TCA9535: Default state of I/O ports on startup // bypass capacitor value

Hello,

The ports should be initialized as inputs after the device completes its power-on reset.  For this to work properly,the VCC ramp behavior should be within the guidelines given in Section 10 of the datasheet.  Can you please confirm that this is the case for your application?

Typically a 100 nF decoupling capacitance on VCC is sufficient.  If the rail is particularly noisy or prone to glitches you may consider a higher value.

When you see one of the ports in an unexpected state, does the corresponding control register have a value that matches?  Or, does it still show the port configured as an input but the behavior is observed to be an output?

Regards,
Max

hi,

the vcc ramp up from 0 to 3.3V is about 40 msec. For the decoupling, we are using a 0.1uf and a 10 uf. I didn't read the register configuration because it is in default state with no access, SDA and SCK are both pulled up at 3.3V. The behaviour is that once every 10 attempts on powerup, we see the pin drive at 0. (with a 10k pull-up on 3.3V). The specification specifies that all io are configured as inputs on startup. 

thanks for the reply,

Hi Gaetan,

Are you able to get a high resolution/zoomed in scope shot of your Vcc ramp when the output pins(like P03) is driving low?

Thanks,

-Bobby

Hey Gaetan,

It looks like the Vcc is sitting around 400mV before it actually powers up. Is there some kind of back biasing occuring on the Vcc line?

Are you able to do a case where the device powers up from GND? (Maybe place a ~100k pull down resistor on Vcc)

I have seen problems before where if the device starts at a potential higher than GND, it can sometimes get a bad power up though this is the first time I've seen it happen with this device....

-Bobby

Hey Gaetan,

I was just following up on this thread, were you able to address the issue from your original post or get the information for BOBBY's question?

Regards,