I have these two power bricks I want to convert one of them to power a Banana Pi and 3.5 HDD.


I want to determine if either is safe and the needed circuit modifications to do so. This whole thing is a curiosity and not needed from a functional stand point, but I'd like to understand why they have been working for other electronics.

The first one has a molex IDE HDD connector, I will call "Fineness" (rated at 2A for each 12v and 5v from the label photo), seems the most questionable to me because of what I would call overvoltage. Right after I connect wall power to the brick, the 12v circuit reads 16.25v. I connected a 12v fan with a label stating it draws 0.41A. Initially that circuit reads 15.8v with the fan then the voltage increased slowly. A few minutes later the voltage stabilized at 16.4v with the fan connected. Right after I removed the fan, the 12v circuit read 17.5v. The 5v circuit stablizes at 7.2v after slowly climbing. I don't have the knowledge to know if this acceptable with the small amp draw of Banana PI and the hard drive. I've seen forum post photos showing the BPi using as little as 200 mah at 5v USB and then contrasting that with the supposed manufacturer USB power rating of 2A: http://www.bananapi.org/p/product.html. My guess is the voltage sag would be minimal and the incoming voltage would still be far too much for a HDD and above the max of 5.25v max for the USB standard. This power brick originally came as part of a USB to SATA/IDE kit. About a year or more back I successfully used this to power a 3.5 HDD.

Fineness Power

My results from testing the molex connectors from the top down:

  • slot 1 is +12v
  • slot 2 actually empty, would be Gnd 1, but instead is a plastic stand-in to provide a firm fit???
  • slot 3 is Gnd 2
  • slot 4 is +5v


  • slot 3 connected to slot 1 reads up to 17v
  • slot 4 connected to slot 3 reads up to 7.2v

Is it worth making changes or even necessary to lower the voltage on this Fineness' power brink before I can trust it will not break the Banana Pi and a 3.5 HDD?

So if the Fineness brick is not an option, I could work on this one instead: Second power brick (see second photo on the imgur link above) has a barrel connector and is rated at 12v up to 6A. When I initially power it reads 16.5v and continues up to 17.75v over a few mintues. I had used this to power a cheap china made audio amplifier with no problem over several months of use and then stopped having a use for it several years back.

I have the same question for this brick about safe voltage. What is needed to create a voltage drop circuit to safely supply 5.25v - 5.0v to a USB plug on the Banana Pi and also 12v and 5v to the HDD?

To me it seems a problem for both of these power bricks to have voltages so far outside of the rated 12v and 5v ranges. There has to be room for a little voltage sag as a load is applied.

Is it actually important to stay within 5.25v for USB? If not, what is acceptable above 5.25v without being damaging?

Is it just by chance that these power bricks have not broken something in the past and they are just too low quality to spend time on correcting?

Do my voltage measurements with no load or a 12v fan not mean much without a more significant load such as 1A or 2A to see how the voltage sags/stabilizes?

Currently, I am using a USB charger for the BPi and the original power brick that came with the external HDD. I'm looking to learn and then apply that by laying out, testing, and then soldering the changes. I'd also like to power the 3.5 HDD without the enclosure's USB adapter which is presently obstructing connecting the HDD to the BPi through SATA directly.

Thanks for your help and consideration.


3 Answers 3


I have an external HDD power supply that is almost identical to yours (it has the same model number, but a different brand name). One notable difference is that when unloaded my PSU puts out a more reasonable 5.13V and 12.6V.

enter image description here

This type of switch-mode supply rectifies the AC mains to produce DC, then chops it up into a high frequency square wave which is fed into a ferrite-cored transformer. For multiple output voltages the transformer can have several windings with different turns ratios.

Output voltage is regulated by varying the on/off ratio of the square wave, but this affects all outputs so putting more load on one output will cause the other output's voltage to rise. The PSU cannot regulate each output independently, so it may be designed to stabilize the voltage of only one output (usually +5V because this is more critical for digital logic), or take an average of both.

Switch-mode supplies often need a minimum load to get good regulation. Cheaper units may have no regulation at all, simply relying on the transformer ratio to set the output voltage at the rated current. Your PSU may have worked OK with a hard drive because it is matched to that load - or it might be faulty. Either way I would not use it with the Pi, which could be damaged by anything over 6V. However you could use it to power a suitably rated step-down voltage regulator (AKA DC-DC converter).


In my experiance modern switch-mode power supplies even the cheap ones are normally within a few-percent of nominal.

Given that all your readings are significantly over by roughly the same proportion I suspect that your real problem is the multimeter. Many multimeters will over-read voltages when their batteries are low. Some of them are nice enough to give you a low battery warning before they start mis-reading, others are not.


I think you're going to need to find out what's inside that Fineness supply if you're serious about modifying its internals.

It sounds from its very poor load regulation that it's likely an unregulated mains-transformer-based supply. This can be often be verified by the "heft" factor. Such a supply of this output power would be pretty heavy.

I expect it unlikely anyone is going to know that particular unit well enough to tell you what's inside and how to modify it. You may be able to place an LDO linear regulator after it to give you a stable voltage close to that required, as long as you don't draw too much current. That would be easy to test with some different loads. I expect for a Raspberry PI it would work fine as long as you stay under say a few hundred mA.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.