Controlling fan speed?

Discussion in 'UDOO BOLT' started by itimpi, Sep 16, 2019.

  1. itimpi

    itimpi UDOOer

    Joined:
    Oct 31, 2013
    Messages:
    136
    Likes Received:
    39
    Does anyone know if the Bolt hardware is capable of controlling the speed of the fan? At the moment the Bolt BIOS seems to allow you to control the duty cycle (the proportion of time the fan is turned on), but when it turns on it always seems to be at full speed.

    Since my Bolt never seems to run above 60C even when heavily loaded rather then setting a duty cycle I would much prefer to run the fan at a slower speed (which should mean it is quieter) all the time rather than having it turning on off all the time.
     
  2. ccs_hello

    ccs_hello UDOOer

    Joined:
    Apr 15, 2017
    Messages:
    536
    Likes Received:
    194
    Need the new (future) firmware (on BOLT's embedded controller MEC1705, which has its own PWM control circuit.)

    Note: it is not for driving the "PWM fan" but a regular DC fan, latter is what comes with the BOLT.
     
  3. someone_passingby

    someone_passingby UDOOer

    Joined:
    Oct 3, 2019
    Messages:
    24
    Likes Received:
    7
    I can confirm this.

    Stock (default) fan: Correctly follows the duty cycle (in %) value.
    Noctua NF A6x25 5V PWM: Correctly works only when the duty cycle is set to 100% (both 80% and 90% do not work).
     
  4. Deluxe

    Deluxe UDOOer

    Joined:
    Aug 13, 2019
    Messages:
    13
    Likes Received:
    1
    Can you share your settings? I'm not able to get the fan to do anything other than "full ON" 100% duty cycle, fan spinning up and down in rather short intervals (about ~95-99%) and "full OFF" (bellow ~95%) at which point fan just twitches (until CPU hits 75*C and fan goes 100% as it should).
     
  5. Zgembo

    Zgembo UDOOer

    Joined:
    Sep 21, 2019
    Messages:
    49
    Likes Received:
    27
    Same here, my "green batch" stock fan (marked with green dot on one side of the fan), behaves exactly the same. Looks like the "green batch" fan is COFAN F-6010H05BIV, which is a 5V 0.45A 3-wire fan, spec here http://cofan-usa.com/pdf/dc_6010iv.pdf.
    One thing that I fail to understand; CN17, being the Bolt "CPU Fan" connector is only 3-wire, but PWM (with duty cycle settings & co) is a standard for 4-wire fans, where 2 wires are GND & voltage, 1 is TACH sense (RPM feedback pulse from fan) and finally dedicated wire for PWM pulse control. On Bolt's CN17, I can only see GND (black wire), TACH (yellow wire), and one +5V\PWM combined pin (red wire). How is this supposed to work at all? The COFAN fan itself has one unused solder point marked as "P", the red wire is connected only to "+", which totally indicates it is only a power source, not PWM. Could be that the "green batch" fans are not PWM at all, or they can only operate as 4-wire PWM fans?!?
     
  6. someone_passingby

    someone_passingby UDOOer

    Joined:
    Oct 3, 2019
    Messages:
    24
    Likes Received:
    7
    Hi Again!

    May I check whether you are experiencing this with a stock fan or the third-party fan you purcahsed separately? In my case (bios v1.05 R.C. 8), I have the same issue with Noctua NF-A6X25 5V PWM, but not with COFAN F-6010H05BIV (stock fan).

    The conclusion is that COFAN (stock fan) with the latest bios follows the duty cycle percentage correctly (when I tested, any value between 10% to 100%). However, Noctua fan (4 wire PWM fan connected to 3 wire connector of Bolt as explained by @Zgembo) still has the exactly same issue explained by @Deluxe and @Zgembo, even when the latest bios is installed.

    Feel free to get back to me for clarifications & misunderstanding correction! I'm still exploring various aspects of Bolt as well and I may misunderstand how Bolt works.

    Thanks.
     
