$2.50 eBay fan controller / Zalman Fanmate schematic

So I wanted to adjust a fan down to 3.3 volt, the fan controller only allows 5-12V. Replacement of R6 was needed:

To understand this circuit, know that any voltage higher than 2.500 volt on the Ref-pin of the Tl431 will lower the zener voltage, and any voltage lower than 2.500 volt will adjust the zener voltage to a higher value:

So what happens if you connect Ref directly to the + of the output? If the voltage on the output is 6V, the Tl431 will adjust its zener voltage down, lowering the voltage on the base of the NPN-transistor lowered, hereby shutting down the transistor. This means that the voltage on the PNP-transistor will be higher because it’s not connected to ground anymore, and because the voltage on the base of a PNP-transistor has to be higher than the voltage on the emitter to allow current to flow, the transistor will close itself.

Now there’s no way to supply current to the output of the circuit, and the voltage will be lowered until the circuit has reached equilibrium at a lower voltage. If we want to get a higher output voltage, we have to lower the feedback coming from the emitter of the PNP-transistor; this is what R6 does. So you lower R6 (a 470 Ohm resistor did the trick) and you get a lower output voltage, which is what I was after: R7 probably has to be adjusted as well to get the maximum output voltage to 12V, but why use a fan controller if you want the fan at >10V?

DIY silent air filter for pollen, dust mites, other allergens and dust

If you own glossy black equipment, you’re probably aware that even after a mere day of not dusting, a thin layer of dust has accumulated on top of your appliance. According to dust.com, this domestic dust mainly comes from cloth (fibers), pollen and dead skin cells, although they don’t cite a source for that claim.

If you’re part of the 55% of the population that webmd.com claims has allergies (source) you probably also know that many allergens circulate in the air around us, causing discomfort and millions of dollars in lost productivity.

So what to do about it? Well, we could filter it out using a commercial air purifier, but those aren’t really quiet enough for day- and night use in a home. This quote is taken from a review of the $70 Hamilton Air Cleaner that comes up as a top result on Amazon:

And it isn’t silent on the lowest setting, unlike some reviewers said. The noise is similar to a fan on low speed or a refrigerator.

So obviously we have to make one ourselves. Here it is:

This slideshow requires JavaScript.

It’s one of the top fans from silentpcreview.com combined with a vacuum cleaner filter, attached with a zip-tie around the fan, with a 6V phone charger as power supply.  The filter has gotten pretty grey from all the dust that the filter has caught in just under 2 months. Cost is $9.99 for a new fan of this sort plus $0.50 for the filter and $5 for the adaptor, so you could make several of these for the price of one commercial unit. It’s also dead silent from more than 30 centimeters away.

The air flow of this Nexus fan at 5 volt is, according to Silent PC review here, 13 cubic feet per minute, or 22 cubic meter per hour. A room of 5*5*2.5 = 62.5 cubic meter should be completely filtered multiple times over in 24 hours: 1/3 of the air gets filtered in 1 hour, so in 6 hours only 0.66^6 * 100% = 8% of the dust is left over, and after 10 hours 1%. This is of course assuming that the fan randomly samples air from the room, and doesn’t only recirculate the air around it. The smallest particles that the filter catches are, according to the label, 0,3 micrometers. That should provide adequate filtering according to this graph, courtesy of air-purifiers-america.com:

Now what to do about the aesthetics of this device? It doesn’t look like something you would want in your bedroom, but an enclosure would function as a sounboard and would amplify the noise.

Thoughts?

Converting a PWM-signal to a DC-signal for powering the fan on your aftermarket GPU-cooler

After replacing the stock cooler on my graphics card (Replaced the stock 9600GT cooler with a Arctic Cooling Accelero S1 rev.2), i quickly found out that it needed a fan under load, as it’d otherwise crash when the VRM’s reached upwards of 100 degrees. But without running a 3D game, there’s really no need to have a fan running. So, what we’re looking for is a semi-passive cooling solution, driven by the GPU temperature.

The speed of the standard fan was regulated by a PWM signal coming from a chip on the PCB. Unfortunately, i couldn’t get the duty cycle below 20% in software, which still keeps all PWM fans running. To get better fan control, a PWM-DC converter was needed.

This is the schematic. First, the PWM signal is compared against a the voltage produced by R3 and R4, enabling the first opamp to output a clean 0V – 12V block wave, basically a cleaned-up version of the original PWM signal. This signal goes through a low-pass filter formed by R2 and C1, converting the duty cycle to a corresponding voltage (10% = 1,2 volts, 50% = 6 volt et cetera). The exact values of these components are not really important, just be sure the time constant of the filter (resistance * capacity) isn’t too small or too big, we’re looking for a RC value in the 1mS-1S range. The second opamp then drives Q1 with the voltage needed to put the exact same voltage on the fan as is outputted by the low-pass filter – if the voltage on R1 is not yet as high as on the + input of the opamp, the voltage is raised on the base of the transistor, allowing for more current flow and a higher voltage on R1 until the voltages are equal.

This is what the finished product looks like and how it’s attached to the video card. I decided against making a PCB, as a $5 breadboard works just fine. You can see that there’s 3 NPN transistors in parallel for driving the fan, as i did not have ones at hand that could dissipate all the heat.

The exact temperature-voltage chart can be made in MSI afterburner or any other program that lets you change the fan speed of your GPU (e.g. Rivatuner). If you’re planning on making a PWM->DC converter for yourself, know that any single-supply opamp is suitable, the only thing to look out for is the voltage swing of the ouput: my lm2902 isn’t rail-to-rail, so i can only regulate the output from 0V to 12 – 1.5 = 10.5 volts, which is suboptimal, but more then adequate for my the lowly 95W TDP – I only chose it because it’s a quad opamp for $0,20 and i had one lying around.

The finished product works really well! When just browsing or listening to music, the fan does not turn on, and when playing videogames or other intensive 3D applications, the fan turns on until the GPU gets cool enough.