12.3W (nb : I added a 1.3 multiplier for the 12V, to take into account the losses in the converter : it might be a bit more or a bit less depending on the efficiency of your converter). So basically, for normal operation, you need 3 USB ports @900mA or 2 @ 2A.

Then there is the question of brigging the drive to spin : you need 5 * 0.9 + 12 * 2.04 * 1.3 = 36.3W : thats about 3 times normal operation, so if you store energy with the same current as in normal use, you will need to store it for (at least) 3 times the start-up duration (no idea how long that is).

In practice, you will have trouble finding capacitors big enough to supply 2A during start up.

Basicaly, for a capacitor, you have :

• Q=CU (Q the charge, C the capacity, U the voltage), which gives dQ=CdU
• I = dQ/dt : I the current, dQ change in charge, dt duration

So you get C = dQ/dU = I * dt / dU

So (for the 12V), if you allow a change of voltage of dU=0.25V and you need 1s for start up (if might be far more), and want a current of 2A, you will need a capacity of C= 2 * 1 / 0.25 = 8 F

You won't find any capacitor that big (I quickly checked on digikey, for 12V, the biggest capacity is 1.5 F (so you woul need 6 of them) : it costs more than 200€, and is 10cm diameter / 25cm high).

So as @thexeno pointed out, the only way to go is to add a small battery (for exemple a small Lipo 1S with the corresponding converters).

PS : be carefull when playing with USB : if you do a mistake, you can destroy the motherboard of your computer.

• Thank you for this detailed explanation. The capacitor scenario is what I imagined it might be. So even as an experiment might be quite challenging to get it to work. If I do attempt it, I'd use an old motherboard or even just USB multi power brick to see if I can even get it to spin up before connecting it for data. Commented Dec 24, 2021 at 16:23
• The capacitor calculation does not make sense, as you would not put the capacitors directly on the 12V and 5V supplies for the hard drive. Like you mentioned the LiPo, it would be between USB input and HDD supplies, so would the capacitor. Commented Dec 24, 2021 at 16:29
• There are different places where one might put the capacitors, but the main goal was to show that capacitors are not the way to go. One might slightly decrease the cost by putting the capacitors instead of a battery (with a converter), so one can empty them entirely. But even then, the stored energy is E=0.5*CU², so C = 2 * E / U². The required energy is 12V*2Aduration (lets say 1s), so 24J. So we get C= 2 * 24 / 12^2 = 0.33F. It's already far less, but still, its a 50€ capacitor, sizing 7.5cm diameter and 12cm high Commented Dec 24, 2021 at 16:42
• USB3.x is limited to 900mA. If you need more, you need to go to USB-PD or the old USB charger specifications. Commented Dec 24, 2021 at 17:52

They don't exist because they are not practical. If you need 10 watts and need to connect three USB cables just to run a single drive, and even require complex energy storage solution to get it started, it will be quite complex to draw power from three cables in a way that it fills USB specs. You can't just short the USB 5V output pins together.

So it not impossible, it can be done, but it all boils down to if you want a complex expensive solution that requires three free USB ports, or cheap single cable solution with external power supply.

• Thank you. I agree, and I had similar thoughts but someone else trying to convince me it's possible and practical. But it got me thinking if it's possible to actually make it work. Fun experiment if nothing else. Commented Dec 24, 2021 at 16:20
• Well, you can't connect the 5V outputs together or you will fry something. If someone says it's practical using 3+ USB cables to run a single 3.5" drive while risking frying the motherboard in the process of doing the experiments, I suggest letting that person do the experimenting with his/her own motherboard first. That's why they are not being sold, as it would not work to sell this kind of product, there is just too much risk that it either does not work or it damages something. Commented Dec 24, 2021 at 16:41

you made the math wrong: from an USB3.0 you can get up to 5V at 900mA (4.5W), and with a converter, you can get an output of 12V at 375mA. At this point you have NO power available for the 5V rail needed by the HDD, and also is not enough for the 12V itself. You can't power an HDD which demands 12V at 0.65A.

Secondly, is not just a capacitor that solve the problem: capacitors will vary A LOT (linearly) their voltage when providing constant current, and they have very, VERY limited energy: ideally you need a battery to sustain the duration of the peak of an HDD. How do you charge it then? When the HDD is off? You get the point :)

You need to design a dedicated converter, which coincidentally, is what the market offers already.

