• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 46 subscribers
  • Views 2682 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AMC6821 PWM cycling

Chris Westervelt
Intellectual 440 points
Other Parts Discussed in Thread: AMC6821

I am designing in the AMC6821 into one of our embedded Linux boxes and I am trying to understand a phenomenon or normal operational modality of the chip.  When the driver loads and initializes the chip, the fan slows and begins surging on and off.  When looking at the PWM before and after, the PWM is a steady output but after the driver initializes, the PWM cycles on and off at about .5 to .75 second intervals.  In other words, the PWM cycles on and off slowly.  Why is this the case in temperature loop mode and can it be set to just run a PWM steady with closed loop temperature control changing the duty cycle according to the loop parameters?

 

Chris

  • 0
    TI__Expert 6925 points

    Hello Chris,

    Welcome to the Precision Data Converters forum!

    We are sorry it took this long for your post to be redirected.

    The AMC6821 has a function called “spin-up” that generates a ramp to 100% duty cycle (See Figure 24 of the datasheet). It gives the fan momentum before sending a PWM signal.

    This function is run after power up or if the TACH detects a TACH value below or above the TACH limits. If the TACH never reads a correct value, the AMC6821 will cycle in spin-up mode forever, which looks like a very slow PWM signal.

    The TACH can read wrong values if the PWM polarity is not the polarity expected by the fan controller input. By default, the polarity is set to low for 100% duty cycle in order to drive the PMOS transistor typically used to control 3 wire fans. Usually, with 4-wire fans, this polarity needs to be reversed by writing 1 to the PWMINV bit.

    Another setting that affects the TACH value is the pulses for revolution setting. The AMC6821 only has two options: 2pulses/rev or 4pulses/rev.

    Besides, sometimes pulses per revolution are confused with poles per revolution. For example, a 4 poles fan is actually a 2 pulses per revolution fan.

    Could you share with us the settings programmed to the AMC6821 including the TACH limit registers and the datasheet of the fan? We would like to check if the AMC6821 settings are correct.

    Also, could you send us the schematics? We just want to check if the TACH input is properly driven.

    Thanks and again sorry for the delay,

  • 0 in reply to Rafael Ordonez
    Intellectual 440 points

    Rafael,

    Thanks for the response.  To answer your questions, I am using the AMC6821EVM board to do my testing before integrating it into my next board update.  I am using the stock Linux driver that is supplied with most kernel distributions.  See http://lxr.free-electrons.com/source/drivers/hwmon/amc6821.c?a=mips for a listing.  I am setting the board up for a three lead fan.  Other than that, I'm not doing anything special.  I have tried inverting the PWM which does not help for 3 or 4 wire fan configurations.  Changing from 2 to 4 polls does nothing either.  The tach readout is always 91.  I don't really understand how the tach is supposed to work right when the ground reference is cut off during the blank period of the PWM especially when the PWM is running much faster frequency wise than the output pulses of the fan.  If the fan is to act as a generator during the off phase to produce current for the tach then it would need some sort of storage device and you cant always expect a phase brushless motor to act as a generator since semiconductors are in series with the windings.  The fan I'm using is a Sunon MagLev model GM0502PFV1-8.F.GN.  If you can get this combination to work, please let me know what settings do the job.

    Thanks

    Chris

  • 0 in reply to Chris Westervelt
    TI__Expert 6925 points

    Hello Chris,

    Thank you for your feedback.

    I have never worked with the linux driver you mentioned before. I don’t think it is supported by TI. However, making the AMC6821EVM work with a 3-leaded fan is not too difficult.

    First, you need to change the jumpers in the EVM for the TACH signal to show up properly. The Sunon fan is a 5V fan and the J5 jumpers in the EVM are set for 12V fans by default. So, change the J5 header to march the following figure:

    As you mentioned before, during OFF time, TACH pulses will not show properly. In order to correct that, the AMC6821 has the option to keep the power ON while taking a TACH measurement. For 3 leaded fans, this option is critical. It is activated if the TACH-MODE bit in configuration register 2 is “0”. So, make sure you have active.

    About the polarity bit, since the EVM uses a NMOS, PWMINV must be 1.

    The Sunon fan generates 2 pulses per revolution, so you need to use the default value in the PSPR bit of configuration register 4. (PSPR=0)

    Try this and let us know what happens. It might take a few more post, but we will help you to get it working.

    Best Regards,

  • 0 in reply to Rafael Ordonez
    Intellectual 440 points

    Those are good settings and I was able set up the system trough the device drivers just fine.  Thanks for the help.