TCA5405: Output IO structure

Hi Tsugumi,

Thank you for sharing the results of your testing here. These findings are interesting to me. I need to look into this more to see if we can get a hold more information on the internal structure. I will get back to you early next week with more information. 

Regards,
Eric Schott

Hi Tsugumi,

It seems the commonality between these two measurements is that there appears to be a current path from Vcc to GND that is not through the 10k-ohm resistor. Can you confirm that there is no other component on the board that would provide such a path? Is the Master in the figure removed? 

Based on the information I have found, it seems that most of the back-biasing current is indeed from the MOSFETs' body diodes. Because of this, any significant positive voltage difference between the Vcc and GPO pins will result in back-biasing to the Vcc line. Because of this, we do not recommend a configuration that would have a voltage applied to the output pins while the TCA5405 is unpowered. Because the outputs of this device are always active when powered, an external biasing resistor should not be necessary for most applications. 

Regards,
Eric Schott

Hello Eric, 

I am pretty sure I removed all components that could possibly create a path, but here are the list of components I depopulated on the EVM for you to confirm as well. 

List of DNP : R1, R2, R3, R4, R5, R6, R7, R8, R9, U1, U2, U3, U5, U6, 

Additionally, I ran the same test with our PCB which does not have the 10K pulldown from VCC to GND, and has no additional components on the VDD line, but this board still exhibited the 1V diode drop from OUT to GND. 

The unknown path could be from VDD to GND as you said, but additionally it could be from OUT to GND directly. Thus, I wanted to see the internal structure so we can possibly root cause the problem. If we need to take this to offline I am more than happy to do so. 

Sincerely, 

Tsugumi 

Hi Tsugumi,

From what I can see of the internal structure, I don't see any such path from Vcc to GND. The only section of the circuit that's not included in the section I reviewed is the ESD structure, but this doesn't seem like a likely place for such a path to exist either. 

Is this currently causing an issue that interferes with your design? I'm curious if we might be able to look into a alternative way to combat this while searching for an internal explanation in parallel. 

Regards,
Eric Schott

Hello Eric, 

If there is no such path from Vcc to GND, then the more I think there is a path from OUT to GND. Can you check the structure for the GPO and see if there can be a conduction path? Again, if you can send us the internal structure that will be helpful for us to confirm and verify as well. We can take this offline and sign any necessary NDA to obtain the information if necessary.  

We currently use this chip in our design where the Vcc gets powered off while the GPO could be exposed to external voltage and thus want clear understanding of the leakage path. 

Sincerely, 

Tsugumi 

Hello Eric, 

Any update on this item ?

Hello Eric, 

Thank you for the update. 

IC partially turning on and changing the current consumption through the 10k pulldown resistor does make sense to me. Do we have an understanding of what turns on at what voltage / current consumed by those partial internal circuits? 

Also, is this the reason for the 1V forward drop from GPO to GND when conducting diode test? I ran the diode test using DMM from GPO to GND and then used a second DMM to measure VDD voltage, but the VDD was around 1mV which makes it hard to believe that partial internal circuit were on at this point. Thus I think there still is an internal path from GPO to GND, but was there anything internal that could cause this? 

Additionally, if you have unencrypted SPICE model (I use LTSpice) for this chip that would be helpful for us to validate this chip further. 

Sincerely, 

Tsugumi  

Hi Tsugumi,

The device you're inquiring about was made over 10 years ago. A lot of the design files from that time aren't around anymore. I'm also not aware of TI making unencrypted models. 

Tsugumi Murata said:

Also, is this the reason for the 1V forward drop from GPO to GND when conducting diode test? I ran the diode test using DMM from GPO to GND and then used a second DMM to measure VDD voltage, but the VDD was around 1mV which makes it hard to believe that partial internal circuit were on at this point. Thus I think there still is an internal path from GPO to GND, but was there anything internal that could cause this? 

I'm not 100% I follow you on this explanation. The diode test you're referring to is the one from the original post correct? And you're saying you're seeing 1V measured from the GPO to GND but have a second DMM for Vcc to GND or Vcc to GPO? 

Have you been seeing any failures with the device during testing with our device? 

My guess is when you did your diode testing, and saw ~0.7V on Vcc, you were basically partially powering on the device's PoR circuit since 0.7V is about half of the minimum recommended operating voltage. This is likely causing leakage through the PoR since that circuitry is mostly CMOS inputs. 

Hello Bobby, 

Yes, I ran a diode test from GPO to GND using one DMM, and while having this on I measured the voltage from VDD to GND using a second DMM. I ran the test again and now I see 0.786V from VDD to GND when the GPO to GND is under diode test (with value of 1.116V) with the first DMM. When I remove the diode test , the voltage at VDD to GND drops down to 0.  This now makes more sense in that the diode test did power up the internal circuit of the IC. 

Hi Tsugumi,

Are you seeing any failures/issues with the device?

I currently still believe the extra leakage you're seeing is going from the PoR circuit since Vcc is partially powered thus leakage is going from Vcc into GND.

-Bobby

Hello Bobby, 

We do not have failures at the moment, but we were observing the reverse conduction during the validation of this chip / board with the design. 

If there is a limit to how much reverse current the IC can handle, please advice us so we can compensate for it. 

Sincerely, 

Tsugumi 

Hi Tsugumi,

Do you have an estimate of how much current flowing through the diode from GPO to Vcc? Currently your set up has a resistor from Vcc to GND, I assume it's not like that in your actual system.

-Bobby

Hello Bobby, 

Currently in the design, we are tweaking the resistor A and resistor B value and thus it will be nice to have the reverse current from GPO to VCC IC can handle. The 10k from VCC to GND does exist due to the other IC we have on that line. 

Sincerely, 

Tsugumi