• Thanks. But I guess I was asking if possible over multiple USB ports (2, 3, 4?) to get the power required. I agree with you, I don't think it would really be very practical, but just wondering what it would take in theory to make it work. 2 or 3 USB ports could deliver proper usage power, and spinup power only need a short burst at 12V/2A. Commented Dec 24, 2021 at 16:17
• Parallel operation of multiple USB ports is not suggested. They have individual restorable fuses, and putting fuse in parallel is not the best. More importantly, some (low quality) USB are from a hub or a splitter, meaning that these 2x USB 3.0 could even deliver 450mA each in parallel. Something to consider even when using more than one external 2.5" USB. So, I'd say no, unless you get dedicated supply for the ports, hence losing the point of the initial idea. But is a good exercise. Commented Dec 24, 2021 at 18:47
• Also if some ports are powered from standby supply and some ports from the main supply, it is a sure way to damage a system, if the USB 5V wires are simply shorted together for paralleling (such as used for early external 2.5" USB to IDE enclosures, such a cable is not allowed to exist). Commented Dec 24, 2021 at 19:03

I know typical USB 3 can provide 5V / 900mA (4.5W) DC.

That's not the power a USB 3.x port can provide "typically", that's the minimum power a USB 3.x port can provide and still meet the USB 3.x spec. I don't know if this is "typical" but I've seen many USB 3.x ports that will provide 7.5 watts, 5 volts at 1.5 amps. I've seen USB 3.x ports on computers rated for 12 watts as well, the maximum power allowed by USB-A under the spec. USB-C ports on some computers will provide 15 watts, that's using the USB-PD protocol on top of USB 3.x on the USB-C port. Drawing more than 4.5 watts from a USB 3.x port is possible while meeting the USB spec but it would be using optional extensions to the protocol.

Pulling a random 3.5" drive off my shelf and I see it draws 12 volts at 0.5 amps and 5 volts at 0.35 amps, so less power than you saw. Then the adapter board will be drawing power on top of that. That's going to exceed 7.5 watts that would be available on many USB 3.x ports but should still be under 12 or 15 watts from other USB 3.x ports. I didn't see any mention of extra spin-up power on the label of this random drive from my shelf. If spin up power gets in the 30 watt range you saw then that's not going to work even under the 15 watts I've seen from USB-C.

To power a 3.5" drive and USB adapter, and not run the risk of frying some USB ports, the drive and adapter needs to stay within the allowed power of the spec. First thing is the adapter needs to ask for power before powering up the drive. To draw even 4.5 watts requires getting permission from the host to draw that much power. To draw 12 or 15 watts means asking for more power using the optional extensions to the USB spec. Not all ports will support more than 4.5 watts, and hubs powered by the host (as opposed to a separate power supply) may provide less than a watt per port. Some hubs don't meet the USB spec and so fail to provide sufficient power for many devices.

The number of 3.5" drive adapters that will run only off USB power will be slim to none because the market of people wanting these will be small. Even smaller will be the people that will have 3.5" drives that draw low enough power, and USB 3.x ports that can provide high enough power, that the two overlap. The people that want to comply with the USB spec will find compliance difficult while keeping costs to something people would be willing to pay. People that don't much care about compliance with the USB spec do so because compliance checking costs money, and if that's the case then they will find it cheaper to make it work with an external power supply. Devices that violate the USB spec still need to work to sell so adapters that fry USB ports from drawing 30 watts won't last long on the market.

• What you say is true for any USB device though. You have to design to min specs otherwise you can't ensure a product will work. You mention you don't see spin up power. The detailed product manual for hard drives (in link provided) shows maximum start current, and usually ranges from 0.5-0.9A on 5V and 1.5-2.0A (or more) on 12V, but typically 25-30W. But you've answered how I believe it to be as well, it's not practical. I was just trying to do the math to see if it was possible and what it would take to make it work even if needs a ridiculous amount of caps and USB ports. Commented Dec 26, 2021 at 14:46
• I wanted to verify your numbers with a drive I had, and I did not see mention of spin up power on my drive so I don't know how big of a problem that would be. Commented Dec 26, 2021 at 19:42
• Using power from multiple USB ports violates the USB spec. Using two USB ports has unfortunately become an all too common means to get more power for USB devices. Trying to spread power draw over three or more ports is just a bad idea beyond violating the spec. Commented Dec 26, 2021 at 19:47
• What drive model exactly were you trying to verify? WD, Seagate, Toshiba all call out "spinup" or "start up" current requirements. Commented Dec 27, 2021 at 1:04