SPCR does IDF, Fall 2002

The Silent Front

September 15, 2002 -- by Mike Chin

SPCR's first participation at the Intel Developer Forum, September 9-12, 2002, at the San Jose Convention Center was a one-man affair by yours sincerely. Naturally, I was run off my feet trying to see and partake in everything of interest to silent computing, finding and shmoozing the right contacts in the industry, and visiting companies in the San Jose area outside of IDF. Combined with the daily 3-hour round trip commute from San Francisco where I was staying with family, they were long days that left little time for either writing or sleep. I chose to skip the long commute for the Thursday morning wrap-up to begin this summary report.

IDF is an industry training event for Intel, with much of the time devoted to classes, some of which are highly technical sessions of interest only to specialists in the area. It is also a marketing event where hardware and software companies show off their latest efforts incorporating the newest technologies. From a big picture perspective, it is also one of the many ways by which Intel tries to maintain its hegemony in the computing world.

The online tech press was strongly represented. Many of you have probably read the daily IDF reports posted this week by big tech news and info sites such as Tom's Hardware Guide, Anandtech, and The Inquirer, to name just a few. Rather than duplicate the efforts of these large sites, I chose to focus solely on issues related to silent / quiet computing.

There were only 2 IDF classes of direct interest, Fire and Ice: thermal management for small desktop systems, and What's That Noise? a system design approach to making your system quiet. These sessions presented few approaches that would be revolutionary or new to dedicated SPCR readers. However, some of the conceptual language and terminology is useful, and some specific hardware implementations show promise. The most promising aspect of these classes is the prospect of new hardware specifications and form factors that may result in a greater range and number of quieter PCs in the marketplace. (The specifics presented in these sessions will be covered in a later article. Briefly, new small form factor designs, revised case airflow models, ducting, decoupled mounting, and fan design optimization are among the specifics.)

Meetings at the IDF with key people from Intel, Microsoft, Analog Devices (ADI), Seasonic, Molex, and Thermacore proved productive. Offsite visits to two other companies in the Silicon Valley area, Antec and Silicon Valley Compucycle (SVC), were also well worthwhile.

While low noise PC noise is far from center stage for most of the companies at IDF, there are encouraging pockets of strong interest and initiative among a handful of companies, and within Intel and Microsoft. While Intel and Microsoft appear like monolithic juggernauts to casual outside observers, among the many hundreds of departments in these PC industry giants, various groups with divergent interests vie for influence over direction and priority. It seems that pro-quiet segments are active behind the scenes at both Intel and Microsoft.

Low Noise is Essential for the Entertainment PC

The single most important motivation behind Intel's and Microsoft's growing interest in quieter computers is their desire to move the PC into the living room as the heart of a home entertainment system, with new interactive control and communications devices, many of which were on prominent display in the IDF technology showcase. PC-as-entertainment-center was a key aspect of the Convergence theme of this IDF. The simple fact is that existing home entertainment products such as stereos, TVs and home theater systems are generally silent except for the sound produced through speakers when playing music or soundtracks. Consumers will expect similar performance from any PC in the living room. The typical 35-55 dBA noise of most mainstream PCs will not be accepted easily in most living rooms. With growth from traditional market sectors crawling at single-digit rates, this new role for the PC is one space that holds potential for real growth.

In a self-congratulatory aside, it was gratifying to hear many individuals from large companies express surprise that silentpcreview.com is only 5 months old. They seem genuinely delighted about the work we are doing, and appreciate our efforts at depth and comprehensiveness in our reviews, articles, and project pieces. Many offered support to help keep Silent PC Review growing. Our great thanks to all of those folks!

ADI acoustic-thermal management in new Intel motherboards

Analog Devices (ADI), one of the world's biggest chip makers and a gold sponsor at IDF, may have been mentioned by SPCR in conjunction with their dBCool ADM1027 IC chip. We have been in touch with ADI since early summer regarding ADM1027 and a working engineering sample of a PCB utilizing the chip that they provided some weeks ago. According to ADI,

ADM1027 is a complete systems monitor and multiple fan controller that can monitor the temperature of up to 2 remote sensor diodes, plus its own internal temperature. It can measure the speed of up to 4 fans and control their speed so that they operate at the lowest possible speed for minimum acoustic noise. Measured values can be read out via a serial System Management Bus, and values for limit comparisons can be programmed in over the same serial bus.

