PCA9539:

How exactly are you measuring the impedance? What is the state of all other pins while you are doing this?

We measure with a Fluke multi-meter

All others output are high impedance (OL)

Hi Emmanuel,

What power is being supplied to the device during this test? Have you confirmed the configuration of the registers by reading registers 6 and 7?

Best,

Danny

All I/Os have clamping diodes to GND. 32 Ω is plausible for a diode.

What is the purpose of your test? What problem are you trying to solve?

Hi!

We did not isolated the component to measure the power.
The card is functional.
The question for us is: can we considered the component compliant despite these impedance differences?

Emmanuel,

Apologies for any confusion with my question. I am just trying to confirm whether the device is powered or unpowered when you're evaluating it.

When unpowered, we don't have a particular specification for this pin's impedance to GND. If it is powered, the values in registers 6 and 7 would be relevant to my response.

Best,

Danny

Danny,

Ok, my apologises.

So, the device is unpowered when we measure impedance with multi-meter.

and the device is powered In functional tests.

New data, on one product, we measure a very low impedance (18 Ohm) 
on PIN N°6 (PO2) to GND. In this case, the high level voltage is too 
low (about 1V instead of 3.3V) and the product don't work.

Danny

With 10KOhm resistor to GND on our schematic, on cards discarded from production we measured (unpowered)

18 Ω, 62 Ω, 33 Ω, 43 Ω, 17 Ω, 30 Ω, 22 Ω, 47 Ω, 30 Ω, 32 Ω, 31 Ω, 35 Ω, 27 Ω, 44 Ω, 37 Ω, 35 Ω, 39 Ω, 30 Ω, 23 Ω, 10 Ω, 56 Ω,

And the corresponding voltage (powered)

1.18V, 2.56V, 2.04V, 2.24V, 1.88V, 1.92V, 1.56V, 2.36V, 1.84V, 2V, 2V, 1.92V, 1.84V, 2.28V, 1.96V, 1.88V, 2.16V, 1.8V, 1.56V, 0.68V, 2.52V,

Should these components be evaluated?

Emmanuel,

Based on your information I'm not exactly seeing what's wrong with the devices.

When measuring these values, what is in registers 2-7?

Best,

Danny

Danny

We are not in the prototype phase. The product is 5 years old now and we have made more than 10K cards with the PCA9539PWR.

We have never seen this problem before. It was detected by our in-situ test which checks all the components of the board.

 At start-up, we initialize the task that manages the IOs in the real-time OS

The output registers (registers 2 and 3) are at 0xFF by default, we write 0x00 in both.

At the next period of the real-time OS clock, we write 0x00 in the two configuration registers (registers 6 and 7) to configure all the pins of the two ports as outputs.

The two Polarity Inversion registers (4 and 5) are at 0x00 by default and are never written.

Then we alternate at each period of the real-time OS between updating the outputs of the IO Expanders (O0.2=1).

Best regards

Emmanuel

Danny

Can you confirm that the only difference between the PCA9539PWR and the TCA9539PWR is the supply voltage which can go down to 1.65V for the TCA?

BR

Emmanuel

Emmanuel,

Thanks for providing your software steps, now I'm a little clearer on what's going on.

So, to restate the concern concisely: When configuring the PCA9539 to set its outputs as output LOW while powered and active, the PCA9539 outputs only appear to exhibit approximately mid-rail outputs. Does this sound like the correct understanding? And at this moment, the registers are reading as:

If you were able to confirm your writes by reading the registers after the write command, and they all show the expected values as above but still were not pulling low, then yes, this would be unexpected.

How often are you seeing this? (i.e. out of how many tested units?) The PCA9539 has not had any design change since its release. And I should also ask, were these devices ordered from TI or authorized distributor? Or were these obtained from a different source?

And regarding your second question:

Emmanuel MURARD said:

Can you confirm that the only difference between the PCA9539PWR and the TCA9539PWR is the supply voltage which can go down to 1.65V for the TCA?

The TCA series of our I2C devices are created in newer process technologies. When they share the same numerals (such as PCA9539 and TCA9539), the TCA device is specifically designed as a drop-in for the PCA component, however it would not be proper to say they're "identical" aside from the voltage supply range difference. The TCA devices are designed to meet or exceed the performance of their PCA counterparts.

Here is an FAQ that also addresses this question: https://e2e.ti.com/support/interface-group/interface/f/interface-forum/815758/faq-what-is-the-difference-between-an-i2c-device-with-the-family-name-pca-and-tca

Best,

Danny

Danny 

Thank you very much for your prompt answer.

Yes we reading the registers after the write command, but still were not pulling low.

We have 21 products on which we see this, with 2 different dates code.

These devices were ordered from ARROW ELECTRONIQUE that we believe to be reliable.

Best regards

Emmanuel

Emmanuel,

Ok, that's very helpful information, thank you for sharing. Would you be able to share a schematic for the IO pins and their connections? If you'd prefer to email me about it, please reach out to the email address in my profile Danny Bacic.

Additionally, for these 21 units, these were detected out of approximately how many total units? Just trying to get an understanding of the number of known good units versus units in question.

Best,

Danny