PF8200 input supply must sink current?

[adevries]

Hello,

I noticed that the PF8200 datasheet was recently updated to include this note on Table 9: "Input supply for switching regulators must be capable to sink current to avoid overvoltage condition during the power down sequence of the device". My application has the PF8200 being supplied by a power supply that can only source current, so I'm trying to understand the implications of this note. Here are my initial questions:

1. The switching regulators on the PF8200 are all buck regulators, so how is it possible that powering down the PF8200 causes the regulator's input voltage to increase above what the input supply is providing the PF8200?

2. Given that my supply for the PF8200 can only source current, how high should I expect the voltage on SWxIN to rise? Higher than 5.5V?

3. If I were to install a device that limits this overvoltage condition, how much current would it need to be able to sink?

4. Can this voltage rise be mitigated by increasing the input capacitance? If so, how much additional capacitance is recommended? I am currently following the recommendation of 4.7uF per SWxIN pin.

5. Does this situation only occur if the PF8200 is commanded to power down the voltage rails? What if the input supply for the PF8200 simply turned off? Would the overvoltage condition happen then?

[reyes]

Hi,

Please find my answers below:

1. The switching regulators on the PF8200 are all buck regulators, so how is it possible that powering down the PF8200 causes the regulator's input voltage to increase above what the input supply is providing the PF8200?

[A] problem is associated with VRE, VPRE cannot accept this current since voltage would rise and damage VIN node.  

2. Given that my supply for the PF8200 can only source current, how high should I expect the voltage on SWxIN to rise? Higher than 5.5V?

[A2] Typically, all power supplies operate in forced PWM, which in theory are capable of making sink current, only VRE operates in something similar to pulseskip which cannot have sink current, and of course LDO are not able to have sink current. So, unless you are using an LDO to supply the PMIC, it is unlikely that your power source is not capable of sinking current.

The issue occurs when VIN is operating at 5V and an overvoltage goes above 5.5V.

3. If I were to install a device that limits this overvoltage condition, how much current would it need to be able to sink?

[A3] You can probably use a clamp diode to prevent an overvoltage event on VIN pin.

4. Can this voltage rise be mitigated by increasing the input capacitance? If so, how much additional capacitance is recommended? I am currently following the recommendation of 4.7uF per SWxIN pin.

[A4] No, it is not recommended to increase the input capacitance.  

5. Does this situation only occur if the PF8200 is commanded to power down the voltage rails? What if the input supply for the PF8200 simply turned off? Would the overvoltage condition happen then?

[A5] If the input supply for the PF8200 simply turned off, the PMIC will not try to do its power down until the VIN voltage gets to UVDET (2.5V), so even if the sink current event occurs, it will generate an overvoltage event, but referenced at 2.5V, so it would not damage VIN.

The issue occurs when VIN is operating at 5V and an overvoltage goes above 5.5V.

So, it is not ideal, this solution can work, but we cannot recommended because the MCU would not have the proper power down sequence, which, could be prejudicial for the MCU.   

[adevries]

Hi Jose,

Have you been able to take a look at my previous response yet?

Thanks

[reyes]

Hi,

Yes, since you are using a Buck converter to supply the PMIC, you should not have any problem, this kind of regulators are PWM based, which in theory are capable of making sink current. But you need to check with the buck manufacturer to make sure.

You should not go above 5.5V.

I’m not referring to TVS, but to normal diodes.

When I refer to the MCU, in your case, I’m referring to the i.MX8M Mini.

You are correct, the recommended way to turn down the i.MX8M Mini is as mentioned din the image you sent. So, the solution of simply turn off the supply of the PF8200 can work, but, we cannot recommend it since it will not power down the MCU in a proper manner.

Regards,

Jose

[adevries]

Hi Jose,

Please find my responses below:

1. The switching regulators on the PF8200 are all buck regulators, so how is it possible that powering down the PF8200 causes the regulator's input voltage to increase above what the input supply is providing the PF8200?

[A] problem is associated with VRE, VPRE cannot accept this current since voltage would rise and damage VIN node. 

What does VRE and VPRE stand for? I don't see those words referenced in the datasheet and I don't know what they are.

2. Given that my supply for the PF8200 can only source current, how high should I expect the voltage on SWxIN to rise? Higher than 5.5V?

[A2] Typically, all power supplies operate in forced PWM, which in theory are capable of making sink current, only VRE operates in something similar to pulseskip which cannot have sink current, and of course LDO are not able to have sink current. So, unless you are using an LDO to supply the PMIC, it is unlikely that your power source is not capable of sinking current.

I believe my power supply acts like a pretty standard buck converter, similar to the diagram I pasted below.

Assuming my power supply works like the diagram shows, do you think it would be able to sink current? If so, what path would the current take? Vs would be greater than 5V, and the diode would block current current going to ground, so I'm not sure how it would sink current.

The issue occurs when VIN is operating at 5V and an overvoltage goes above 5.5V.

Okay, it's good to know that the overvoltage would go above 5.5V. Do you know how much above 5.5V it will go? Will it go above 6V? Because that is the maximum voltage allowed on SWxIN.

3. If I were to install a device that limits this overvoltage condition, how much current would it need to be able to sink?

[A3] You can probably use a clamp diode to prevent an overvoltage event on VIN pin.

When you say "clamp diode" are you talking about a TVS? or do you mean some other kind of device? I don't believe there exists a TVS that can withstand 5V without conducting and clamp the voltage below 6V during an overvoltage condition. If a simple device exists that can be added to prevent this overvoltage condition, that would be awesome.

4. Can this voltage rise be mitigated by increasing the input capacitance? If so, how much additional capacitance is recommended? I am currently following the recommendation of 4.7uF per SWxIN pin.

[A4] No, it is not recommended to increase the input capacitance. 

Understood, I will keep the input capacitance the same.

5. Does this situation only occur if the PF8200 is commanded to power down the voltage rails? What if the input supply for the PF8200 simply turned off? Would the overvoltage condition happen then?

[A5] If the input supply for the PF8200 simply turned off, the PMIC will not try to do its power down until the VIN voltage gets to UVDET (2.5V), so even if the sink current event occurs, it will generate an overvoltage event, but referenced at 2.5V, so it would not damage VIN.

The issue occurs when VIN is operating at 5V and an overvoltage goes above 5.5V.

So, it is not ideal, this solution can work, but we cannot recommended because the MCU would not have the proper power down sequence, which, could be prejudicial for the MCU.   

What MCU are you referring to? In my application, I am using the PF8200 to power the i.MX 8M Mini. Based on the processor's datasheet, it looks like all the voltage rails can be turned off at the same time. Wouldn't removing the power supply accomplish this? Please see images below for the processor's power-down sequence, and please note that I am not using rail VDD_MIPI_1P2, it will remain disconnected.

Thanks for your assistance.