AMC6821: spin-up noise issue

Hello CK

How is the AMC6821 being configured for PWM startup. Can you please share the configuration that is being done?

Hi Amit,

I am Scott, I co-work with CK for this issue.

Below is our AMC6821 setting.

0x00 - BIT5 | BIT4 | BIT3 | BIT0
0x01 - BIT2 | BIT1 | BIT0
0x3F - 0
0x04 - BIT7 | BIT5
0x20 - BIT4 | BIT3
0x23 - BIT6 | BIT5

other than these register, we didn't change it.

After temperture increate, the controller will setting PRM as below.

RPM - 0x1E & 0x1F -> 1900PRM
Low limit 0x10 & 0x11 -> 950RPM

Then, fan speed will change to 100% duty cycle.Note:the fan PRM is over 3000~4000RPM.

Due to fan speed generates noise, so we need to reduce/fix this problem.
We try to disable FSPD bit, it can fix this symptom, but it will cause Fan Fault detect no function.

Currently, we think AMC6821 should not trigger Spin-Up process feature(Please refer SPEC page.24), because fan just start to spin-up, it is not fan fault condition.

Is it AMC6821 know issue?

If yes, how to fix this problem?

Hello Scott,

Does the fan have a Tach output when no PWM is applied? I am trying to get as much information based on the device setting and description to get the issue that you are seeing.

Dear Amit,

Thank you very much for your support.

Now we just want to confirm that:

1. Weather the spin-up process is always applied when the fan starting from a stopped state??

2. If we disable the spin-up function (setting FSPD=1), will the Fan Failure Detection perform or not??
We found that if we disable the spin-up function, the Fan Failure Detection will not perform. Is it because that the Spin-Up Time Count is always zero??

Look forward to your reply. Thank you very much.

Hello CK

When the FSPD bit is cleared i.e '0', the Fan Spin Up process is disabled. That means it will go directly to fault measurement. The correct procedure must be

1. Set the Tach Limit
2. Enable the Fan and Spin Up process with a larger Spin Up Time STME2-0
3. Let the Fan Spin up correctly since there is an inertia for every Fan which it must overcome.

Once this happens the Tach pulses should have stabilized and not cause a fault or a noise.

Hello Amit,

Let me re-phase - just two questions:

1. When the fan is stopped and FSPD=0, does the default spin-up process always apply upon power on (Yes/No)? 
2. When we disable the spin-up function (FSPD=1), the FAULT PIN doesn't appear to work.  Is that expected behavior when FSPD=1?

Best, Ishaan

Hello Ishaan,

As per the datasheet flow chart and register description

1. When the fan is stopped and FSPD=0 the spin up process will apply at power on with the default values.
2. When the spin-up is disabled i.e. FSPD=1, the fault pin should work if the conditions for fault occur

Dear Amit,

That is what I confused. As per the datasheet flow chart, I think the fault pin should also work even if the spin-up is disabled.

When we set the FSPD=0 i.e. the spin-up is enabled, the FAN-FAULT# pin (pin8) can work well and will turn to low if the failure conditions occur.

But when we set the FSPD=1 to disable the spin-up process, the FAN-FAULT# pin (pin8) will not turn to low if the same failure conditions occur.

All the settings are the same. We only change the register FSPD.

Would you please help to clarify it??

Thank you very much.

Hello CK

So with FSPD=0 the function of the controller is as expected. However as was mentioned in the earlier post by another user, the noise is excessive which is what you are trying to solve.

CK Ho said:

After temperture increate, the controller will setting PRM as below.

RPM - 0x1E & 0x1F -> 1900PRM
Low limit 0x10 & 0x11 -> 950RPM

Then, fan speed will change to 100% duty cycle.Note:the fan PRM is over 3000~4000RPM.

Due to fan speed generates noise, so we need to reduce/fix this problem.

We try to disable FSPD bit, it can fix this symptom, but it will cause Fan Fault detect no function.

Currently, we think AMC6821 should not trigger Spin-Up process feature(Please refer SPEC page.24), because fan just start to spin-up, it is not fan fault condition.

1. What if instead of updating the RPM directly to 1900 RPM, it is phased into smaller rise? Does that help control the noise?

2. Also in the register at 0x23 why is not RAMPE bit being set to '1' for slower Ramp?

Dear Amit,

Yes, the excessive noise is the main problem we want to solve. And it can be solved if we disable the spin-up process. But we want to keep fan-fault detect function.

So I want to know why the fan-fault detection can not work when spin-up is disabled. I think it should work well.

By the way, the customer is using "software-RPM control mode". So I think the RAMPE bit in register 0x23 is ignored.

Below is captured from the datasheet:

Hello CK

Checking back on the datasheet the Fan Fault section Figure-21, the Fan Fault is generated only after the 5 spin up fails. So it seems that the FSPD=0 is required for the Fan Fault to work.

Since it is Software-RPM mode, can you clear the Configuration Register 0x04 Bit-5 (TACH-FAST = 0) to reduce the update rate and also apply the value 00 the DCY-Ramp Register 0x23 Bit 6-5 (STEP1 and STEP0) to see if it makes any difference?

Lastly, as mentioned in the thread earlier, sudden jumps to the RPM must be avoided to make the controller go to high duty cycle which would result in noise. Instead a stepped approach in software should be taken to update to the Target RPM.