Pullups for SPI voltage level translator

Hi all,

I am using an 8-bit bidirectional voltage translator (TXB0108; http://www.ti.com/lit/ds/symlink/txb0108.pdf) to step down SPI signals from the Arduino to a display from 3.3V to 2.5V. My question is whether or not pullups are needed between the arduino and the voltage translator.

Here is the schematic of the translator:

I have a pull-down on the output enable pin as suggested in the datasheet. My understanding is that when the outputs are disabled, all pins are in a high impedance state, to prevent excessive current drain/sourcing while the MCU initializes its outputs. Once the output is enabled, I'm a little less sure. My reasoning is that , because there are inputs on the "low voltage" side (nets marked with _L) - i.e. those on the display - pullups are needed to keep the pins in a definite state when they aren't actively being driven low. But I'm not sure about the MCU -> voltage translator. Should there be pullups on this line also?

The device cannot tolerate pullups or pulldowns unless very very weak (greater than 50k). It only drives weakly in the steady state, with strong drive only on transitions. Its a smart device.

So if you are worried about initialization time use 100k pull-ups / downs and enable the outputs once everything is set up. Otherwise just hard-wire the enable on and it should "just work".

FWIW, I do not use pullups on this device on the 'downstream' side (in your case 2.5V). I do use pullups (10ks) on the 'upstream' side for the SS pins to make programming the chip via SPI easier. HTH