  7. someone_passingby

    someone_passingby UDOOer

    Joined:
    Oct 3, 2019
    Messages:
    24
    Likes Received:
    7
    Looks like COFAN (the stock fan) I received is different from yours. Mine does not have a green dot and the model number is F-6010M05BII. The last letter is I instead of V. This fan does not seem to be PWM and it follows the duty cycle setting in Bios correctly. The new Noctua fan I purchased is 4 wire PWM fan, and it shows the exactly same issue you are having now with COFAN.

    So my gut feeling is that the "green batch" fan you received is the 4 wire PWM version, while the new fan I received is the non-PWM version. Maybe we can dig the COFAN USA website more to find the differences (https://www.cofan-usa.com/fans-blowers/#_legacy-fan-part-numbering).

    Please correct me if my understanding is wrong.
    Thanks!
     
  8. someone_passingby

    someone_passingby UDOOer

    Joined:
    Oct 3, 2019
    Messages:
    24
    Likes Received:
    7
    IMG_20191030_014154_544.jpg
    Photo of my COFAN.
     
  9. Zgembo

    Zgembo UDOOer

    Joined:
    Sep 21, 2019
    Messages:
    49
    Likes Received:
    27
    Hold your horses, I clearly stated that I have PWM duty cycle issues with COFAN F-6010H05BIV (stock fan). Exactly same issue as @Deluxe;
    FYI, my Bolt is on BIOS 1.05 RC13, Embedded Controller FW v0.14 / Library v0.15.

    I tried some old 12V fans on the CN17, they work fine in low RPM mode, no PWM of whatsoever.

    @someone_passingby is your stock COFAN fan perhaps marked with green dot on its side?
     
  10. Zgembo

    Zgembo UDOOer

    Joined:
    Sep 21, 2019
    Messages:
    49
    Likes Received:
    27
    @someone_passingby Just saw the picture you posted, hmm, you actually have COFAN F-6010M05BII, and I (and some other people here on the forum) have COFAN F-6010H05BIV. That must be it, looks like H05BIV variant has broken PWM, but M05BII works just fine as you claim...
     
  11. someone_passingby

    someone_passingby UDOOer

    Joined:
    Oct 3, 2019
    Messages:
    24
    Likes Received:
    7
    Aha. So our fans are clearly different:

    F-6010H05BIV
    F-6010M05BII

    I guess these two fans must have different specs and that is why our fans are behaving differently?
     
  12. someone_passingby

    someone_passingby UDOOer

    Joined:
    Oct 3, 2019
    Messages:
    24
    Likes Received:
    7
    yeah.. Mine is Mouser-shipped. I'm not a Kickstarter backer. Maybe this new stock fan is otw to Kickstarter backers..?

    But even with the new stock COFAN, it is damn noisy anyway at full speed lol. So I ended up buying the Noctua 60mm 5V PWM and configured it to always run at 100%, which is waaaaay much quieter than the new stock COFAN running at 100% haha.
     
  13. Zgembo

    Zgembo UDOOer

    Joined:
    Sep 21, 2019
    Messages:
    49
    Likes Received:
    27
    Here is some COFAN spec for B II variants; http://cofan-usa.com/pdf/dc_6010ii.pdf. It doesn't say anything about PWM besides listing it as an "optional feature", who know what that really means...
     
  14. Zgembo

    Zgembo UDOOer

    Joined:
    Sep 21, 2019
    Messages:
    49
    Likes Received:
    27
    Interesting, sent you PM case you don't need stock COFAN anymore...
     
  15. itimpi

    itimpi UDOOer

    Joined:
    Oct 31, 2013
    Messages:
    136
    Likes Received:
    39
    I have successfully been using a Gelid Silent 6 fan which has the great advantage of (just) fitting in the metal case.
     
