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.

Can't get PWM signal to vary from "HI" using AMC6821EVM-PDK kit

Other Parts Discussed in Thread: AMC6821

Hi All, 

I'm new to the AMC6821, and am using the AMC6821EVM-PDK kit to evaluate it.  

I have a problem where I can't get the PWM signal to change at all.  The o-scope shows it's high all the time, with no pulses.  I've tried several of the automatic loads in the VI software tool, and the fan rpm never changes from full rated speed - as verified by the tach signal.

I can update/change duty cycle registers in the VI tool, but the actual signal to the fan never changes.

I have verified that the fan responds correctly to a PWM signal from an function generator when the EVM is not in the loop.

I've looked at the HW jumpers and think everything is setup per the manual.  I'm now at the point where I don't know what to check next.

Any suggestions would be appreciated.

Thanks,

Michael

 

 

  • Hello Michael,

    Where exactly are you connecting the o-scope probe to? Is connected to J4 pin 4 or J5 pin 11?

    J5 pin 11 or 9 are connected directly to the PWM output of the AMC6821, so I hope you are connecting the o-scope there.

    A simple test is to disconnect the fan, and turn off and on the board. By default, the AMC6821 starts up with a PWM signal of 33% duty cycle (33% low, 67%high). So, without starting the software, simply connect the USB and check for the 33% duty cycle signal in J5 pin 11 or 9.

    Also check if any red LED is on. All red LED should be of after power up.
     
    If the signal is there, then we can assume that the problem is with the jumper configuration of J5 and W5.

    If the signal is not there, it might be a power related problem, so make sure that LEDs D1 and D5 are illuminated with good intensity. If D1 has less intensity than D5, then change W6 in order to drive the VDUT voltage properly.

    By the way, could you attach the datasheet of the fan you are using? I would like to know more details like operating voltage and frequency, and be carful with the wires because different companies have different color coding.

    For the TACH register, what is the value for you read from it?

    Let me know if the previous recommendations help?

    I have the same board here, but since I made some modifications to it I have to put it back together, so after you send me the fan datasheet, I can send you a picture of how the board should be setup to make you fan work. Assuming the board doesn’t have a damage component.

    Regards,

  • Rafael - 

    I had the probe on the PWM pin on the J1 fan connector.

    I ran the board on just USB, and probed J5 Pin 9, and got a 5V square signal signal with pulses divided 33% low, and 66% hi.

    When I start the USB-MODEVM on just USB, the leds show as follows:

    D2 - Yellow, D3 - Green, D4 - Green, D5 - Red, D6 - Not lit, D7 - green, D8 - green

    When I power up USB-MODEVM with the AMC6821-EVM attached, D1 and D5 on AMC6821-EVM are bright green, and the LEDs on USB-MODEVM board remain unchanged.  I tried to change W6 and saw no change in D1 or D5.

    I connected the Nexus fan to J1, and 12VDC to J7.  (I have verified the pinout of the fan connector is correct based on my external testing with the function generator and scope on the tachs.).  During the power up, I observe the scope and watch the PWM pulses start at about %50 HI, shrink to about 10% HI, and then go to about 67% HI - all in about 3 seconds.

    I start up the Eval Tool SW, and reset the AMC until I get all green indicators.  I start fan monitoring, and the PWM DC shows 33%, and RPM is "INF".  (Note that the fan appears to be running at full speed).

    I then go to "register auto programming" and load a "FAN Max Speed" profile, and I observe the PWM signal change to pulses about 33% HI.  The fan monitoring is now showing ~1960rpm and 33% duty cycle.

    I then load a "FAN software DCY control" profile, and note that the DC still shows 33% on the tool and 33% HI on the scope.  I change the "PWM Duty Cycle" parameter to 100% (255) and observe the DC read back as 100% on the tool and 100% HI on the scope!  (Milestone!)  (Fan speed appears constant).

    Now, I change the DC to 10%, and 50%, and 75% and 99% - and each time the results are the same.  The tool reads back the correct PWM, the scope shows the correct signal - but the FAN RPM never changes.

    Note that I'm using two fans:

    1) Nexus D12SL-12

    2) Delta FFB1212EH (Data sheet attached).

