January 28, 2003 by Mike Chin
In the second of two reports about the Intel Developers Forum (IDF) last September, I mentioned some work being done by Intel on Small Form Factors, quiet PC, and fan design. Two specific sessions were mentioned. I was chagrined when someone pointed out recently that the additional information promised about these sessions were never delivered.
This follow-up is done in haste. The presentations for the two sessions mentioned in the above articles are located on the Intel Developers Forum website and can be downloaded as PDF documents directly via the following links:
Fire and Ice: Thermal Management for Small Desktop Systems (561 KB)
Acoustics: "What's that noise?" ¬ó A system design approach to making your PC quiet (1347 KB)
You can also obtain any presentation from the Fall 2002 IDF from this complete listing of PDF format docs from the Fall 2002 IDF.
Another Intel document of interest to SPCR readers is
Developing Small Form-Factor MicroATX Systems by Greg Schlechter of Intel Desktop Products Group. This article relates to the Fire & Ice article linked above. It first appeared the June 2002 issue of the Intel Developer Update Magazine.
A document closely related to the SFF MicroATX Systems document above is TFX12V Power Supply Design Guide version 1.01, downloadable as a PDF in the linked page at the Desktop Form Factors (http://www.formfactors.org/) web site.
ABOUT FIRE & ICE
Prepared by Michele Szleper of Intel's Desktop Platforms Group, the subtitle of the piece tells all. To be frank, there is little here hardcore SPCR enthusiasts haven't already tried or at least thought about. The overall conceptual approach is not that unique, either. I suppose given the narrow boundaries of the problem and the range of practical tools available to anyone, this is not unexpected. But it is encapsulated nicely in technical terminology with nice illustrations and pictures. A few comments on specifics that are noteworthy.
Venting, p.6: We all know fan grills block air, but this is the first time I've ever seen figures on how much. The worst one here, which passes only 56% of the airflow, is better looking than the vast majority of vents I see in cases and PSUs. I guess this vindicates my instinctive, cut-and-slash approach to all grills to date.
Ducting, p.11-12: Some creative ductwork here, including splitting airflow to better share the cooling. One has to wonder about the effect on turbulence and noise. Does anyone know of a cheap plastic molding machine?
Next Generation Fan, p.18-19: This is the most interesting part of the doc, IMO. The fan blades are larger and there is no frame. It is not clear what Intel is shipping with their hottest P4 processors these days. One wonders if this design has found its way into production. Can anyone with a recent retail P4-3.06 tell us in the forums?
ABOUT ACOUSTICS: What's that Noise?
This one was put together by Les Fisher of the UK for Intel's European Application Design-in Center. It is at the heart of SPCR's interests. As with Fire & Ice, core approaches and concepts are not new for hardcore SPCR enthusiasts, but some of the details may be, and the photos, charts and graphs are definitely of interest. The actual session presentation (in Sept at the IDF) was incorrectly booked for just one hour instead of the 2.5 hours that it was intended for, so a lot of the material did not get covered. I am thus unable to shed much additional light on the contents of the presentation.
The hemispherical sound power setup photo on p.8 is interesting to anyone who hasn't seen the details of how this is done, as is the ISO 7779 derived operator-position sound measurement setup on page 9.
PC noise has stayed constant from CPU clock speeds of 300 MHz and up, according to a chart on page 10. This is a curious contradiction to Dan Quinlan's article, Computer Noise in the 21st Century, which predicted in 1999, a 10 dB rise in PC noise in the coming decade.
Use multiple fans for low impedance systems and a single fan for high impedance systems is the message of page 16, advice I don't quite understand.
Innovative blade design of Intel's new fan is touched on again on page 17. Note that fan, as shown in the extracted page above, has 8 blades. Someone at the session asked why it was an even number; apparently, an odd number tends to minimize harmonic noise generated by the spinning blades. (Again, if this fan is in production as a HSF for a high end P4, and someone has one, please count the number of blades and tell us! I will also inquire Intel contacts directly about this fan.)
While all of these various endeavors by Intel are encouraging, one can't help feeling their noise targets are probably not very ambitious, at least not by SPCR standards. The Acoustics... presentation describes ambient noise levels thus:
- Typical Office: 35-45 dBA SPL
- Typical Home: 30-35 dBA SPL
- Quiet Home: 20-25 dBA SPL
In the Intel-sourced TFX12V Power Supply Design Guide, the acoustic target for the PSU is described thus:
"4.7 Acoustics - Sound Power: The power supply assembly shall not produce a declared sound power level greater than 3.8 BA... at 43C, 50% of maximum rated load."
Even generously translated, 3.8 Bels (A-weighted) is not much better than about 30 dBA @ 1 meter SPL. It is a target many SPCR members have surpassed with their entire systems, using quiet commercially available components.
Hopefully, all this is good food for thought to encourage more thinking, creativity and experimentation that will surface in the forums and in new article submissions to SPCR!
* * * * *
Comment on this article in the Forums.
|Help support this site, buy from one of our affiliate retailers!|