TCA9548A: Defect serie of this compontents?

No.

Please show the schematic, and specify what the defect is.

Hi

Schematic is linked to in the bottom of this side about the module: https://wiki.dfrobot.com/Gravity__Digital_1-to-8_I2C_Multiplexer_SKU_DFR0576

As part of error searching I connected a esp32 wroom to the Gravity module (linked above), and 4x BME280 sensors. This worked. Then I desolder the TCA9548 and hand soldered a new one on under microscope. The new one was taken from tape. I only disconnected usb power, alle other connections was kept. With new TCA it did not work. Then I put the old TCA back on and it worked again.

I stille have 5 ic's in not  opened tape.

I get 10new from rs-online tomorrow. I will make above test again, and then replace on our own pcb's. Above test was to make sure that it was not something else in hw or sw. 

I have never seen more defective new ic before. I have been in production where 10^9 components where mounted....

Br David 

What exactly do you mean with "did not work"? Does it acknowledge writes to its configuration register?

All chips sold by TI are tested. Did you buy those chips from an authorized distributor?

All data/signal in is ok, no data/signal out. No contract with ic.

IT test when produced, not when sold. You are saying they are not produced by IT? Which do you thing is IT.

Hi David,

Happy to help here.

Looking at our internal system it looks as this our devices produced by TI.

However can you please explain more about the new device "did not work"? This is the same question Clemens asked earlier in the thread. Would you happen to have waveforms that can also explain what you mean here?

In addition did you test the other 10  units you received from ti.com? How did those test go?

Please let me know what you think and we can look into this further

Regards,

Kameron

With the default setting of the configuration register, no output would be the correct operation. But you are saying that the old chips work, so I have to assume that you are setting the configuration to another value. When you do this, does the I²C access work, i.e., does the chip return an ACK?

Hi Clemens

This is my test setup(Photo below). To get all other but the chip away, that could be the cause, I ended tested with brack out modules: The ESP32 is programmed to look for BME280 on all gates 0-7 and both addresses of BMW280. When the program finds a BME280 it write the data on the serial port. This works... with both this two setup, until I remove the org. TCA9548A and mount a new one. I only remove the USB cable nothing else, do the soldering (I am IPC-610 certified), and pluck in the USB back in again, and it dos not work. Then I replaced the new with the old again and it worked again. 

I cant see it different that a IC error. I still have 4 left under the tape out of 10. One of the coming days I will receive new ones. Based on this test, and the fact that I never have see series error on IC like this before, I expect the news TCA's to solve the problem.

Can there be different configuration register settings? I have not see register settings in the data sheet, please share a link.

Does your software check for ACKs?

Anyway, all TCA9548As behave the same. This looks as if your chips are damaged or counterfeit.