    7230.Delta PWM Fan FFB120x120x25mm.pdf

    Both fans are very different, but perform exactly the same when uses with the eval kit.  (They also function properly when not on the kit).

    Thanks,

    Michael

     

     

     

     

  • Rafael - have you had a chance to review my response?  

    I'm curious what your thoughts are on why I get a PWM signal from the IC, but not at the fan.  

    Thanks for the help.

     - Michael

     

  • Hello Michael,

    I'm sorry for the delay, but I'm glad you were able to read and control the PWM signal.

    As for the PWM signal, I cannot think of anything else besides the jumpers in J5.

    Previously, you mentioned that you were reading the signal from J5 pin 9. This tells me that the third jumper in J5 is connecting pins 11 and 12 instead of pins 9 and 10.

    With pins 11 and 12 connected, the PWM signal is send to a MOSFET that is used to control 3-wire fans. Since you are using 4-wire fans, the third jumper should connect pins 9 and 10 as follows:

     

    By the way, the fan speed is calculated by this formula:

    RPM  = 6,000,000 / TACH_DATA

    TACH_DATA is a 16bit number made with the combination of registers:
    TACH-DATA-HByte
    TACH-DATA-LByte

    IF RPM reads INF, it means that TACH_DATA registers are zero, or that  no TACH measurement has been performed.

    Examples:

    (1) FAN is stopped or speed is below 91.5RPM
    TACH-DATA-HByte = 0xFF
    TACH-DATA-LByte = 0xFF
    RPM = 6,000,000/65536

    (2) FAN speed is 26,077 RPM or faster
    TACH-DATA-HByte = 0x04
    TACH-DATA-LByte = 0x01
    RPM = 6,000,000/260

    (3) FAN speed has not been measured (TACH reg. have their default values after reset)
    TACH-DATA-HByte = 0x00
    TACH-DATA-LByte = 0x00
    RPM = 6,000,000/0 = INF

    In order to enable the TACH input, these bits must be set:
    Configuration Register 0, Bit 0 (START Bit)
    Configuration Register 1, Bit 2 (TACH-EN Bit)

    Let me know if any of this info helps,

     Regards,

  • Rafael - The J5 11-12 jumper was the problem, and moving it to 9-10 solved it for me.  Thank you very much.

    Part of the problem is that the jumper settings for J5 9-10 and J5 11-12 are not very clear in the documentation.  Looking back in the document "SLAU195A–December 2006–Revised February 2009", Figure 9 on page 13 shows the same J5 jumper setting for both 3-wire and 4-wire fans.  I suspect this is an error with the document.  Can you look at it and confirm?

    Again, thanks for the support.  The EVM worked great in all four modes until today.  I will start a new thread for my latest problem.

    Cheers, 

    Michael

     

     

  • Hi, I am working on the evaluation board. I read your comment here, and I found it is something close to my problem. So I dont know if you can help me.

    I am using the 3 wire fan, but I tried to measure the direct pwm-out without the fun. The high level is only 200mv. I can see the output change correctly at the spin-up process. But only the 200mv high level output can not drive the transistor. So I ordered another AMC6821EVM separately without the USB board. I still got the same problem. I have measured the other pins of amc6821. Power, ground look fine. D1 and D5 are also bright. I cannot find something else could make the problem.

    Besides, I would like to use the Software RPM control mode. But from my experience before, I had problem about the tach-mode. After I set Configuration Register 0, Bit 0 (START Bit), Configuration Register 1, Bit 2 (TACH-EN Bit), and kept the tach-mode clear, I measured the pwm-out directly. I can see there are gaps between pwm signal with frequency. On this condition, if I read the tach-data, the value is not stable. So it definately will not work at the rpm control mode. 

    I wrote down what I have tried,

    With the tach-mode bit cleared,

    If I disable tach-en, or clear start, bit0 of CR1, then the pwm-out will output one continuous signal;

    But if I enable both, the pwm-out will be discontinuous with gap between the pwm signals (there is tach input).

     

    With the tach-mode bit set,

    No matter if I disable any of the two or not, the pwm-out will be always continuous without gaps.

     

    So it looks like when the tach-mode bit is set, the pwm-out is kept ON. (opposite to the datasheet)

     

    I dont know if you have succeeded to make it work at the Software RPM control mode. If so, could please supply your settings with the registers?