Paul Errico, marketing manager of ADI's Thermal and System Management Products, was instrumental in sponsoring SPCR's Summer 2002 Survey on PC Noise. A major challenge for ADI is to convince major OEMs such as Dell, HP, and IBM of the marketability of low noise systems using the ADM1027. With the current state of the world economy, most PC makers virtually seek proof of profitability before embarking in any new direction. On the component manufacturing end, Taiwanese motherboard makers whose products sell on the street in the US$80 range are completely dubious about embedding a $3.75 chip without seeing real demand for sophisticated thermal / noise control from PC OEMs.

Hence the primary question of our survey, How much more would you pay for a quiet PC? An aside: the answer by over 1400 respondents was, the quieter the machine, the more we are willing to pay. The most surprising result: over 17% would pay a surcharge of US$500 or more for a silent or inaudible PC! More on this in another article.

The exciting news is that the chip is embedded in a new series of Intel motherboards which are due to be released in about a month. A product marketing manager at Intel has promised to provide SPCR with a final pre-production review sample. This Intel document provides reasons for implementing the extensive active hardware monitoring and control on these motherboards. Look for a SPCR review in coming weeks.

Newer versions of the 1027 were also announced: the ADT7460 and ADT7463. The fundamental improvement is that once a thermal target for a device is chosen, the chip takes care of controlling fan speed(s) to maintain that target. There is no need for users to consider fan speeds or airflow as these factors simply become part of the feedback loop. All that is really needed is for the user to determine his level of comfort about how hot the components are allowed to run. This is the usual noise vs. heat decision silent PC modders face. ADI aims to persuade motherboard makers and major OEMs to utilize the new chips, especially for entertainment PCs.

Noise Standards: Intel and Microsoft

One of the questions raised by Tomas Risberg of The Silent PC is why Microsoft no longer includes any specific noise guidelines desktop systems. Tomas notes,

In the early months of 2002, but possibly as early as in 2001, to about May 2002 did Microsoft talk on "Designing a consumer desktop PC optimized for Microsoft Windows XP" and they made this statement:

  • The target for the declared sound power level of the PC should be 37 dBA in the sleep state and idle mode, but no more than 50 dBA. The target level should be less than 55 dBA in active modes. The sound power level must be measured according to ISO 7779 and reported according to ISO 9296...

Comment: Now is it impossible to find any statements on the IT noise issue at Microsoft's Web sites.

Intel also does not provide specific guidelines about total desktop PC system noise.

In separate conversations with Intel's Greg Schlecter, a technical marketing engineer involved in new small form-factor desktop projects, an Intel product marketing engineer who preferred not to be identified, as well as Jason Anderson, a Microsoft technical evangelist focused on Windows hardware platform strategy, I learned the following:

  • TFX12V (Thin Form Factor) power supplies for Intel's small form-factor programs code-named Tidewater and Big Water have a target noise level of 3.8 Bels. This specification is too high for a truly quiet PC, but a start, at least. Tidewater and Big Water represent Intel's current R&D about thermal management in small PCs, including airflow, fans, heatsinks, layout geometry, vents, ducts, etc.
  • Because PCs are used in a variety of applications, a single noise recommendation may be unrealistic. Simply stating a target noise level without demonstrating how it can be achieved would upset Intel's hardware partners. Again, the Tidewater and Big Water projects represent part of Intel's efforts to show how lower noise PCs can be built. When asked, Why not a noise recommendation for entertainment PCs, at least?, the answer was that while this option is also being reviewed, it is important to establish what can be realistically achieved without pushing cost too high.
  • Microsoft's recommended noise target referred to by Tomas Risberg above was originally published on the basis of Intel initiatives. The recommendation was removed after further review within Microsoft deemed it too conservative. Currently Microsoft is working closely with Intel and other industry partners to study what is feasible and to establish realistic noise guidelines.

Both Microsoft and Intel admit that SPCR and other silence-oriented modders on the web are leading in the area of PC noise reduction. This is partly because we modders are unconstrained with standards / safety adherence issues or the inertia that must be overcome to make changes in large companies. But I believe the main reason is the low priority that noise has been given thus far. Until the recent push to move the PC into the living room, there has been little in the way of pecuniary motivation to create a quieter PC. Profit and revenue drives the PC industry like any other.

