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.

AMC6821EVM-PDK: Shorted PWM and 12 VDC

Part Number: AMC6821EVM-PDK

Hi,

I have the AMC6821EVM-PDK and I was testing a 4-pin 12 V fan with it. I made an error with the wire connection and the PWM and 12 VDC connectors were shorted through the J1 connection. A puff of smoke rose from the board when I plugged in the USB cable. A 12 V adapter was also connected to the board. Something obviously blew up judging by the smoke, but what? I have a new AMC6821EVM coming, but if there's a chance that the original can be repaired, I will do that also.

What are the components on the AMC6821EVM that likely need to be replaced and is it possible that the USB-MODEVM-board survived the short, or do I need to replace the whole PDK? 

  • I'm looking into this.
  • Hi Pasi,

    Thank you for reaching out to us. The fastest way is to get a new EVM so that you can continue evaluating the AMC6821. Please let me know I can send one for you immediately. I need more time to review the schematic, and get back to you since we inherited this product several months ago.

    Aaron

  • Pasi,

    You most likely damaged the PWM_OUT pin with the 12V.
    The EVM may be saved with a new AMC6821 IC.

    -Kelvin
  • Thanks for the reply! I should be receiving a new AMC6821EVM tomorrow, so we will see shortly if the USB-MODEVM was undamaged by the short. I will also order a new AMC6821 IC to replace the damaged one on the old board. I will reply to this thread when I can confirm that the kit is working again.
  • I received a new AMC6821EVM board today and connected it to the old USB-MODEVM. However, it doesn't work. I run 4-pin 12 V PWM fans with it, and they only spin if the PWM-pin is disconnected. Even the TACH reading was shown in the software. If I plugin the connector all the way so that the PWM pin is also connected, the fan stops. In my understanding the fan should spin at 33 % PWM when the board is powered on and the software is not running.

    I am pretty sure this is the same problem I had BEFORE I accidentally shorted the 12 V. I had made a temporary, bare wire, connection with a fan to the board via a cable that had the standard connector at one end and bare wires on the other. When I plugged in the 12 V power and USB the fan spun up. However, it was upside down on the table, so I disconnected the USB cable and removed the 12 V adapter from the wall socket. I turned the fan around, joined the wires again as they had bundled up touching each other, and powered the board up again. The fan spun up only for a moment and then it stopped. This began the 2 hour troubleshooting which ended up in the unfortunate smoky short.

    How can I troubleshoot this? The EVM board is new, but the USB-MODEVM is the same that was in use during the short. It would seem that the PWM is the problem, since the fan(s) spin(s) with the PWM pin disconnected. I've attached a photo of the setup, where you can see the jumper and switch positions. The setup should be for PWM-controlled 12 V 4-pin fan.

  • Hi Pasi,

    There are a few things you can do to debug the board.

    You will need a pulse generator and an oscilloscope.

    Let's first make sure the 4-wire fan is working like you expect it to.  Bypass the PWMOUT of EVM and use a the pulse generator to control the fan.  Play with duty cycle and see if the fan is operating like it should. Pay attention to the polarity of the fan PWM control (i.e. is it active low or active high).  Scope the TACH output and see if it changing accordingly.  Since this this a 4-wire fan, it probably has an integrated fan driver (NMOS/PMOS?).  

    Note: Upon power up, the PWMINV bit of configuration register 1 (address 0x00) is reset to zero, which assumes the PMOS device driver polarity.  So if your fan has an NMOS driver, then the fan spin up will have a similar problem to Figure 15 of the user's guide.  There is a relatively clean hardware (HW) workaround for NMOS driver of 3-wire fan mode (Figure 16 of EVM user's guide) involving U2A, R10, and R16.  There's no instruction for a HW workaround in the user's guide for 4-wire fan mode.  But you can unplug the jumper on J5 (pins 9 and 10) and short pins 10 and 12 on J5.  This will effective invert the default PWM_OUT waveform for the 4-wire fan mode.  I assume you have R10 and R16 populated at this point.

    You can also change the PWMINV bit in software also.  This would be the quickest way to find out if the default PWM polarity is the root cause.

    If the above steps aren't the answer then you will need to debug the PWM_OUTPUT waveform without the fan.  Play with the control in the GUI and see if the waveform is what you expect.

    -Kelvin

  • Hi,

    I debugged the setup with a scope and a pulse generator, and I think I now know the reason why the fans don't spin: the PWM voltage amplitude is too small. I connected the scope to the board's PWM pin and you can see the output in the picture below. The duty cycle setting worked fine, but as you can see, the amplitude is only 100 mV when it should be 3.3 V.

    I then tested the fan by disconnecting the PWM wire from the connector so that all the other pins were still connected to the board. The (Noctua) fan spun up to full speed and the TACH reading was correct in the software. I then connected the pulse generator to the PWM wire and input 25 kHz, 3.3 V square wave with 50 % duty cycle. The fan slowed down and the readings were again correct. I tested other duty cycles as well and they also worked. I then lowered the voltage amplitude and the fan stopped spinning when it was below 0.8 V. So that's the reason why the fan doesn't spin. But what is the reason for the low voltage?

  • Pasi,

    Good work on narrowing down the issue.

    Quick question: In the picture, I noticed that you have W6 and W7 configured so that the DUT and digital circuits operate off 5V (ref Table 4 of user's guide). Is this the case? Check the VDD voltage to make sure it's 5V.

    The PWM high level should be VDD since it's pulled up to VDD by the 2.2kohm (refer to schematic). The PWM_OUT is an open drain output so the high logic level is set by the pull up VDD. Just to be sure, is the PWM low level around 0V?
    When you tested the PWM_OUT, did you have the fan connected? Make sure to scope the PWM_OUT without the load (i.e. the fan).
    I just want to make sure we don't have a loading issue.

    -Kelvin
  • Hi,

    The W6 and W7 pins are in their default positions from the factory. I will confirm the Vdd voltage tomorrow. The oscilloscope picture in the last post was taken when the fan was (totally) disconnected. Only the scope was connected to the PWM pin. The PWM amplitude was the same with the fan connected. The PWM low level is around 0 V (the low level is in the middle of the screen in the scope picture).

  • Hi,

    I did some additional measurements with a multimeter. I measured the voltage from either side of the R13 (2.2 kOhm) with the PWM DCY set to 100 %. On the J5 side the voltage was 0.3 V, the same as in the J1 PWM pin. On the other side it's 5 V. So it seems that the PWM voltage isn't pulled up.

  • Pasi,

    That's interesting.  The PWM_OUT is an open drain output so it's supposed to be pulled up the VDD rail.  Did you check the voltage of VDUT?

    When you change the PWM_MODE (W4), does the PWM frequency change? 

    Does this happen for both EVMs?

    -Kelvin

  • Hi,

    The frequency changed correctly when changing the W4, but the voltage was still too low. I measured the VDUT from C1/C5 and from R23 and it was 5 V. I also tried to change the voltages to 3.3 V with W6 and W7, and that worked also. Although this resulted in a further drop in the PWM voltage (about halved).

    I haven't even measured the old EVM, because I released the smoke out of it. As we know, the smoke should stay inside the components for them to work correctly. :)

  • Pasi,

    This is strange.  I still waiting for the AMC6821EVM to ship to me.  I will let you know if I get the same result on my end.

    -Kelvin