12 V Computers

It is possible to buy a UPS that fits inside a desktop computer and does indeed produce the multiple elv that the computer requires and feeds it direct to the mobo and other components - without going via "mains" conversion.

However, an external UPS producing these ulv would require the computer to be modified, as the only power input socket normally provided is "mains in". The UPS would still have to have a SMPSU to generate all the elv required and the design of the elv. There would also have to be a changeover pod within the computer - the design of which would be a "challenge", bearing in mind the very low voltages and very high current involved, and the length of wires going to the UPS.

Sue,

Sue, thanks for the reply.

My problem is that the UPS that I currently use can allow me to continue to use my PC for a maximum of half an hour usually with a warning siren continuously distracting me.

I'm thinking of using a bank of 12V batteries which would allow me to continue to work during brown-outs and load shedding for longer than half an hour.

Furthermore I am considering using solar power cells to further decrease my reliance on the mains grid.

I am aware that cabling 12V is not very practical for safety reasons and that converting from 12V to AC is the most practical way to go.

However, it does seem that converting from DC to AC so that the PC Power Supply Unit can convert it to DC again seems to be a little inefficient.

Dave Coventry

You are aware that the computer needs +- 12v and 5v ? Just wondering what is unsafe about 12v cableing? I would think you might be better off with a purpose built 12v computer ?

It might be easier, and take less power to use a laptop.

| You are aware that the computer needs +- 12v and 5v ?

Only those that are designed that way.

| Just wondering what is unsafe about 12v cableing?

High fault currents.

| I would think you might be better off with a purpose built 12v computer ?

Lots of small and embedded computer boards do use just +12v. Also, you might want to see this small paper:

http://216.239.57.110/blog_resources/PSU_white_paper.pdf

Just buy an inverter. They are efficient and avialable for under $50 at many department stores. John

Power supplies currently convert 120 volt AC to 300 volt DC; then create those DC voltages. We designed a machine once using a supply that converted 120 volts to 12 volts DC, charges a lead acid battery, and then used that 12 volts either from AC or from battery to create 5 and 12 volts DC outputs. Company was discontinuing the design complaining of no demand. People would rather get same by buying a $100+ UPS.

Power supplies currently convert 120 volt AC to 300 volt DC; then create those DC voltages. We designed a machine once using a supply that converted 120 volts to 12 volts DC, charges a lead acid battery, and then used that 12 volts either from AC or from battery to create 5 and 12 volts DC outputs. Company was discontinuing the design complaining of no demand. People would rather get same by buying a $100+ UPS.

Phil,

Really interesting article and covering much about which I was asking.

Actually it does not. The power still must be delivered to computer at 120 or high voltages for efficiency. If you think providing 12 volts to the computer is more efficient, well, that voltage is so inefficient that auto makers will soon (will have to) upgrade to something like 48 volts.

The regulation requirements of a MB far exceed the power quality of a bank of batteries.

I'm not sure about 12V, but there are ATX power supplies designed to operate from 48V battery banks (commonly found in telco central offices).

Well, no, that was the point of the OP.

Is it more efficient then, to use an inverter to take the 12V back up to mains AC?

It is more efficient to convert 12v (if that is what you are starting with) to the required dc output voltages and not via a 240/120 v 50/60Hz intermediate stage.

However, it is not efficient to transport 12v dc other than very short distances. Thus, if the 12v supply is some distance from the load, it can indeed be more efficient to convert it to a higher voltage for transmission.

Also, 12v > 120/240v ac inverters are standard items and benefit from the cost savings of mass production. They can be used to power a very wide range of equipment.

I live in the middle of nowhere, with power lines going miles over open field and moorland. So I do have one large bank of very large batteries (6v Trojans) plus one large surge arrest unit and one large inverter. Maintaining a scattering of small 12v batteries adjacent to equipment would not be practical.

There are also supplies like this:

?sc=8&category=13 or search google for "dc-dc PC power supply"

| On Mar 13, 7:27 am, "Cov" wrote: |> Really interesting article and covering much about which I was asking | | Actually it does not. The power still must be delivered to computer | at 120 or high voltages for efficiency. If you think providing 12 | volts to the computer is more efficient, well, that voltage is so | inefficient that auto makers will soon (will have to) upgrade to | something like 48 volts.

I would agree. The distance squared should dictate your voltage.

Still, a single common AC to 12VDC power supply would be convenient for a rack full of computers. I would speculate that Google could be putting as many as 4 to 8 computers per 1U of rack space, a couple hundred in one rack cabinet allowing for some power supply space. It would be practical to power several racks from a single power supply if care is used to deal with the really high fault current such a setup would have.

Telco offices typically run their 48VDC over a larger distance than just a few racks in one row. But at 48V, you can go further than with 12V.

My plan is to bring power to my computers at 240V (instead of the usual

120V in the USA). But that might end up being 240VAC to one big 12VDC supply if server mainboards start coming out with single supply voltage.

I believe the Google issue is more about having a single voltage and avoid the 5 or so voltage/polarities you get in an ATX powered computer.

| On Mar 12, 6:02 am, "Cov" wrote: |> It occurs to me that a UPS is a device that takes AC current from |> mains and stores it in a battery. |>

|> When the power supply fails, the stored battery power is converted |> back to AC to provide a temporary supply. | | Power supplies currently convert 120 volt AC to 300 volt DC; then | create those DC voltages. We designed a machine once using a supply | that converted 120 volts to 12 volts DC, charges a lead acid battery, | and then used that 12 volts either from AC or from battery to create 5 | and 12 volts DC outputs. Company was discontinuing the design | complaining of no demand. People would rather get same by buying a | $100+ UPS.

So what about standardizing on a 360VDC delivery voltage and cut the PC power supply in half?

It wouldn't be half. More like 2% or less of the supply components are involved in converting the input supply to DC.

That makes sense when trying to simplify the system, make maintenance such as hot-popping easier, and other solutions. One power supply for multiple servers does simplify hardware design. But the author somehow confuses this with massive efficiency increases. Power supplies must be at least 68% efficient. Higher efficiency numbers are easy with better designs. Other problems such as excessive harmonics and power factor can also be solved with better designs - more efficiency results. Somehow that paper instead says thousands more little power supplies all over a motherboard would be more efficient than one large one. What he describes and what is being solved in that paper are not consistent.

The point is that the objective and what is done do not coexist. The article claims many tiny supplies will be more efficient than one main supply. That just does not make sense. Standardizing to one inputs voltage on blade servers makes sense for other and different reasons. I suspect the author has confused the solution with what that solution is trying to solve.