Eaton 9SX3000I UPS Fan Replacement

Hardware

• x2 80mm Fans - Sanyo Denki 9A0812G4D01
• x1 Eaton 9SX3000I UPS

The UPS I recently acquired had some loud (~45 dB) fans, since the UPS is going to be running my study I wanted something quieter. The stock fans were loud enough that you could easily hear them elsewhere in the house with the study door closed.

I wanted to replace these with some Noctua fans which are known for being quiet. I chose NF-A8 FLX since they seemed to be a good match for the specs of the Sanyo Denki fans just significantly quieter.

Swapping Fans

Replacing the fans was not as straight forward as I hoped, the pin pitch on the UPS' PCB header was different to the standard fan header provided on the Noctua fans. Additionally the pin order was different.

Initially I tried popping the pins out of the Noctua header to fit the Sanyo Denki header but found that the female crimp connection on the cable was the wrong size.

Instead I plugged the Noctua into the provided extension cable, chopped the female header off and soldered the individual wires to the corresponding pins on the original fan header.

The fans worked great, much quieter than the originals, however the UPS detected a fan fault.

Resolving the fan fault

Disclaimer: I don’t really know what I am doing, this worked for me and has not yet caused a fire. Use at your own risk.

You may have noticed that the Sanyo Denki fans have a rotor lock sensor, this sensor tells the UPS if the fans have stopped spinning. The Noctua fans do not have this sensor.

Some reddit threads suggested that this fan fault could be “resolved” by placing a non-polar capacitor on the sensor/signal/yellow wire and DC-/black wire. I gave this a try but it made no difference to the fan fault.

With my multimeter I took a few readings:

• Sanyo Denki signal wire voltage - 0.125v
• Noctua signal wire voltage - .41v

Now the datasheet for the rotor sensor seemed a bit light on details but did seem to indicate that the voltage would not normally be constant. It should briefly go “high” when the fan first starts, but whilst the fan is running it should stay lowish (but not zero.)

So the problem boils down into converting the 0.41v signal into a 0.125v signal. Now its likely that the Noctua signal voltage is a bit more transient than a static 0.41v and might vary depending on UPS load. However this has been good enough for me so far (YMMV).

We can get the right voltage by creating a voltage divider with two resistors. I used this calculator to determine what resistors I needed. I chose 220 and 100 Ohm.

First I tested that the design would work using a breadboard and then implemented it. I used the signal wire as the voltage input (v1) (fan side) and the PCB side as the output.

Note that the UPS did not detect a fault with this fix applied to a single fan. I did not apply it to both.

Please let me know in the comments below if you spot any problems or can suggest any improvements.