What's new
What's new

Possible to Detect Fan Failure Electrically?

RC Mech

Stainless
Joined
Jul 21, 2014
Location
Ontario, Canada
Knowing little about wrangling electrons outside of programming and PLC's, what is an elegant solution to detecting a fan failure on a monitored circuit?

Mercedes, Audi etc. have a bulb detection system that measures resistance on the circuit to alert the driver of a burned bulb. Infinite resistance = warning light, and position of said bulb.

If an enclosed fan (plain-Jane 6" square, electronics cooling, 24VDC OR 120VAC are options) fails, I'd like to be able to have that as an input to a controller. The fan failure would be a burn-out like in the case of a bulb. Measure infinite resistance on the circuit? A method to measure current draw in real time and then program based on that input?

What's the sophisticated, elegant way of doing this?
 
Knowing little about wrangling electrons outside of programming and PLC's, what is an elegant solution to detecting a fan failure on a monitored circuit?

Mercedes, Audi etc. have a bulb detection system that measures resistance on the circuit to alert the driver of a burned bulb. Infinite resistance = warning light, and position of said bulb.

If an enclosed fan (plain-Jane 6" square, electronics cooling, 24VDC OR 120VAC are options) fails, I'd like to be able to have that as an input to a controller. The fan failure would be a burn-out like in the case of a bulb. Measure infinite resistance on the circuit? A method to measure current draw in real time and then program based on that input?

What's the sophisticated, elegant way of doing this?

Dunno what the "sophisticated, elegant" way is, but the brute-stoopid way the past 20-odd years is to pay attention to the third wire most of even the cheapest of Dee Cee box fans have had for about that long.

It's a very basic but good-enough tachogenerator speed feedback, raw pulse-train, usually, and when it goes away, you have a "situation".

Short ton of IC's to make use of it as have been on PCB's for two decades or so already, too. Ditto cheap printed-on-flexible-substrate thermistors with long leads meant to be trimmed intended to bump fan speed up when the case CPU, or such gets too warm. BIOS on planars, and OS'en that ride them have "syscon" or equivalent entries to read their
status.

All old, old stuff, by this late date, and sitting right in front of you if you were able to post, here, or ever heard a PeeCee, network shared storage device, printer, boxed hard drive, or laptop change its fan speed.

Where yah been?
 
I don't think most fan failures are caused by an open in the windings. I think you will find that it is usually the motor bearings lockup or at least get tight.

You might be able to sense the increase in winding current depending on particular fan but I'm not so sure if the current will actually increase enough to be easily detectable.

Fan motor winding failure is only one of the failure modes of a cooling fan. The more common and serious is dirt buildup and or fan obstruction.

I think the the best way to detect a failure would be to monitor temperature. Easier to implement and will cover all of the failure modes.
 
I've seen Fanuc drives monitor both the temperature and RPM of the fans. Anything out of spec sets an alarm.
 
The small fans these days are electronically commutated, as brushes are a reliability problem. This uses hall effect sensors to drive the stator windings, as the rotor has a permanent magnet. The hall effect switches control the timing of the pole switching of the stator coils.
As Monachist indicated, they sometimes bring out one of the hall effect outputs on a 3rd wire.
I've made a circuit in my former job that used a frequency to voltage converter IC, then an analog comparitor circuit that tripped a monitor output that a 'confuser' reads in, to report this to a diagnostic program running. There may be simpler ways, but this was aircraft avionics, and the fan was something that needed to be monitored. It was complicated by requiring the fan to operate at a slower speed, during a thermal heating event, and high speed once the temperature exceeded 25C. They were expensive EGG Roton fans. Bastards charged us $250 to bring out the 3rd wire in a standard product.
 
It's a very basic but good-enough tachogenerator speed feedback, and when it goes away, you have a "situation".

Short ton of IC's to make use of it as have been on PCB's for two decades or so already, too.

Where yah been?

Brilliant. I've only worked with these types of fans in 35YO+ electronics enclosures, so ASS-U-MED they all didn't have the tacho leg. That simplifies things greatly. I was thinking in terms of a load type monitor that would use current as a parameter for fan health, but I could plot a baseline of fan performance and simply extrapolate it to cover "fan degradation" as fluctuating or diminished revs in the software, assuming a reliable/dead-constant power supply.