Heatpipes in the PC Mainstream?

Thermacore International Inc has been designing advanced thermal solutions since 1970, but is virtually invisible to PC consumers because of their almost exclusive focus on industrial and OEM applications, especially in notebook PCs, where many Thermacore heatpipes can be found. Thermacore have a great deal of experience and expertise with heatpipes, a technology that is the subject of growing interest in the PC world in recent months. The heatpipe CPU cooler used in the popular SSF Shuttle SS40G is actually manufactured by Thermacore.

The technical details of heatpipe technology is well explained on the Thermacore website; suffice it to say that its most salient property for CPU cooling in desktops is the ability to move the heat efficiently away from the source. This means, for example, that the "condensor" portion of the heatpipe can be located on the outside of the case, much like radiators in water-cooled systems; with a large enough condensor, it may be possible to use just air convection even with hot CPUs.

Thermacore's Gregg Baldassame and Ellen Cruse explained that while custom heatpipes can be made to solve almost any specific thermal problem, it is very difficult to make one that is adaptable for a wide variety of form factors. A heatsink for socket A can be easily made to fit virtually all socket A motherboards, in a wide variety of case form factors. Ditto a heatsink for socket 478. This is not true for a heatpipe cooler; the aforementioned Shuttle heatsink, for example, was custom designed for that particular motherboard in that particular case. It may be possible to adapt it for other motherboards in other cases, but there is no guarantee. This has to do with the rigidity of the copper tubing needed for heatpipes.

This is why it is unlikely a quiet, effective, modestly priced, generic cooling solution using heatpipes will appear on the component market any time soon. On the other hand, custom heatpipe solutions for barebone systems like the Shuttle or for complete systems from majors may appear in greater numbers soon. Here is exciting news: Thermacore is currently working with one of the majors (Dell, HP/Compaq, IBM, NEC, Gateway, etc...) on a custom heatpipe cooling system that may allow fanless CPU cooling! Whether this project comes to fruition as a line of silent PCs available to anyone remains to be seen. Look for more info on this topic at SPCR.

Intel Innovative PC Award Winners: The SPCR perspective

At every IDF since 1999, Intel has recognized PC designs that measurably improve setup, usability, expansion and maintenance as well as innovative design, new features and ease of use. The 2002 fall IDF winners are shown in full frontal glory on this page, and detailed on many other hardware sites. From the SPCR point of view, the back ends of the desktop PCs are much more interesting.

The photos below show the back panels of the Gateway 700XL and the Legend Tianqui 9220. Please examine them carefully. Notice anything different?

Neither of the power supply air exhaust grills have a fan behind them! That's right, there is no fan on the PSU back panels. In fact, there is no fan mounted anywhere on the back panels of these PCs.

Before you get too excited, I have to tell you that these are not fanless power supplies. In both, a fan appears to be mounted on the bottom panel of the PSU case -- that is, the panel directly over the CPU. This is the normal location of the second inner fan in most 2-fan PSUs like the Enermax or Antec units. Warm air could be felt flowing out of both PSUs. Seasonic's Vincent Chang, who I met with at IDF, informed me that this fan position was not new, that it had been the original ATX specification some years ago, at least briefly.

Another notable feature of these PCs is the openness of the PSU air exhaust grill. The square-holed grill on the Gateway is obviously more open than the more conventional rectangular slots on the Legend, but both PSUs have far larger vent exhaust area than on any PSU I have seen. This is similar to the case airflow venting in a Dell SFF system that was used to power a PC-TV demo setup.

These are the conjectures that came to mind about the air flow configuration in these PCs:

  • Placed deeper in the interior of the case, the fan is less audible.
  • The fan provides better cooling for the CPU by drawing air over it, by removing the hot air around the CPU more efficiently.
  • The more open, large air vent area minimizes turbulence noise and maximizes cooling airflow.

Company reps could not be found to confirm or correct these conjectures; technical departments for Gateway and Legend will be contacted in the near future to obtain more information about the above.

* * * * *

There is much more material from IDF and San Jose to cover. Alas, I have a life to live, family and friends to spend time with, paying work to do, and my own sanity and health to consider. Please look for part two of IDF coverage in the coming week.

* * * * *

Click here for Part Two of SPCR's IDF Fall 2002 coverage.

Discuss this article in our Forums.

Help support this site, buy from one of our affiliate retailers!