  16. Zgembo

    Zgembo UDOOer

    Joined:
    Sep 21, 2019
    Messages:
    49
    Likes Received:
    27
    Gelid Silent 6 is a 12V fan, the likely reason it works well is that it is either constantly off or constantly on in lowest voltage mode - hence the most silent one, yet enough to properly cool the APU... I seriously doubt Bolt's "PWM" would work with it at all.

    So far the only fan that is confirmed working against Bolt "PWM" is F-6010M05BII, that can't be purchased anywhere...

    @evaloverde Very unlikely that there is a way that any BIOS/firmware update can fix F-6010H05BIV (aka "green dot batch" fan) situation, it is technically impossible 'cause F-6010H05BIV can only be controlled by changing input voltage, but on-board embedded microcontroller can not control and change +5V "PWM" voltage on CN17 pin (unless there is some super secret undocumented feature in the controller chip), only its pulse and "PWM" frequency.

    Please escalate this to Udoo tech team, F-6010H05BIV is mismatching fan for Bolt's embedded controller, and will never work as intended. The ability to only switch fan on/off can't be considered as satisfactory in 2019, ability to control this particular fan RPM is broken, unlikely to ever be fixed.

    Where / how do I submit request ticket for fan replacement?
     
  17. nophead

    nophead UDOOer

    Joined:
    Aug 18, 2015
    Messages:
    87
    Likes Received:
    14
    Why can that fan not be PWMed? I think all 3 pin fans can be slowed down by pulsing the 5V at some appropriate frequency.
     
  18. Deluxe

    Deluxe UDOOer

    Joined:
    Aug 13, 2019
    Messages:
    13
    Likes Received:
    1
    In general that should be the case (PWM signal can be seen as averaged DC). However, I played around with stock fan I got with Bolt and it works only in about 4.75 to 5V range. It was almost imposible to nail a voltage at which it wouldn't go either full 4k RPM or 0. If I do something similar with 60x60mm 12V fan I can get smooth transition from its 4k RPM (at 12V) down to 4V.
    I'm thinking about using this 12V with Bolt's PWM, but I'm not sure if it will be capable to cool it at 5V.
     
  19. nophead

    nophead UDOOer

    Joined:
    Aug 18, 2015
    Messages:
    87
    Likes Received:
    14
    PWM isn't the same as changing the voltage. Reducing the voltage reduces the torque until some point where the electronics stop working. With PWM the fan always has 5V or is off. The result is while it is on the electronics should work and the fan should drive at full torque. While off the fan free wheels, so the rotor inertia averages the speed out.

    It is very odd that a 5V fan stops working at 4.75V. It should not be that critical and probably explains why some fans don't work at all. A small 5V fan I am using on my Raspberry PI works down to about 2V.
     
  20. Zgembo

    Zgembo UDOOer

    Joined:
    Sep 21, 2019
    Messages:
    49
    Likes Received:
    27
    Looks like COFAN produces multiple variants of 3-wire 5V 60mm fans, with different electronic circuits on the fan PCB. By the reports in this thread, F-6010M05BII works OK with Bolt's 3-wire PWM (which is highly unusual; try searching the net for 3-wire PWM, virtually nothing comes out, heck, even 4-wire 5V 60mm PWM fans are rarity), where most other fans just stutter and squeak with PWM duty cycle set to anything other than 100%. Trust me on this, I've tried 5-6 different fans connected to Bolt's CN17, against various PWM duty cycle and pulse frequency BIOS settings. With high enough pulse freq. some fans would behave just as if continuous voltage was supplied (probably some capacitor on fan PCB is averaging voltage pulses), with about the same effective RPM as when normal continuous +5V is supplied. I have an impression that 99.99% 2-pin and 3-pin DC fans out there are designed just to be voltage-controlled and nothing else, they would behave unexpectedly if some sort pulse voltage is supplied via 5V wire instead of continuous voltage...

    In theory, it could be possible to control the fan speed via 2 wire connection by sending PWM-like 5V pulses, but then the PCB on the fan side has to be built specifically for such purpose. Could it be that F-6010M05BII is that sort of super special fan?!
     

Share This Page