Replace 3-wire fans in sophisticated circuit

Discussion in 'General Electronics Chat' started by jodo, Dec 21, 2009.

  1. jodo

    Thread Starter Member

    Dec 9, 2008
    Inactive lurker here, been so busy with my day job as a software guy, I haven't done electronics for a while.

    Just in time for Christmas, the DC fans in my Sanyo Z4 projector broke down.
    It's too old to be worth a paid repair job, so I'm going at it myself.

    I need to replace two hard-to-find 70 mm 12V DC 3-wire fans.
    They are called fan 1 and fan 3 (FN901 and FN903) in the service manual.
    Both are NMB-MAT 2810KL-04W fans, but have slightly different specs.

    Fan 1. 2810KL-04W-B39 2.28W 0.19A Sanyo part 645 078 8390
    Fan 3. 2810KL-04W-B49 2.40W 0.20A Sanyo part 645 078 8383

    I have not found these parts anywhere, and I would really prefer to change them for something better anyway, since they are very noisy.

    The challenge is that 3rd wire is used by the projector, to verify that the fan speed is "not too low" -- as told in the service manual.

    The projector shuts down immediately if it is not pleased with the fans' 3rd wire signal.

    Here's the diagram for the fan control circuit.


    Another diagram shows that the desired fan speed is a linear function of temperature, and there are three temperature sensors around the unit for this purpose.

    Is there an easy way for me to fool this circuit that the fans are running well? A pulse generator of some sort?

    I am willing to power the fans externally, and so would settle for a suggestion of how to please this circuit without even connecting these two fans to it.

    Edit: I have found good alternative fans of the correct size, but lacking the 3rd wire.
    If I understood this correctly, the fan 3rd wire signal must "match" the factory fans specs.
    I'm guessing this is a linear map of expected pulses per second to PWM duty cycle. Is it really so?
    As I mentioned above, I'm willing to use 2-wire fans and power them externally with manual speed control.
    I guess I'm looking for a way might be to override or please the rpm and fan status part of above circuit.
    I need to keep the temperature shutdown mechanism though, don't want to cook it.

    I would love to have some help on this.
    Ideas much appreciated!

    Last edited: Dec 21, 2009
  2. ifixit

    Distinguished Member

    Nov 20, 2008

    a) Can you remove the tachometer form a bad fan and cobble it on to the new fan somehow?

    b) leave the old fan in the airflow so that it turns and supplies the tachometer signal.

    c) Subsitute the tach signal (difficult and challenging)...
    1. use a scope and measure the tachometer output of one of the remaining working fans, or manually turn one of the dud fans and measure the signal.
    2. use info from 1 to build a cct to replace the signal using a 555 timer ic. Modulate the frequency based on fan voltage since there is no tach.
    d) Use an optical sensor pointed at the fan blades to generate a tach signal that looks like the original.

    e) Some computer fans have tachometers in them that may give out a compatable signal. There are lots of these fans around.

    f) Dispose of the Sanyo at the e-waste center near you, and buy a new one.:)

    Seasons Greetings
  3. eblc1388

    AAC Fanatic!

    Nov 28, 2008
    I tried Google "70mm 12V Fan" and get back thousands of hits. Many of the products have a three wire connection.

    It looks like 70mm is a popular size for 478 CPU cooling so many just cost a few dollars each.
  4. JMD


    Dec 9, 2009
    Most fans have 3-pins - its been ages since ive seen one without. Newer fans even got 4 (4th is PWM).

    My advice: Find out what speed the old fans were running at (RPM), and then find some fans that match that, or maby even some that goes a bit faster.
    Last edited: Dec 22, 2009
  5. jodo

    Thread Starter Member

    Dec 9, 2008
    Thanks a lot for the comments and ideas!
    Ifixit -- I like the way you think! :)

    First off, please help me clarify what the 3rd wire IS.

    As far as I've understood, the red, voltage wire of a fan is fed with PWM that drives the fan.
    The duty cycle and current limit of the PWM set the fan speed.
    Then the yellow, tachometer wire sends out a short impulse for each completed turn.
    Is this correct? Why does some texts on the net refer to the 3rd wire as "PWM output" ??

    I was hoping that someone here was extra familiar with these status-based shutdown circuits that must be around in lots of equipment.

    Your most recommended solution so far is "search for the matching fan."

    The absolute requirements for such a fan is:

    1. 12V DC
    2. 70x70x25 mm dimensions
    3. 2000-3000 rpm

    I have found two fans that match:

    * Spire crap fan for $2. Louder than the originals and a time bomb but has 3rd wire. It's just too crappy to put in a projector.

    * Great Sunon MagLev fan. Extremely quiet and good flow. Only has two wires. I would love to use a pair of these fans with manual control.

    The speculative requirements regarding RPM/3rd wire are a gamble and currently one of:

    * The fan must reach this RPM at the wattage/current that the projector gives.
    * Or maybe the whole PWM-duty-cycle-to-rpm curve must be matched (within some error margin).

