TPS22958NDGNR intermittently failing to turn on

Thread Starter

robotDR

Joined Mar 17, 2020
68
I have three boards and 3 switches per board. 1 Each for 1.8V, 3.3V, and 5V.

4/9 of the switches wouldn't turn on, out of the box. So I replaced them and got some working. now I'm down to one board that still has two failing. I replaced again and they now barely turn on. If I keep power cycling they eventually turn on sometimes. Sometimes not.

All have 5v VBIAS. CARRIER_PWR_ON is 1.8V but even when I use 3.3V it doesn't change anything.
I have changed C218 (controls rise time) to 1000 pF but that doesn't help.
I can't imagine I'm using it wrong. I ahve accounted for the 330 mA max inrush. I only have 16 uF max output capacitance (on 3.3v) and my rise time is plenty long enough.

Is there something I'm missing?
Are some parts just really fragile and this is a bad design from TI?
Datasheet: https://www.ti.com/lit/ds/symlink/t...tps%3A%2F%2Fwww.ti.com%2Flit%2Fgpn%2Ftps22958

1634860633801.png

Thank you for taking the time to look.
 

Thread Starter

robotDR

Joined Mar 17, 2020
68
Also would like to add:



The working switch (VIN is 1.8V), the voltage on CT pin is 9.17V

The failing switches the voltage on CT pin is ~0.6V -0.75V

Side note: how does the chip get the voltage up to 9V with Vbias = 5V and VIN=1.8V?
 

Papabravo

Joined Feb 24, 2006
19,572
Why do you have a capacitor on pin CT? A: You want to control the slew rate. (Really??). 10,000 pf is pretty big for that application.
A: You can use a charge pump to make a higher voltage. See section 9.2 - Functional Block Diagram
 

Thread Starter

robotDR

Joined Mar 17, 2020
68
Why do you have a capacitor on pin CT? A: You want to control the slew rate. (Really??). 10,000 pf is pretty big for that application.
A: You can use a charge pump to make a higher voltage. See section 9.2 - Functional Block Diagram
Thank you for the response.
I wanted to control in rush current. Even if that's unnecessary, it surely shouldn't stop the device from functioning.
I just again replaced with two fresh parts and these ones are working perfect. correct rise time expected from data sheet.
 
Top