     

    Thanks a lot,

    Lei

  • Hello Lei and welcome to the precision data converters forum!

    I have used the software RPM control mode before and I hope I can help you, but before giving you detailed instructions, I need more information about your fan:

    Is it 5V or 12V powered?
    What is its PWM frequency?
    What is its maximum speed (RPM)?
    Is a 4 poles fan or 8 poles fan?
    What is its maximum power consumption?

    If possible, could you attach the fan’s datasheet with your reply?

    Thank you,

  • Hi Rafael,

    It is really nice that you have used the software RPM control.

    From my last message, I have 2 problems. 

    1. About my own design, The pwm-out is not continuous with the tach-mode cleared; (so I tried to go back to the evaluation board, but)

    2. About my evaluation board, the high level of pwm-out is 200mv and the after I write 0x01 to CR1  the pwm output become not continuous either;

    My final purpose is to make the software RPM control mode work on my design, but I think I should mention first that both the above issues happen also without the fan. 

    About the fan, I am using 

    SUNON HA40201V4-0000-C99.

    I am sorry I dont have the complete datasheet at hand. Maybe you know it. Or I will forward to you later.

     

    Thanks a lot for your help. I hope I can solve this problem soon.

    Lei

  • Hi Lei,

    Thanks for all the details, scope pictures, and fan info. All was very useful to get an idea about what is going on. About the fan’s datasheet, I was able to find it, so you don’t need to send it anymore. It is attached it in case you want to see it.

    1307.HA40201V4-0000-C99.pdf

    The non-continuous PWM is due to bad TACH readings keeping the AMC6821 looping. This is explained in page 26 of the datasheet.

    First, let me start with the possible hardware issues and then move to register setting recommendations.

    The AMC6821 pin 11(PWM-Mode) should be tied to VDD or left floating because 3-wire fans are typically meant to work at low frequencies. This is critical when trying to control speed as the fan’s TACH sensor may mal-function at high frequencies.

    In the evaluation board with 4-wire jumper settings, J1.4 (PWM output) has a 2.2kohm pull-up resistor. If the SUNON fan red wire is connected here, it low impedance should lower the voltage significantly. This could be why you see about 200mV when high. Therefore, the fan’s red wire should not be connected here.

    All 3-wire fans require an external NMOS or PMOS to work with PWM. The evaluation board includes an NMOS to ground for 3-wire settings (W5:3-2). Thus, the SUNON fan red wire must be connected directly to 12V supply. The SUNON is rated 12V according to its datasheet. Also, the SUNON black wire should be connected to J1.1 pin (FAN GROUND) that is controlled with the NMOS.

    For 3-wire fans, the PWM-Out must only be connected to a NMOS or PMOS gate. It cannot be connected directly to the fan’s power supply, this can damage the AMC6821.

    The following diagram shows how to setup the evaluation board for 12V 3-wire fans.
    The six jumpers for 12V 3-wire fans:
    W1:1-2
    W5:3-2
    W8:1-2
    J5:3-4
    J5:5-6
    J5:11-12

    For RPM control, I recommend doing the following:

    1. Write 0xFF to Reg. 0x01 (Reset the AMC6821)
    2. Wait 1.5ms
    3. Write 0x28 to Reg. 0x00 (PWMINV=1 for NMOS and SW-RPM-Ctrl mode)
    NOTE: The start bit should NOT be set yet
    4. Write 0x05 to Reg. 0x01 ( TACH-Mode should be “0” for 3-Wire fans)

    The maximum speed for the SUNON fan is 4700RPM+/-15%, so 75% of this speed or 3525RPM should be easy to get.
    TACH_SETTING = 6,000,000 / 3525 = 1702.1 = 0x06A6
     Let’s set this speed:
    5. Write 0xA6 to Reg. 0x1E
    6. Write 0x06 to Reg. 0x1F

    Then, we can change the start bit to 1 by:
    7. Write 0x29 to Reg. 0x00 (This starts the speed control loop)
    8. Wait a couple of seconds for the speed to stabilize

    Finally, register 1E and 1F can be changed again if desired to a different value
    Let say we want 50% of 4700RPM or 2350RPM
    TACH_SETTING = 6,000,000 / 2350 = 2553 = 0x09F9

    9. Write 0xF9to Reg. 0x1E
    10. Write 0x09 to Reg. 0x1F
    After this the speed will be changed accordingly

    For this SUNON fan, the minimum speed should be about 30% of 4700RPM or 1410RPM. This establishes the maximum value for the TACH_SETTING

    TACH_SETTING_MAX = 6,000,000 / (1410) = 4255.3 = 0x109F

    I hope this can solve the issues, if not, could you please send the schematic of your design?