    I was trying to eliminate the speculative requirements, so that I can buy a fan without risking waste, by separating the cooling solution from the solution for pleasing the sophisticated circuit.

    Coming from the software world, it doesn't seem that difficult or crazy to replace the fans with dummy pulsers and modding it by adding manual cooling control pots. This particular thing is more difficult in electronics than in software, obviously.

    Thank you very much for your help so far.
    Anybody else have some input?
  6. JMD


    Dec 9, 2009
    Normaly you have a red, black and yellow wire (can be other colors). Where red is the positive connector, black the negative and yellow the tachometer output.
    You can read up on it here:

    Your projector uses the tachometer-output from the fans, to make sure the fans are running, and that they are running at a high enough speed (thereby making enough airflow).

    You can fake this with an oscillating circuit. This makes it possible for you to choose whatever fans you want.
  7. jodo

    Thread Starter Member

    Dec 9, 2008
    Thanks JMD.

    The wikipedia article does not explain the tachometer output, it just mentions it. I have already been over wikipedia and about three hours of further net reading.

    Unfortunately, the information I encountered also referred to the 3rd wire as the "PWM output" -- I'm wondering what that is all about.. It's conceivable to report RPM as PWM "as a factor of the maximum RPM" -- but much more complex so I doubt that it's actually done that way.

    Thanks for concurring on the oscillator faking bridge idea. I am leaning towards a test of that.

    Just trying to figure things out ! Relentless. ;)
  8. bertus


    Apr 5, 2008

    The 9 in the number B39 or B49 indicates that the speed of the fans is checked.
    If the speed is to low the signal on the white wire will change state.

    (We use this type of fans in transmitters, if the state of the white wire changes,
    we shut-off the transmiter to protect it, lighting the overheat led).

  9. jodo

    Thread Starter Member

    Dec 9, 2008
    Thanks Bertus!

    Intriguing, does that mean the fans are self-checking and only set a high/low signal on failure?

    Is there a name for this technology, so that I can inform myself some more?

    That would mean the circuit is much easier to "please" -- maybe I can even do without the manual fan speed control if it's like that.

    I get a lot of junk searching the net for these things. I'm not very experienced with electronic components characteristics, though I know some stuff about it.
  10. eng1ne


    Dec 4, 2009
    Presumably you have three fans left. What are the part numbers of these?

    One would assume that the 'PWM Output' to which you refer serves as feedback to the controller; hence 'output'.

    Though I think the scenario that bertus implies is more likely. If it is as simple as that, then I imagine the output of the yellow will be high e.g. 5 V when the speed is in range, and will drop below X V when the fan drops out of range.

    A device whose output is dependant on a signal higher or lower than a high and low threshold is essentially a comparator. If the fan produces a PWM output, although unlikely, I doubt it would be driving a comparator.

    My favourite option would be to run the projector and scope the output of the yellow wire of the remaining working fans. Failing that...

    If you have two working fans of the type that have broken, you could perhaps feed the output of the yellow wire of those into the inputs of the controller where the output signal of the broken fans were previously.
  11. bertus


    Apr 5, 2008

    On this page I found some info:
    Here is info on the numbers behind the B:

    9. Special Control Function

    0 Standard type
    8 2-speed type
    7 Temperature detecting variable speed type / PWM control type
    6 Temperature detecting variable speed type / Sensor type
    5 2-speed type / Sensor type
    9 Sensor Type

  12. eng1ne


    Dec 4, 2009

    If you have a look at the above link - well found by bertus - and hover over the lock signal text, you should see a circuit and brief description.

    My initial thoughts would be that the 'white' wire on that circuit, i.e. the lock signal, corresponds to your yellow wire. The description would suggest that when your fan is locked (broken), the signal will go high.

    If that is the case, then connecting the pcb end of the yellow wire, i.e. removing it from the fan end, and tying it to ground should bias it to the 'on' condition and fool the controller.

    Firstly though, I would check by turning on your projector with your broken fans in place and measuring the voltage on the yellow wire with reference to ground.

    Then, measure the voltage on one of the working fans' yellow wire with reference to ground.

    Of course, I very well could be wrong!
  13. jodo

    Thread Starter Member

    Dec 9, 2008
    Great info, Bertus and eng1ne!

    This info made me think about the fact that the unit worked for a long time with near-critical friction in one of the fans, so it must not be very RPM-sensitive, if at all.

    Also it makes sense that the 3rd pin can have various specialized uses, beyond tachometer -- it is that way with all kinds of electronic components. Various versions specialized for this and that. Also explains the ambiguous descriptions found when searching the net.

    Phew. :)
    If this all checks out, it will be a easy job.

    I'm going to measure and try your suggestions tonight.

    Thanks so much.
  14. eng1ne


    Dec 4, 2009
    Good luck, let us know how you get on!
  15. jodo

    Thread Starter Member

    Dec 9, 2008
    Back from Christmas break. :)

    Out of the two original fans, only one will rotate at all, and it makes an awful noise while running.