I like Ziggy's idea of a temp monitor as well. The machine has many temp inputs as-is, what's another one... The below is a diagram of a bulb failure circuit that comes up in a Gizoogle search. Forget that they put an opto-coupler @ PC1- looks like my inclination that it was too convoluted for what I'm attempting was correct. Thank you both.
Bulb_Fail.jpg

Edit* Where have I been indeed! Thank you to Wes and ignator for your replies. Toying more with modern electronics offerings is now a priority in my futile attempt to become familiar with what's available.
 
My 1977 Datsun had a little box under the seat that sensed a burnt out bulb and illuminate an idiot light on the dash to show which one. It also had a sensor in one of the battery cells to indicate if the acid level was too low.
Bill D
 
The 1980's solution I saw on a HiAk servo drive was a big power resistor Above the drive electronics, with a thermal cutout switch attached to it.
If the fans ( and filters) were OK, it keep the power resistor and the thermal switch, cool, if not it alarmed out via the thermal switch.

So, if you include the filter, fan speed alone might not be enough.
Depends on your level of concern.
 
I'm not sure what your fan is cooling, but the IBM mainframes that I serviced in the 80s & 90s had thermal sensors above the airflow path. If things got too warm an error was indicated or the machine powered down.

Bob
WB8NQW
 
If an enclosed fan (plain-Jane 6" square, electronics cooling, 24VDC OR 120VAC are options) fails, I'd like to be able to have that as an input to a controller. The fan failure would be a burn-out like in the case of a bulb. Measure infinite resistance on the circuit? A method to measure current draw in real time and then program based on that input?

What's the sophisticated, elegant way of doing this?

A modern way is to drive the fan with a PWM signal and monitor the input current. Modern PC fans use internal feedback to the motherboard. You could us a IR emitter and detector across the blades on a old two-wire fan.
 
Monarchist has stated the solution. When the tach output goes away the fan is no longer spinning. Probably the simplest circuit would be to use a watchdog/reset IC. These are normally fed continuous pulses from a microprocessor and if the pulses stop it is assumed the processor is "hung" and a reset is forced. Instead of resetting a microprocessor the output could trigger a "Fan Failed" light or other warning. Most of them also do a power-on reset so there would be a brief warning on power up of the circuit that would need to be ignored.
 
Monarchist has stated the solution.

Well wadayah figure? There have been a billion or three of these shipped and BOUND to be at least one if not several at-work right under our own roof, too? Not just 'puters, NAS, routers, switches, laptops and printers. Even the Microwave in the kitchen.

I'm just puzzled as to how he missed the obvious, given it got HARD to even find fans as did NOT have the third wire. Not forced to use it, so they cover two-wire needs also, making all of volume production, OEM builder, store, and repair shop inventory easier.
 
No he has not. The question is based on a "plain jane" type fan. "plain jane" = 2 wires.

It was thirty years ago, yes.

Fair certain one can still buy leather saddles, goat-leather wine skins, Amerindian footwear, and even Apples not all that different than the snake and Mother Eve dealt with.

The electronics industry has moved on, though, and even "Plain Jane" now understands soap and water, wears makeup, owns underwear, rides her own John Deere tractor or Harley-Davidson, and has even ceased going about barefoot - pregnancy optional.

:D
 
With a 2-wire fan the simplest way to detect failure is with a self-heating bulb thermistor in the input air stream to the fan. When air flow stops it heats up and raises its resistance, triggering a simple comparator circuit. Doesn't even need an IC, just a small transistor (usually NPN) to trigger an LED, buzzer, etc. and a series resistor.

This is the same method used to detect low liquid levels for years. When the liquid falls below the sensor bulb it heats up when in air.
 
With a 2-wire fan the simplest way to detect failure is with a self-heating bulb thermistor in the input air stream to the fan. When air flow stops it heats up and raises its resistance, triggering a simple comparator circuit. Doesn't even need an IC, just a small transistor (usually NPN) to trigger an LED, buzzer, etc. and a series resistor.

This is the same method used to detect low liquid levels for years. When the liquid falls below the sensor bulb it heats up when in air.

All extra work, "Engineering", fabrication, components - higher costs in general - when the fan makers basically provide a proven low-hassle solution for free.
 








 
Back
Top