TXB0108: Output voltage level issue

Hi Dylan, 

Thanks for the support. The only voltage level issue I see is on pin3 (A3) which doesn't have any pull-down resistor. All other pins have the correct waveform and don't have such an issue. Do you still see it's worth removing all the pull-downs? 

The waveform shows that there's something pulling the line down -- could there be an internal pull-down on something connected to that line?

The line is connected to Jetson Nano's SPI MISO directly (SCH attached). If it's a MISO input, I would assume it's Hi-Z. 

Shane Chang said:

I would assume it's Hi-Z

I'm certainly not an expert on that device, but glancing through the datasheet I saw a mention that there are weak pull-up and pull-down resistors on all of the pins -- and it indicates a "Power on Reset" state for the MISO pin of "PD" which I can only assume means "Pull Down" -- so it seems that it does include one. I'm not sure what their definition of "weak" pull-downs are, but you could pretty easily calculate it based on the waveform shown and the known 4k resistance in the TXB translator. I did some quick back-of-napkin math { 1.4 = 1.8*R/(4000 + R) } and figured it's about a 14k resistor.

There may be a way to disable that through software - unfortunately that's not my area of expertise.

Hi Emrys, Thanks for your reply. I have disabled the pull-down in the PoR setting but it didn't make any difference. That makes sense because that's for the PoR phase and the issue I'm seeing is during active phase so it's unrelated. I will try to confirm the 14k today by adding an external resistor (perhaps another 14k) to the line to see if the voltage settles to a lower level. Curious, will capacitive load on the trace cause the voltage drop as well? Let me know if there's any test you can think of for debugging? Thanks

Have you tried this same test on a second board?  What about swapping the part to a new one? Could the trace be shorted to something else?

Yes we have seen this on a second board. There's no part-to-part or board-to-board variation. 

I'll ask Dylan to take a second look, as he's our translation expert, but in my experience when we see something like this, there's something overlooked on the board. 

Thanks. BTW I attached a 14kohm external pull-down resistor to the line and the voltage was pulled down further to 1.1V. So the calculation of the weak pull-down of 14kohm is correct. 

Hey Shane,

If you refer to Figure 10 in this app note, you will see a similar waveform to what you captured. Any issues with DC levels will typically come from a resistive load contending with the 4 kohm internal series resistors (typically pull-downs as digital inputs are high impedance). Capacitance won't affect the DC performance rather the AC performance (rise/fall timing). If you can isolate the I/O from everything, this will a good way to determine.

Will go over the app note. Thanks Dylan

In the meantime, is there a level translator IC you can recommend that has a stronger output drive than TXB0108 ? 

Any unidirectional or direction controlled translator will have strong output drivers. Auto-bidirectional devices have the trade-off of not having a strong output driver.