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.

PCF8574A: High consumption and heating - latch-up?

Part Number: PCF8574A
Other Parts Discussed in Thread: LM2841

I'm having a problem in my company which I think may be related to a latch-up condition.

The system which I'm debugging is composed of 1 MCU, 2 IO expanders (PCF8574A), LEDs and displays, plus some memories and transceivers.

In some boards, just after the system is powered up, one of the IO expanders draws a lot of current (~ 500 mA at 3.3V) and heats up a lot.

Each one of its 8 outputs is connected to a PMOS. The IO expander itself is not directly connected to any low impedance load.

This IO expander is located close to a buck switching regulator (LM2841). The problem never occurs with the second IO expander, which is farther away from the regulator. This one also is not connected to any low impedance load.

The problem seems random. The greatest part (~70%) of our boards are working fine, in some boards (~20%) it happens intermittently and in some others boards (~10%) the problem happens every power on.

I have already done:

- The INT pin was left unconnected, I've connected it to 3.3V through a 10k resistor.

- Added 100 nF caps close to the IC, on VCC pins.

None of these modifications have solved the problem.

The next thing to try is to insert 10k pull-ups in all the IC outputs. Currently, each output is directly connected to a PMOS, without pull-ups or pull-downs.

I'm suspecting this is a latch-up condition, causing overheating. Has anyone ever seen a situation like this with this part? What have you done to avoid it? Any suggestions?

  • Hi Luis,

    Thank you for providing a detailed description of the problem. This is a strange one - I haven't heard of something like this related to this device. Once the device is power cycled, does it return to normal operation? If you swap the device from a consistently failing board and a device from a consistently working board, does the issue tend to follow the I/O expander? Have you tried probing the device's I/O and power pins to check for any overvoltage conditions that may be occurring (e.g., due to noise coupling from the nearby switched-mode power supply)?

    Regards,
    Max
  • Hi,

    I was able to minimize the problem, and I'm reporting here because it may be useful to someone.

    Apparently it was really a latch-up condition affecting the IO expander, mainly due to poor PCB layout design.

    The board has a LM3841 switching regulator, converting 12/24V (automotive supply) to 3.3V, located very close to the IO expander. The big problem was that the GND trace connecting the buck regulator diode to other regulator components was long, and the IO expander GND pad was connected to this trace. Consequently, the noise on trace was affecting the operation of the PCF8574A. The noise could be easily seen with a scope on the IO expander I2C and output pins.

    Even the boards that were working good were in fact noisy (but a little less than the ones which failed constantly).

    The problem was solved by inserting a jumper wire between the diode and the input/output capacitors, creating a more direct and lower impedance connection between them.

    Fortunately, this fix could be inserted on completed boards without much trouble, and the next step is to have the PCB redesigned for the next batches.

    This situation just shows how important it is to have a properly designed layout.