Connect with us

Detectig fan failure - modes of failude

Discussion in 'Electronic Design' started by SioL, Sep 15, 2004.

Scroll to continue with content
  1. SioL

    SioL Guest

    I'm designing a device with two convenient solder pads on the PCB
    providing LC filtered power to the fan to. This will be a smallish 1W
    12V fan, with only two connections (those are cheaper and I can't get
    the 3-pin version of 40x40x20 easily).

    It occurred to me that it might be great to have failure detection. Now
    it'd be easy if these were the 3-pin fans, but they're not. I wonder whether
    someone has some experience on modes of failure. Does it consistently
    draw more current, no current or should I watch for some AC ripple on
    supply line during normal operation, I'd expect there would be some and
    it'd stop if the fan stopped rotating.

    I'll possibly settle for a NTC if everything else fails, but I have some
    PCB real-estate left for other solutions.

    SioL
     
  2. Yes, that will work. It might be patented however.


    Best regards,
    Spehro Pefhany
     
  3. SioL

    SioL Guest

    Oh, well, most complicated solution. I hate it when the only
    solution is the most complicated one. I'll probably go the NTC route.

    Thanks,
    SioL
     
  4. SioL

    SioL Guest

  5. Jim Thompson

    Jim Thompson Guest

    They have horrible current ripple, so bad that I once designed a
    quick-fix circuit to fill-in the ripple, because the ripple was
    disturbing a nearby CRT in a piece of OmniComp/GenRad portable gear.

    Did someone actually patent looking at the ripple ?:)

    ...Jim Thompson
     
  6. SioL

    SioL Guest

    Judging from the datasheet
    (http://ww1.microchip.com/downloads/en/DeviceDoc/21688c.pdf)
    a small 100mA 12V fan generates about 125mV of ripple over a series 4.7 ohm
    resistor. With
    sharp edges, to make things worse as far as interference is concerned. And the
    connecting
    wires are a nice antenna.

    SioL
     
  7. No capacitors inside them. ;-)
    Seems kinda obvious. I had a vague notion M*x*m did, but I've never
    had a reason to delve into it further. More expensive fans can have
    tacho output or locked rotor output, of course. Here's one of M*x*m's
    chips:

    http://pdfserv.maxim-ic.com/en/ds/MAX6684.pdf

    Best regards,
    Spehro Pefhany
     
  8. B

    B Guest

    By NTC you mean using a termsistor in the airflow and moitor the temprature
    rise if it stops overheating?
    This is the most reliable method as this will actually monitor the desired
    thing - airflow.

    If (very very unlikely) all the blades broke the motor will still turn but
    there will be no airflow.

    Thanks
    Darren

    http://www.speff.com
     
  9. Spajky

    Spajky Guest

    check my site under electronics/fans-III how to monitor fan RPMs with
    fans w/o 3rd wire; maybe you will get some idea later after exame of
    the first stage of that circuitry ....
     
  10. SioL

    SioL Guest

    What's more likely is that someone places a piece of paper on top and blocks the
    airflow. The fan will spin, but airflow won't, well, flow.

    SioL
     
  11. Ian Stirling

    Ian Stirling Guest

    This can be detected - the speed goes up.
    The ideal solution is probably monitoring the temperature of the
    equipment you care about.
    Do you care if it's getting too hot because the fan has failed, or the A/C
    has gone offline, the room is at 40C, and the sun is shining on the case
    heating it to a nice toasty 80C.
     
  12. John Fields

    John Fields Guest

     
  13. Joerg

    Joerg Guest

    Hi SioL,

    Now this is really low-tech but I have seen it used in most equipment
    where plugging of the airduct can lead to some level of catastrophe: A
    regular end switch with a small piece of metal fastened to its switch
    lever. They look like the roller type switch but without the roller and
    with the tongue bent up.

    This works nearly unconditionally. The system can check switch action
    (fan off, switch must be off). When anything happens that prevents
    enough airflow, whatever it is, the switch alerts. This solution is even
    installed in our gas furnace and these things have to go through
    rigorous testing before release.

    Regards, Joerg
     
  14. SioL

    SioL Guest

    I've used two NTC's, one for the heatsink, the other for selfheating/airflow.
    That should cover everything.

    Thanks for the sugestions, guys.

    SioL
     
  15. Aiiiieeeee!"!!!

    I worked at a place - JET, UK - where they had several foot-ball-field-sized
    buildings full of interesting electric machinery three stories high. Apart
    from broken wires and loose connectors, *the* most unreliable pile of thrash
    were those "flap-switches"!

    Everything mechanical is evil -

    But those things are particularily evil, being difficult to place correctly
    in the first place, hard to calibrate properly with almost random detection
    range (especially after being covered in dust) and then sporting non-sealed
    contacts liable to being covered in Dust, SO2 or SF6 - or just sensitive to
    plain vibration. "Cheap" is about the only positive adjective that can be
    attached to those "devices". ;-)

    The heat-based flow sensors, O.T.O.H, *work* - and they can be easily
    calibrated to max and min limits too. The only deficiency is that, after
    some years, the air flow tends to erode the leads to the sensor so that it
    fails open (But this is detected too).
     
  16. Jon Elson

    Jon Elson Guest

    The advantage of the PTC thermistor is that it detects AIRFLOW failure,
    not just loss of fan rotation. (The NTC thermistor is usually used as
    an inrush surge limiter, the PTC is better for airflow detection.)
    So, if a piece of paper is sucked over the fan inlet, the PTC will get
    hot. The fan tach would never know there was a problem.

    AC ripple detection or current level is not totally reliable, as some
    fans shut off completely after several seconds of stall, others start
    pulsing, so you get AC.

    Jon
     
  17. Ian Stirling

    Ian Stirling Guest

    Plus, if you get the numbers just right, it can do not only airflow,
    but overtemp detection too, which is a bonus.
     
  18. John Fields

    John Fields Guest

    ---
    Since it's in the nature of the beasts' resistances to decrease with
    increasing temperature, NTC thermistors are _always_ used as passive
    inrush surge current limiters. However, I don't believe it's true
    that PTC thermistors are better suited for airflow detection than NTC
    thermistors. What evidence do you have to support the position that
    they are?
     
  19. Zak

    Zak Guest

    If you use a voltage source a PTC is better suited. NTCs may burn out as
    current keeps increasing above a certain temperature.

    Use a constant current source and an NTC is more suitable. But contant
    voltage is more common...


    Thomas
     
  20. John Fields

    John Fields Guest

    ---
    That's ridiculous. Neither sensor is used alone, and you seem to
    have forgotten (or you never knew) that in order to use an NTC
    thermistor as a self-heated flow sensor it's almost always configured
    as one of the elements of a voltage divider or a self-balancing
    bridge, and that that divider or bridge is fed by a voltage source.
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-