Are you really bound to hear fan noise when operating a PC? Not by principle. Can an energy efficiency be allocated to the activity of a computer? Not by principle, but the energy consumption can be reduced. What does one of these have to do with the other? Perhaps more than you expect!
The Codex of Business Writing Software for Real-World Solutions 2.pptx
Fan noise
1. »Energy efficiency«
in data processing?
Stefan Fassbinder
Deutsches Kupferinstitut
Am Bonneshof 5
D-40474 Düsseldorf
Tel.: +49 211 4796-323
Fax: +49 211 4796-310
sfassbinder@kupferinstitut.de
stf@eurocopper.org
www.kupferinstitut.de
2. The German Copper Institute, DKI, is
the central information and advisory
service dealing with all uses of copper
and copper alloys.
We offer our services to:
Commercial companies We can be contacted by:
The skilled trades post
Industry phone
R & D institutes fax
Universities e-mail
Artists and craftsmen internet
Students online database, or
Private individuals personally
4. This brand new PC power supply
unit
at 39.95€
promises to be 60% less noisy! 4
5. Well, 60% less noise
is at least a statement!
Though not really any more than that …
… as long as the question remains
unanswered:
Less than what?
6. This was not the way
Now what to do?
Loss heat can be
• either blown away
• or avoided!
So a different solution had to be found:
A fanless PSU (at 79.50€) works
totally noise free!
7. How is this reduction in
noise level achieved?
By improvement of passive cooling or by
reduction of cooling need, say losses?
By improvement of passive cooling and by
reduction of losses:
• Substantial enhancement of cooling
surface on the aluminium heat sinks
• Substantial enhancement of copper cross
sections in the windings of HF converters
8. But loss reduction holds
the greater share!
140W
Starting up a PC with
120W subsequent idle operation
P
100W
80W
60W
40W
Original PSU Antec True 380 SP
20W Fanless PSU FSP ZEN (2009)
0W
00:00min 05:00min t 10:00min
9. How to reduce losses? – Use more material!
Conventional Fanless
420 W 300 W
10. Still, the actual net power
consumption of a common PC
(built in 2005, 800 MHz) is:
• 50 W when idling (whereas writing a text
or composing an Excel table is idle for a
PC!),
• 100 W at peak processor demand
(gaming),
while the common power supply unit is very
much over-dimensioned!
11. Therefore the 300 W model
can quite conveniently
replace a 420 W unit.
PC power Antec FSP ZEN
True 380 with active
supply unit
SP with PFC,
comparison active fanless
Power rating 380W 300W
Weight 1995g 2223g
Material intensity 5.250g/W 7.410g/W
Factor 100% 141%
If this were not so the copper content would
have to be 41% higher rather than just 11%
12. Who else should have an
interest in this topic?
This is the reply to a request to the
colleagues from www.aluinfo.de sent
March 31, 2009:
I‘m still waiting…
13. Conventional PC power
unit, 420W
PC power supply LC4620, S/N: 7J090402
Rated power 420W
Input
30.0A 28.0A 14.0A Output
P 3.3V 5.0V 12.0V P P /P N η
4.4W 0.0A 0.0A 0.0A 0.0W 0.0% 0.0%
13.5W 0.0A 0.0A 0.0A 0.0W 0.0% 0.0%
64.0W 0.0A 0.0A 3.3A 40.1W 9.5% 62.6%
71.0W 17.5A 0.0A 0.0A 57.8W 13.8% 81.3%
109.0W 0.0A 0.0A 6.7A 80.2W 19.1% 73.5%
148.0W 0.0A 0.0A 9.4A 113.2W 26.9% 76.5%
161.0W 0.0A 25.0A 0.0A 125.0W 29.8% 77.6%
286.0W 0.0A 0.0A 18.8A 226.0W 53.8% 79.0%
307.0W 28.5A 24.0A 0.0A 214.1W 51.0% 69.7%
450.0W 28.5A 25.0A 7.0A 303.1W 72.2% 67.3%
740.0W PSU failure after ≈2s with <20% overload
14. Conventional PC power
unit, 300W
PC power supply T.I.P. 300 ATX
Rated power 218W
Input
14.0A 20.0A 6.0A Output
P 3.3V 5.0V 12.0V P P /P N η
0.0W 0.0A 0.0A 0.0A 0.0W 0.0% 0.0%
3.1W 0.0A 0.0A 0.0A 0.0W 0.0% 0.0%
21.4W 0.0A 0.0A 0.5A 6.1W 2.8% 28.6%
43.3W 0.0A 2.2A 0.9A 21.7W 9.9% 50.0%
56.8W 0.0A 0.0A 3.2A 38.9W 17.8% 68.5%
70.0W 0.0A 0.0A 4.3A 51.4W 23.5% 73.4%
97.0W 0.0A 5.0A 3.4A 66.1W 30.3% 68.1%
122.0W 0.0A 5.0A 5.2A 87.2W 40.0% 71.5%
148.0W 5.8A 5.0A 5.2A 106.0W 48.6% 71.6%
181.0W 5.8A 9.5A 5.2A 129.0W 59.1% 71.3%
226.0W 14.0A 9.5A 5.3A 157.4W 72.1% 69.7%
