Faking the fan rotor lock signal in a Tripp Lite SU1000RTXLCD2U UPS
For Reasons1, I needed2 to connect my refrigerator to a double-conversion UPS so it receives good, clean power regardless of what’s coming from the wall outlet. In my case, this power could be from the local electric utility or from an onsite natural gas generator.
I found a good price on a used Tripp Lite SU1000RTXLCD2U UPS on eBay.3 I failed to consider that anything designed to be mounted in a rack will certainly have a much louder fan that is acceptable in a home kitchen.
When I first powered the UPS up and noticed the fan noise, I wrote: “I’m assuming [swapping the fan for a quiet one] won’t actually be too bad.” lolsob. I chose a Noctua NF-R8 1800 RPM fan as a replacement. This fan doesn’t move as much air as quickly as the original, but given that the load on this UPS is very low — over the past several weeks, I never see it go above 20% load, and typically it’s around 4% — I don’t think this is a problem.
In addition to a large battery, a UPS contains capacitors that can store a lot of energy and discharge it very quickly and circuitry that produces mains voltage from the aforementioned battery. Danger is present even if the UPS is unplugged and turned off. You should only open a UPS if you completely understand these hazards and know what you're doing.
This post is not a tutorial and it does not cover anything you need to know to work safely on such a system.
Once the fan arrived I opened the UPS up and swapped the fan in. (This required swapping the connector from the old fan over to the new one.)
At first I only connected power to the fan and didn’t connect the sensor wire at all. The UPS didn’t like that; it shut down right away with the message “FAN FAULT.” Fair enough. I tried connecting the new fan’s sensor output directly to the UPS’s fan sensor wire, but I got the same result.
After some research I found that the original UPS fan has a rotor lock output signal, which is common on UPSs and network gear, while the replacement Noctua fan has some sort of speed sensor output signal, typical for PC fans. The rotor lock signal should look to the UPS like an open collector when the fan is not spinning, and should be pulled to ground when the fan is operating properly.
I started planning a circuit to adapt the fan speed signal into a rotor lock signal, but despite trying different combinations of voltages and pull-up resistors I couldn’t make sense of the signal coming from the Noctua fan. It just didn’t match their published specs. I also found a Reddit post from someone else complaining of the same thing: a fan speed signal that didn’t at all match what Noctua says it should look like4. After reading that, I gave up and decided that just faking the rotor lock signal would be good enough; the fan will probably keep running fine, and anyway I’m placing so little load on this UPS it probably barely needs a fan at all.
I tried a couple quick & easy workarounds from various threads online: just wiring the fan rotor lock signal to ground, grounding it through a capacitor, grounding it only through resistors of various values. None of them worked; the UPS is too smart. I think it checks not only that the rotor lock signal is valid when the fan is supposed to be running, it also checks that the rotor lock line isn’t grounded before powering up the fan.
So grudgingly (though impressed with the UPS’s engineering) I accepted that I needed something that looks like an open collector when power isn’t applied and is pulled to ground once power is applied. Readers, that’s just an NPN transistor.
My final solution — a BC337 NPN transistor, alongside a voltage divider to keep the base-emitter voltage ≤5V — is illustrated in this schematic:
After assembling the circuit on a tiny piece of protoboard and testing it, I encased the whole thing in epoxy putty, making sure that nothing on the board can move or short-circuit on contact with anything else. I don’t have photos of the final installation, but I zip-tied the epoxy-encased board and wiring in place securely alongside the fan. This solution has been operating without issue for weeks now.
My house’s backup generator, discussed earlier here, produces relatively dirty power that seems to kill the fridge’s built-in ice maker. ↩
Well, “needed.” ↩
This is not a recommendation for this particular UPS; it’s just what was available used when I was looking. ↩
I don’t have any scope screenshots, but all I could get from the fan’s speed sensor were some weak — like, 50mV — and very brief pulses whose period depended on the fan speed. Not at all the nice waveform in the “datasheet.” Perhaps this is just what you get with a 3-pin fan instead of a 4-pin fan? ↩