MP9572 doesn't work at low temperatures

Hi everyone!

I’m using MP9572 step-down converter in my PCB design.
The parameters as follows:

  • Vin= +16V…+32V;
  • Vout=+3.2V;
  • freq=800kHz.

My device works fine at normal and high temperatures with input voltages from 8 to 36 volts.
But I have encountered issues when testing the PCB at low temperatures (below minus 30 degrees Celsius). The MP9572 cannot deliver sufficient power and and the device fails at start-up. I also noticed that the situation worsens when the input voltage decreases. For example, some devices start normally at 24 V supply voltage, but will no longer start normally when the voltage is lowered. The lower the voltage, the more devices will not start normally.

To test step-down converter I cut off the power consumers and connected power load. I used electronic load and load resistor.

The test result when using electronic load and input voltage of 24 volts:

  1. At normal temperature the converter can provide:
  • up to 2.8 A current in case of slow load increase;
  • up to 1.8 A current in case of instant load connection.
  1. At low temperature (minus 40 degrees Celsius) the converter can provide:
  • up to 2.9 A current in case of slow load increase;
  • only 0.15 A current in case of instant load connection i.e. a 12-fold reduction.

I also used load resistor (1.5 Ohms with long wires). The ammeter showed about 1.7-1.8A at +3.2V in normal conditions. After freezing the current drops to 0.6A and output voltage drops to +1…1,5 V.

All capacitors are X7R. I tested them separately and found OK (capacitance and ESR remain almost unchanged when frozen).

In an attempt to solve this problem, I tried the following:

  1. Replace the 10 μH inductor with 6.8 μH and 15 μH.
  2. Removing the 2 Ohms resistor in the bootstrap circuit and using Cboot = 0.11 µF (two 0.22 µF in series).
  3. Adding capacitors to the input and output.
  4. Float CCM pin to switch from FCCM to AAM mode.
  5. Changing frequency to 1350 MHz and 588 kHz.
  6. Connecting input voltage directrly to input capacitors of the converter (i.e. bypassing the input EMC filters).

All of the above did not have any significant result.

I’ve also tried to replace MP9572 by MPQ4573 but it works about the same, no significant improvement.

Can you suggest how to fix the problem? I’m out of ideas.

Some times at low temperatures you can get condensation on the board, which could provide a partial short between pins causing erratic operation. You could try heating the board up and then conformal coating it to keep any moisture out.

Thank you for answer!

  1. I have one PCB with conformal coating and its behavior is no different from the others.
  2. I have five PCB and they all behave the same at low temperatures.

I solved - R26 has to be 22 Ohm.
I used some “cold spray” for my testing and I noticed incorrect start-up
When we have normal conditions - startup looks like in datasheet
When MP9572 is frozen:

When I put 22Ohm(R26) startup is always correct. Tests more PCB in the freezer have confirmed it.

I don’t understand why it happens. MP9572 has operating temperature -40…+125 degrees. Why does MP9572 work incorrect in cold temperature?
I’m thinking about replacing MP9572 in the next my projects.

Wow, good job that sort of thing is hard to find. I think you get better tech support on MPS Now, over here you end up speaking to internet weirdos. This looks like a new chip and you may have blundered into a bug, and a bug fix. Do you have a local rep to talk to?