    Thank you,

  • Hi Rafael,

    Thanks a lot for your detailed explanation. That would be a great help.

    The non-continuous PWM is due to bad TACH readings keeping the AMC6821 looping. This is explained in page 26 of the datasheet.

    I got the non-continuous PWM when AMC6821 worked on the Software DCY control mode. If it worked on Software RPM control mode on my design, the speed will either increase to the max speed or decrease to stop gradually. So I tried the software DCY mode first to see if the PWM stable. It is the garantee for a stable tach signal. Then I get all the scope pictures. At DCY mode, TACH feedback can still keep it looping?

    Both of the above pictures happened at the DCY mode, when there is no fan connected (no TACH feed back). When there is a fan connected. the gap becomes very narrow but at the same frequency. So I can see spikes after the transistor. If I read the tach signal from TACH_DATA (CR1=0x01,CR2=0x05), the value also jumps among very different values. Only when the tach_mode is set, the PWM can be continuous at DCY mode.

    s

    About the evaluation board,

    I am sure I did the same as you mentioned with the 6 jumpers. I have even tried to remove the transistor and measure directly at the pwm-out pin. But it is still 200mv. Strange!

    I know it is really hard for you to say where the problem is from. Probably I will order another new one. So maybe we can focus on the RPM mode.

    For the your recommended RPM control register settings, 

    I have to say I did not do the same calculations. Later I will try it after I get the new evaluation board first. Then I can go to my own design.

    For my own design,

    According to your comment, first of all I tied the AMC6821 pin 11(PWM-Mode) to GND. I don't know how critical it is. But that should be a problem.

    Secondly I am using 24V to drive the 12V fan, so there is one high rate resistor in the serial. The resistance is based on the measurement of the fan. And we want the fan always to connect the GND but not like the evaluation board, so we use 2 transistors instead. You can see from my schematic,

    Could you please make some comment on my design? I will try your RPM setting first on the new evaluation board later. I am really expecting to see it works

    Thanks again for your great help,

    Lei

  • Hi Lei,

    Sorry I didn’t clarify the TACH-MODE functionality before.

    When TACH-MODE is ‘0’, non-continuous PWM is to be expected because the PWM is forced “ON” during a TACH measurement.

    When using a 3-wire, TACH-mode must be ‘0’ because TACH measurement should be done when the PWM is forced ON. The gaps you see are used to measure TACH. These gaps should be high instead of low. This is because PWMINV is ‘0’. PWMINV should be ‘1’.

    The ON state or 100% duty cycle polarity depends on the PWMINV value. If PWMINV is ‘0’, ON state or 100% duty cycle is 0V. If PWMINV is ‘1’, ON state or 100% duty cycle is VDD. In your design and in the evaluation board, PWMINV should always be 1.

    The 10uF capacitor at the fan power pin can be a problem because it can rectify the PWM signal. In order to avoid this, the PWM frequency should be small and the capacitor size reduced. This is why the PWM-MODE should be tied to VDD or left floating to use small frequencies (30Hz) and avoid rectification. Also, small frequencies help the TACH sensor to produce cleaner pulses.

    Regarding the fan’s TACH output, the open collector expects a pull up resistor to the FAN’s supply. It is not recommended to use a pull-up to VDD at this pin. Three 10k resistors can be used to make sure that the voltage at TACH pin goes down when the fan supply is 0V during the OFF time of a PWM cycle.

    I also noticed that the PMOS (PMV65XP) is rated for 20V and using it with a 24V power supply can damage it.

    Since the fan rated current is 46mA, I believe that a 300ohm resistor in series with the 24V supply is a better option. The fan operational voltage is 6V to 13.8V, so it is better to have a voltage lower than 12V.

