I want to explore the notion of an "ideal add-on fan-controller" in addition to what's been said here.
First, it should depend on your choice of an airflow strategy within the case, and the number of fans you intend to use.
So let's lay a groundwork. Assume you use a tower-heatpipe cooler which sits in front of an exhaust fan at the case rear. Assume you wish to pressurize the case. Assume that you want to minimize the number of necessary fans (total), minimize the power draw of fans, and maximize the CFM in overall case airflow. There are tradeoffs among these objectives.
To me, the ideal fan-controller simply extends motherboard fan control and allows you to deploy beefier fans with extra wattage. The ideal fan-controller provides "thermally-adaptive dynamic control" of fan-speeds (and thus CFM), and we assume our motherboard provides the same, limited by (a) overall or total amperage @12V for fans connected to the motherboard, and (b) the maximum limit of amperage per motherboard fan-plug.
What we discover with midrange motherboards: You may be able to control CPU_FAN and CHA_FAN1 through BIOS settings for "thermally-adaptive dynamic" control, but the other motherboard plugs may fall short. They may fall short because you've extended the total amperage beyond either limit, or they may fall short because the BIOS simply does not provide thermally-adaptive control for all the fan-plugs.
You really don't want manually-adjustable fan-control through rheostat knobs. Further, you would like to be able to use CPU and motherboard sensors, as opposed to using "stick-on" sensor-wires provided with a great many fan-controllers.
In order to reach this last objective, you need a controller that can read the onboard sensors, and to do that, it should likely connect to the system via USB. If the controller contains its own microprocessor, you would want to set fan profiles and temperature parameters through software in the OS. Once those profiles were defined, the controller would retain the information and it would continue to control the fans based on the CPU and motherboard sensors as well as any other sensors you care to use.
nVidia attempted to push the envelope in this direction with a feature called "ESA." Silverstone developed and marketed the "Silverstone Commander" which was "ESA-certified," and the notion of "ESA-certified" was defined by nVidia: "certified" meant that you had an nVidia motherboard and chipset.
However, the Commander was released with a certain version of firmware that did not allow creating "profile curves" of temperature versus fan-speed. You could "create" such a profile but you could not save it and make it operable. Silverstone promised a "firmware upgrade" in the works, but never followed through. This, of course, occurred at the time of the great nVidia-Intel feud. So you could only control the static speed of fans (and set them at static speeds) through the nVidia software (downloadable from the nVidia web-site and still available there). That is, I could connect a fan with top speed of 2,000 rpm, set the ESA software to run the fan at 75% to achieve a speed of 1,500 rpm. Although I could do this for each and every fan connected to the Commander, I could still only set static speeds, and there is no thermally-adaptive dynamic control of fans.
The second thing the Commander demonstrates is that nVidia and Silverstone were inept in their marketing. The Commander and the nVidia software also work with Intel chipsets, even if those chipsets and motherboards are not "ESA-certified." They could have sold more product just for giving up on that as a "requirement," because in effect -- it wasn't a "real requirement."
You could still use the Commander (USB-connected) to set static fan-speeds without having a front-panel with manual adjustment knobs. That, by itself, is an improvement. And once you had done this, you wouldn't need to run the nVidia software on the Intel motherboard continually, since the controller would simply remember the profiles you created.
Another product I've looked at -- really several products that integrate and connect to each other -- is the T-balancer Big-NG. It comes with both digital and analog sensors -- the latter being those we're more used to seeing. The digital sensors are little tiny circuit-boards which you would attach to the item being monitored in the same way. And you can chain and add these sensors -- even on the same wire. The Big-NG requires a USB connection and comes with its "T-Balancer" software -- allowing "curve profiles" for dynamic fan control.
However, the T-Balancer needs another monitoring program to read the CPU and motherboard sensors and use them in the profiles. The instructions tell you to use "Motherboard Monitor" or MBM. MBM is a shareware software extant five or six years ago. The author stopped updating it around that time. It won't install successfully on Windows 7, even as it was supposed to be XP-64 compatible.
So in order to use T-Balancer and Big-NG, you need to install a sensor near your CPU and calibrate it. Lots of tedium and trouble. I won't go into the fine machine-work needed to place a sensor at the Intel-spec "TCASE" location between the cooler-base and th CPU IHS.
Maybe the Aquaero or some other manufacture/model overcomes these limitations. All I can say to add further to the discussion: nVidia ESA won't handle fan amperages exceeding 0.50A. The Big-NG allows for amperages up to 1.66A under "analog" control, and closer to 3.0A under PWM "digital" control -- for each of four fan-control channels. You can connect fans in parallel up to the amperage limit.
A final point. "Choice of fans." Many people get around the tradeoff between noise and cooling in their choice of highly efficient, low rpm and CFM-limited fans. For this approach, you'd hardly need a fan-controller if you planned to connect the fans directly to the PSU.
I -- on the other hand -- like the idea of using "beefy" fans which I've found to run at low noise in their lower rpm ranges, which lack "bearing noise" or rattle at their top-end speeds, so that the noise is just "white-noise" from air-turbulence. You can muffle fans; use rubber fan-mounts; shroud the fan in foam or rubber material, and deaden the noise reflections that occur within a metal case.
That way, I'd be able to dynamically control fan speed by CPU temperature so that high CPU load would be the only situation generatiing more noise.
Facebook
Twitter