This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TXS0108E: corrupting signals, could it be broken ?

Part Number: TXS0108E
Other Parts Discussed in Thread: ADS1298, TXB0104, TXB0106



I use a TXS0108E for interfacing an Arduino Leonardo (works in 5V) and a sub-board featuring two ADS1298 ADC (work in 3.3V). The TXS is mounted on a HVV-221 adafruit-like board that has two capacitors between the 5V (resp, 3.3V) rail and the ground. It is used to shift the voltage on the SPI bus, chip selectors and other stuffs.

A simplified schematics is joined, note the transistor used to make sure that the HIGH level on the Output Enable pin is 3.3V and not 5V. (It is hastly drawn, my apologies)

Up until Friday, it used to work like a charm and I could properly pilot the ADCs and retrieve 500 samples per seconds of data from them.

This weekend, I let the whole system powered on (the PSU also powers ancillary systems that need some time to stabilise) and when I resumed data aquisition on Monday morning, the Arduino only sent 0xFF's to the computer.

I unplugged the sub-board, pulled down the MISO to the ground via a 8.2k resistor and proceeded to record signals on the SPI bus via a Saleae Logic Pro 16 signal tracker and this is what I've got :

In each case I'm trying to read the identification of the ADS1298 : I'm sending three bytes (read registers starting from REG_ID ; read 1 + 0 register ; then a 0x00 and the answer should be sent on the same time.)

On the first picture, the level-shifter is also unplugged from the Arduino and the signal is directly taken from the Arduino pins. You can see 3 * 8 clock clicks on the SCK as expected, on the MOSI : 0x20; 0x00; 0x00, and on the MISO, only 0's as nothing is plugged in. So everything is normal.

Now if I plug in the level-shifter board and take the signals on the 5V side, here is what I've got : the MISO follows the SCK, which explains the 0xFF's that I recorded earlier this morning.

Finally if I take the signals from the 3.3V side, this is what is recorded : only one pulse out of eight on the clock and two pulses on the MISO.

I don't understand what happens. Could it be that the TXS0108E be damaged from being powered on for almost 3 days ? I welcome any lead to help me restore last week's state !

Thanks in advance !

  • Hey Julien,

    Are you able to provide a schematic that includes the passive components around the device? Please also capture the waveforms with an oscilloscope as logic analyzers may not tell the full story of what is really going on (will not show any signal integrity related issues). 

    For QSPI signals, we typically recommend the TXB family devices (such as the TXB0104, TXB0106). 



  • Hi Jack,

    Thank you for your fast answer. The schematics of the passive components around the device is just as simple as it gets, but I'm unfortunately not able to provide you the values of C1 and C2.

    I injected a 2MHz (the frequency of the SPI. I could lower it but it has worked like that last week), 0-5V squarewave between B8 and the ground (ch1, green on my oscilloscope) and measured it at A8 (ch2, blue), here is what I've got

    (sorry for the poor quality of the photo with the reflexion and all)
    On CH1 (input), I have Vpp = 5.6V (due to the harmonics I suppose) and Vrms = 3.58; on CH2, I have Vpp = 2.88V and Vrms = 2.1.

    If it happens that the chip is dead, I will try to find some TXBs as you suggested. Thanks again !

  • Hey Julien,

    Thanks for capturing the waveforms. It looks like the output on CH2 isn't reaching 3.3V as we expect. Just to make sure- there aren't any pullup/ pulldown resistors used on the the I/Os correct? (The TXS family devices have an internal 10k pullup, so the use of external pullups/ pulldowns will contend on the output). 

    Also make sure that the total combine loading condition (length of PCB traces, connectors and cables + other parasitic capacitances) around the I/Os are kept under 70pF, as well as using a shorter ground loop on the o-scope probe to clean up the signal rising/ falling edges.