Shut-off – no more measurements possible
15. Conventional PC power
unit, 380W
PC power supply Antec True 380 SP
Rated power 380W
Input
28.0A 35.0A 18.0A Output
P 3.3V 5.0V 12.0V P P /P N η
0.0W 0.0A 0.0A 0.0A 0.0W 0.0% 0.0%
4.4W Fanless PC power
0.0A 0.0A 0.0A 0.0W 0.0% 0.0%
28.3W 2.2A 0.0A 0.0A 7.1W 1.9% 25.1%
64.8W unit, 300W
2.5A 0.0A 0.9A 19.1W 5.0% 29.4%
90.0W 2.5A 4.0A 0.9A 39.1W 10.3% 43.4%
109.0W
148.0W
Rated efficiency:
6.1A
6.1A
4.4A
4.6A
0.9A
3.3A
53.3W
82.6W
14.0%
21.7%
48.9%
55.8%
174.0W 5.6A 9.3A 3.3A 105.0W 27.6% 60.3%
282.0W 89% (at full load)
27.0A 9.3A 3.4A 176.2W 46.4% 62.5%
381.0W 27.0A 17.0A 6.7A 254.5W 67.0% 66.8%
463.0W 27.0A 17.0A 12.9A 328.9W 86.6% 71.0%
561.0W 27.0A 17.0A 19.0A 402.1W 105.8% 71.7%
16. How high is the »efficiency« of a PC?
Much higher than earlier, anyway!
Ancient XT: Latest model:
f = 4,7 MHz f = 4,7 GHz
Pel ≈ 40 W Pel <<400 W
This looks like an efficiency improvement by far
more than a factor of 100! With similar values for
• storage units,
• hard disks,
• graphic adaptors
you end up with improvement factors around 106!
17. Where will the journey
still take us?
A limit will be reached only when for one data bit
only more one electron needs to be moved.
So long, however, it‘s still 10,000 to 100,000
electrons that need to be employed for the smallest
occurring »elementary process« in data processing.
So the theoretical potential
• with miniaturisation,
• with acceleration of procedures,
• with reduction of energy demand
still lies somewhere near a factor of 104 … 105!
18. Hence the major part
of the job is done
• The first computer »Zuse Z1« had the
dimensions of an electrical locomotive…
• …along with its power consumption…
• …and this with the performance of a credit
card sized pocket calculator which you get
as a giveaway nowadays and which is
supplied from 2 cm² of solar cells or with a
button cell battery for no less than a year!
19. But progress with
processors is continuing:
PC built upgraded
2005 2011
Processor 2.6 GHz, 2*2.7 GHz,
RAM 512 MB, 4 GB,
400 MHz 1.6 GHz
Idle power 59 W 45 W
measured with the same fanless PSU
20. Only what‘s the use of it all
if you do anything you can
think of in order to e. g. implement all
available efficiency techniques in a
modern diesel engine and then add a
gearbox with 30% of losses?
Nobody would do that in a car –
in a PC it‘s the usual approach!
For the PSU is sort of a
»gearbox for electricity«.
21. The noiseless operation
comes for free!
Calculation of power consumption costs for a PC (built in 2005)
before and after replacing the conventional PSU with a fanless one
Conventional PSU with fan Fanless PSU
Output values P S Q LF cosφ P S Q LF cosφ
Full load 100h/a 135.5W 138.9VA 51.4var 0.975 0.977 113.0W 123.2VA 67.0var 0.890 0.940
No load 3000h/a 82.0W 87.1VA 29.3var 0.939 0.945 59.2W 70.5VA 38.1var 0.843 0.886
»off« 5660h/a 6.3W 24.5VA 23.6var 0.256 0.307 2.0W 24.1VA 24.0var 0.080 0.112
PSU 39.95 € 79.50 €
Price
Power 14.0c/kWh
Consumption 294.91kWh/a 200.36kWh/a
Electricity costs 41.29€/a 28.05€/a
Payback time 2.99a
The saved losses pay back for the price
premium within 3 years. The global savings
potential in 1 bn PCs would be 100 TWh/a!
Morals: I am a fan of the fanless design!
22. Then why do portable PCs provide
the same computing performance
with that much less electricity?
This notebook PC
is equipped with an integrated
55 Wh battery and an additional
95 Wh battery and runs for about
13 hours on these, corresponding to
12 W net DC power requirement!
23. Metering conditions:
• WLAN off,
• no UMTS device,
• no »action games«,
• moderate screen brightness.
Whereas it is remarkable with the laptop PC
• that the screen is already included
• and the whole PC uses less power …
24. … than a desktop flat screen uses
for itself alone!
25. So in the end there is one
question remaining open:
Why can this technique not be built
into a desktop PC?
Energy is still way too cheap.
The availability of electricity from the
accumulator is a reason to save
energy – the electricity price obviously
is not.