[LINK] Climate Savers Computing Org

[Geoffrey Ramadan]

Stephen Loosley wrote:
> At 10:41 AM 14/06/2007, Stewart wrote:
>
>   
>>> an average PC wastes about half of the power it consumes.."
>>> <http://www.climatesaverscomputing.org/>
>>>       
>> This is one of those silly techno-economic statements where the
>> figures you reveal rest totally on the assumptions you make.
>>     
>
> Agreed in principle, Stewart .. but many computers are business tools,
> not just consumer items, and DO waste significant amounts of energy:
>
> ** High-efficiency power supplies for computers and servers **
>
>     <http://labs.google.com/papers.html>
>
> Most likely, the computer you’re using wastes 30-40% of the electrical 
> power it consumes because it is using an inefficient power supply. It’s
> difficult to believe that something as basic as a power supply could be
> responsible for that amount of waste, but it’s true. 
>
> The problem with power supplies is that they generate heat, which saps
> away energy meant to power the computer. That happens when the power 
> supply converts AC current into the DC current needed by computers.
>   
I am not sure I would put it this way. The power supply consumes power 
as well as delivering power to the attached circuit.

You design power supplies to typically give a regulated voltage at some 
rated current. You also typically rate power supplies by what they 
deliver or Output Power. A 100W power supply delivers 100W to the 
attached circuit... and would in itself consume say 30W, so in total the 
systems consumes 130W.

So an inefficient power supply does not sap energy away from the 
computer, but adds to it as wasted heat/energy.

> At Google, we run many computers in our data centers to serve your 
> queries, so our energy conservation and efficiency are important to us. 
>
> For several years we’ve been developing more efficient power supplies
> to eliminate waste from power supplies. Instead of the typical efficiencies
> of 60-70%, our servers’ power supplies now run at 90% efficiency or better,
> cutting down the energy losses by a factor of four.
>
> We believe this energy-saving power supply technology can be applied to 
> home computers, too. So we’ve been working with Intel and other partners
> to propose a new power supply standard. 
>
> The opportunity for savings is immense ­ we estimate that if deployed in
> 100 million PCs running for an average of eight hours per day, this new
> standard would save 40 billion kilowatt-hours over three years, or more
> than $5 billion at California’s energy rates.
>
> The technical changes we propose are very small and low-risk. 
>
> For reasons dating back to the original IBM PC in 1981, standard
> PC power supplies provide multiple output voltages, most of which are
> no longer used directly in today’s PCs. Back in 1981 the chips actually
> did need all these voltages. But  those times are long gone.
>
> Why then do power supplies continue to be built to produce multiple 
> voltages? The answer is simple: because the standard never changed,
> and because the actual voltage needs of many chips in a computer
> change every year as they become more energy efficient themselves. 
>
> But the changing voltage needs of chips are now met by voltage regulator
> modules (VRMs) that computer manufacturers put on their motherboards. 
>
> These VRMs take one of these voltages (say, 5V) and  transform
> them down to the actual voltage needed (say, 1.7V) making multiple 
> voltage output capability of power supplies unnecessary.
>
> Providing multiple output voltages complicates the design of power 
> supplies, and it makes it harder to build efficient power supplies.
>
> In essence, manufacturers have to build four different power supplies:
> one each for +12V, -12V, 5V, and  3.3V outputs, 4 power supplies in one.
>
> Because each motherboard may draw different amounts of power on each 
> voltage, manufacturers overprovision the supply for each individual voltage
> in order to support multiple options. Since power supplies are most efficient
> near their maximum rated loads, this overprovisioning leads to lower efficiency. 
>
> The VRMs (voltage regulator modules) used internally are also a significant
> source of loss. Typical current efficiencies (including power supply and VRM
> losses) are in the 55-60% range today, i.e., power supplies use 65-80% more
> power than necessary.
>
> Google servers, and the new PC standard we propose, use a simple 12V
> power supply. The power supply generates a single voltage, and all other 
> voltages required by motherboard components will be generated on the
> motherboard itself via VRMs.
>   
You would save more power by lowing the 12V to the minimum your PC needs 
probably 5V or 3.3V and use a step-up DC to DC converter for the 12V (I 
am assuming that the 12V power consumption is minimal and that most of 
the power is consumed by the 5/3.3V circuitry) as VRMs should operate 
more efficiency with lower input voltages.

Reg
Geoffrey Ramadan B.E.(Elec)
Unique Micro Design

> The net result of these changes is a dramatic improvement in efficiency 
> (including the power supply and the regulators) to about 85%, at virtually no
> cost. In other words, you won’t have to pay more for a higher-efficiency PC,
> because the power supply is actually getting simpler, not more complicated.
>   
> By spending another $20 or so extra, it is possible to use higher-quality
> components and achieve efficiencies well over 90%.
>
> You won’t be able to buy such computers for a while, and Google isn’t 
> planning on selling you any. But we’re working with industry partners such
> as Intel to make this technology an open standard that everyone can use,
> and that all vendors hopefully will adopt. It’s the right solution technically,
> and the right thing to do for the environment.
>
> If you'd like us to keep you posted on our progress, please send us a 
> note at efficient-psu at google.com.
> --
>
> Cheers, Stewart
> Stephen Loosley
> Victoria, Australia
> .
>
>
> _______________________________________________
> Link mailing list
> Link at mailman.anu.edu.au
> http://mailman.anu.edu.au/mailman/listinfo/link
>