    I have been using a 12V delta fan with the evaluation board without any issues. The SUNON should work fine as long as the 12V supply is properly applied to the evaluation board.

    If you want to see the PWM signal, you don’t need to remove the NMOS. The PWM signal can be read from J5 pin 9.

    Let me know if I can help you in any other way,

    Thanks,

  • Hi Rafael, 

    It is very nice to get your detailed comment on my own design.

    Regarding the three 10K resistors for the fan’s TACH output, should I use 4.7K as on the evaluation board instead of the 3rd one to lower the voltage to the TACH In pin?

    Regarding the transistor's rate, there are one resistor and the fan in series, so 20V is OK?

    Regarding the gaps used to measure the TACH, I still have doubt. Even if they are high instead of low, would they also influence the stable fan speed?

    But I will try it on the evaluation board first. Many thanks to you.

    Lei 

  • Hi Lei,

    Regarding the three 10K resistors for the fan’s TACH output, should I use 4.7K as on the evaluation board instead of the 3rd one to lower the voltage to the TACH In pin?

    A 4.7K resistor produces a TACH signal at the AMC6821 TACH pin that swings from 0 to 2.3V with a 12V supply. Since you are using 5V VDD, I thought it would be better to increase the size of this resistor to have a TACH signal that swings from 0 to 4V.

    Regarding the transistor's rate, there are one resistor and the fan in series, so 20V is OK?

    The drain-to-source voltage is maximized when the transistor is OFF because it becomes a big resistor. Also, the resistor in series is insignificant when transistor is OFF. In OFF state, the FAN brings the drain voltage down to 0V, and the drain-to-source voltage becomes 24V. Your transistor might be working now, but I’m sure it will break down eventually.

    Regarding the gaps used to measure the TACH, I still have doubt. Even if they are high instead of low, would they also influence the stable fan speed?

    I understand you concern, but I haven’t seen significant speed changes using low frequencies (PWM-MODE = ‘VDD’).

    As you noticed, the 100% duty cycle gaps are long when the fan isn’t connected. This is because the AMC6821 is waiting for TACH pulses. If the fan TACH is connected, the TACH pulses are detected fast and the gap duration is very short.

    TACH sensors in 4-wire fans don’t need the gaps because their TACH sensors are always ON, but this isn’t the case for 3-wire fans

    3-wire fans TACH sensors turns ON and OFF with the FAN. Thus, 100% duty cycle gaps are required to make sure the TACH sensor is ON when reading TACH pulses.

    Also, a fake TACH pulse is produced during a PWM cycle, and this is avoided while reading in the 100% duty cycle gaps.
     
    In other words, 100% duty cycle gaps make sure that the TACH sensor is powered to get good readings and avoid fake signals. Please use TACH-MODE = ‘0’ with your fan. Also, use PWM-MODE = ‘VDD’ and PWMINV = ‘1’

    Let me know what happens with the new evaluation board.

    Thank you,

  • Hi Rafael,

    Today I finally received the new demo board this morning. I am very happy that it finally works with a low pwm frequency as you suggested. Thanks a lot for your help again.

    But I still would like to confirm on the RPM mode, the slowest speed is 30%? If I want to stop the fan, I have to change to the other mode, right?

    Regarding the demo board, I have a comment on the inversion gate, which can be bypassed by the pin 11/12 of J5. You also suggested to use the jumper for it. But if I bypass the logic gate, a very big noise will come to the system, everywhere even for the power. Then the board doesnt work at all. I dont know what has happened on your board. But I guess bypassing the logic  gate will make a loop (feedback). It seems not good. So I dont bypass the inversion and keep PWMINV=0(logically the same). Then it works at this way. 

    I will try my own design later. I hope I will have some more good news.

    Thanks,

    Lei

     

  • Hi Lei,

    It is great news that you got it to work!

    Regarding the 30% limit, yes you are right. If you want to stop the fan, you need to change, for example, to DCY mode and change the duty cycle to 0%.

    Thank you for pointing out the logic inverter. Previous revisions of the AMC6821EVM don’t have R10 and R16 installed, so it was required to use a jumper at pins 11 and 12 to send the signal to the NMOS transistor. My board doesn’t have those resistors, so I needed to use a jumper and PWMINV=1. Since your board is a new revision with those resistors installed, you setup is perfectly fine without a jumper in pins 11 and 12 and PWMINV=0. I’m sorry I didn’t notice that before.