    I connected it nonetheless and scoped the 3rd wire.
    Edit: Clarification: I connected it to the projector circuit.
    That was of course the obvious missing step for understanding the problem. Thanks for pointing that out, maybe next time I'll successfully analyze my stuff on the first go. ;)

    Anyway, surprisingly, it appears to be a PWM-signal at 3.3V, 50 hz.
    The duty cycle seems to converge towards 50% as the fan speed stabilizes.

    If I put all this together, it appears to work like this:

    1. The circuit supplies 3.3V on the third wire of the fan.
    2. The fan modulates the input voltage to PWM by shorting it to ground with a transistor. The choice of 50 Hz frequency might be arbitrary or something.
    3. The circuit reads the duty cycle of the PWM-signal.
    4. I guess it shuts down if the duty cycle is too short or too wide, but is satisfied by anything near 50%.

    So the next step is to build a simple square wave oscillator @ 3.3V, 50 Hz, 50% duty cycle. :)

    Any tips on suitable highly available parts for that?
    Thanks so much for your help.
  16. bertus


    Apr 5, 2008

    Beware that when you make an oscillator to "fool" the circuit,
    the projector will not be protected at all.

    You will probacly need fans with pulses at the sense output.

  17. jodo

    Thread Starter Member

    Dec 9, 2008
    I know, it's a bit risky perhaps.
    But there's an ultimate resettable temperature fuse at the lamp, which shuts down everything, and all temperature sensors will still be in place, so I guess this just weakens the protection to a degree.
    I'm willing to go ahead on that basis, but thanks for the heads-up.

    Pulses on the sense output -- sure, if anybody has a circuit that will easily turn a regular tach output into this kind of RPM-deviation PWM-signal, I'd be interested in that too. But it appears to me as a more complex task than the faking oscillator.
  18. jodo

    Thread Starter Member

    Dec 9, 2008
    It worked!

    I used the Sunon MagLev fans with two wires.

    For the "fooling" oscillator circuit, I used a TI cmos 555 timer chip in astable mode, with a cap and a resistor, to get the 50 hz square wave. The square wave was fed to a C945 NPN-transistor, which was connected to the 3rd wire of fan1 and fan3 (same signal, same transistor for both.)

    I had to find the +5V power supply line in the projector to feed the 555-chip. It worked like a charm. The wafer-board with the few components fit nicely into a little isolation-taped package that I could easily tuck into the projector guts.

    I had to start service mode and set the fan speeds to manual, because the current was very high for the new fans (sounded like a vacuum cleaner). Now it's quiet and running fine.

    I didn't take any photos, my cameras all suck but diagrams can be had if anybody is interested, let me know in that case.

    Thanks for all the advice and ideas!
    Happy hacking ;)
  19. creakndale

    Active Member

    Mar 13, 2009
    I've been following this thread with great interest as I have a 6+ year old Sharp Projector developing a similar issue. Months ago, one of the fans began making a racket. I was able to narrow it down to a small 3-wire blower. I searched the internet for a replacement. I found a guy on eBay who was parting out the same model projector and wanted $40 for a used replacement blower. I eventually found a place that sold new blowers but they are only 2-wire. I purchased two of the 2-wire blowers as they were cheap at $6 each.

    My short term solution: I carefully took apart the noisy blower and was able to oil the two tiny ball bearings. The bearing noise subsided but it's probably just a matter of time before it starts becoming noisy again. I was toying with the idea of dissecting the moving parts from a new blower and putting them into the worn out blower. Failing that I would try fooling the projector circuitry as your successfully did. I don't know what signal is on the 3rd wire as I haven't scoped it yet.

    I would like to see your schematic if it's not too much trouble.

  20. jodo

    Thread Starter Member

    Dec 9, 2008
    Thanks for showing interest.

    Here's what I used:


    If Vdd is above 5V, one might want to put a resistor at the base of the transistor. I used 5V without the extra resistor. The cap was a 35V tantalum.

    A few notes:

    The formula for frequency is f = \frac{1}{1.4RC}
    The good matching resistor for 50 Hz and 680nF is actually 21kΩ but 14.3k was the closest I had which results in about 74 Hz. The PWM receiver doesn't care about exact frequency, use the parts you have at hand and match R and C using the above equation and you will be fine.

    It turns out I was actually wrong about the purpose of the 3rd wire PWM-signal, it seems the duty cycle converts to RPM. I know that since the service menu now reports a constant fan speed of 1600 RPM for these two fans. The start-up phase used this RPM which made the square wave about 50% -- and that fooled this scope n00b.

    One could install a variable resistor to control the PWM duty cycle of the 555, which would set the RPM reading seen by the unit's software. As I'm using manual speed now anyways, and the signal does fool the projector that the fans are running adequately, I feel no need to improve it.

    Feel free to ask if you have questions.