    Good luck with your design and I’m looking forward to hear more good news.

    Best regards,

  • Hi Rafael,

    I think it is better for you to change the user manual. Or it will not work with the jumper on pins 11 and 12.

    Now I am looking at my design. I have modified my design as you suggested as below. But I found a problem about the tach signal. Because the transistor Q4 is not close all the time, the power for the fan is not always there. So is the pull-up power for the tach signal. Probably this circuit would not work.

    Regarding this, you wrote "Regarding the fan’s TACH output, the open collector expects a pull up resistor to the FAN’s supply. It is not recommended to use a pull-up to VDD at this pin." I don't know the reason. But if so, it seems not possible for my case (one resistor in series to share the 24V). Even if I put the resistor and the fun to the other side of Q4, the pull-up power will also jump between 12V and 24V. I don't know if you have any experience to use a pull-up to VDD. Or you think it is better for us to generate one 12V separately instead of this resistor divider set.  

    It is hard to find any useful information about this tach signal. I hope you can give me more suggestions.

    Thanks a lot

    Lei

  • Hi Lei,

    Regarding pins 11 and 12, I will make sure the manual is updated as soon as possible.

    I think the main issue now is that the voltage is going from 12V to 24V. This tells me that the capacitor in parallel with the fan is not able to protect the circuit from the current produced by the fan when it turns OFF. Since the fan is an inductive load, it is expected to produce current when it is turn off.

    With a big enough capacitor this current only increases the voltage a little bit, but with a 100nF, this voltage goes all the way to 24V. However, a big capacitor and the resistor in series with the fan can rectify the PWM, and this will not allow you to control the speed of the fan.

    Therefore, the simplest solution I can think off right now is to use a 1W 12V zener diode like 1N4742 or 1N4742A and change R5 to 200 ohm. This can regulate the fan supply when Q4 is ON and protect it from going to 24V when Q4 is OFF.

    Also, I don’t think you need to worry about the TACH signal during OFF time because, with TACH-MODE = 0, the AMC6821 makes sure that Q4 is ON before a TACH measurement is taken.

    Let me know if this circuit helps your design to work:

    Best regards,

  • Hi Rafael,

    I can understand your point, the AMC6821 makes sure that Q4 is ON before a TACH measurement is taken. But does the measurement happen during the whole monitoring period? It's 250ms at least. If the Q4 is not always ON, how can the controller count the right rotations? 

    I think VDD  is most stable, I don't know if you can explain why it is not recommended? It is hard for me to understand.

    Actually I have tried to debug on my old PCB design. Because it's not easy to change the Tach pull-up, so I just keep VDD as the Tach pull-up voltage but I have modified all the other part of the circuit as we have discussed. Now it is working. But I don't know if there will be any problems with the VDD pull-up. But it looks like the PWM frequency is the key point.

    It is nice that my design works on the right mode now. 

    Thanks for your great help.

    Lei

  • Hi Lei,

    It is great news that your design is working! Also, I appreciate the good feedback you provided us, and I ‘m glad I was able to help.

    As I understand, the AMC6821 measures the rotations during the monitoring time when using TACH-MODE=0, so it should measure the rotations only when Q4 is ON as long as the PWMINV bit is the right value.

    The fan’s TACH output is an open collector, but sometimes the fan manufacturers don’t isolate this output completely. This could mean that there is other internal circuitry connected to this pin.

    In the case of the SUNON fan, its datasheet mentioned that the pull-up voltage for the TACH out is the “FAN Vcc,” so I assumed the SUNON fan doesn’t have a completely isolated TACH output.

    Without a completely isolated TACH output and a pull-up to VDD(5V), current from VDD might go inside the fan circuits when Q4 is OFF and affect/damage the TACH sensor.

    Also, the voltage at TACH output might need a pull-up to the fan operational voltage to work fine, but since you have the TACH output working fine with a 5V (VDD) pull-up, I don’t think this is the case.

    You could try running the fan with a PWM signal overnight using a pull-up to VDD and see if the TACH signal gets noisier. If it doesn’t, it might mean that the TACH sensor is not affected by the VDD pull-up.

    Best regards,