LSF0102

Hi Guy Lemire ,

VrefB is required to operate the part and bias the internal FET to control the device apart from being required for pullup . If unidirectional voltage is concerned , you can consider LV1T family of parts . It works with single supply with high to low or low to high active translation .

Hi Shreyas,

I appreciate the response,but my question is if I can power VerefB and EN with a 1.8v supply instead of a 3.3V supply, but still have B1 and B2 operate with a 3.3V (off board) pullup. This will enable me to remove a 3.3V supply from this area on my PCB.

regards,

Guy

The recommended values of VrefB indicates that there needs to be 0.8V min difference w.r.t VrefA (VrefB >VrefA ) to operate .

Hi Shreyas,
this parameter does appear to preclude me from using this part with lower voltages as I wished. Thank you for the response

Hi Guy ,

I am sorry but the spec on the datasheet recommends to have that difference between Vrefa and B .
I have tested some units in the lab with VrefA equaling VrefB and works alright , you could try in your setup to have VrefA and VrefB and use 3.3V pullup .
In any case , if you are using the 3.3V and 1.8V on your board , will it not be easier to have it power up VrefB with it ?

Hi Shreyas,
my reason for asking this is so that I can run both VrefA and VrefB off of a 1.8V rail and not need 3.3V. I am finding that the majority of my power supplies for my PCB designs are migrating to 2.5V and below, with many now at 1.2V-1.8V, so if i can remove a 3.3V supply that would help in my layout.

The 3.3V pullup is located on another PCB through a PCIe edge connector.

Regards,
Guy

Hi Guy ,

Since TI should bind by the datasheet , the recommended conditions needs to be followed .
If you want to try with same Vref A and Vref B you can try but it may or may not work always and you should accept the reliability issues if any .

Hi Guy ,

You can also consider some transceivers which will work with VccA and VccB being the same but the output cannot be pulled up using resistors . It kind of defeats the purpose of the transceivers actually.

Guy,

Do you need bi-directional translation here (ie is data going from 1.8V to 3.3V side and vice-versa on the same channel?).  If the answer is 'no,' I would recommend looking at the LVxT family of translation parts.  You can do 1.8V -> 3.3V translation easily with SN74LV1T125

This does not defeat the purpose at all......I am translating from an open drain at 3.3V to a 1.8/1.5V IO, so I need a translator chip...the who idea gets back to my original question..using one supply (1.8V) to power the translator chip versus two (1.8V, 3.3V)

regards,

Guy

Guy,
Please correct me if I'm wrong, but from what all I've read so far, it sounds like both lines are uni-directional translation from 3.3V (open-drain output) to 1.8V. This could be accomplished with a pull-up resistor to 1.8V on the open-drain line and no translation device between the two. Am I missing something?

Hi Emrys,
both the CLKREQ# and WAKE# are by PCIe spec 3.3V open drain signals. the pullup resistor is usually on the root complex (mother board)

Because I am using an fpga with 1.8V IOs, I cannot expose my fpga to a 3.3V signal. I need a bi directional translator because I have to both monitor and control each of the signals.

Not sure I understand your comment about using a 1.8V pullup on the open drain..

regards,
Guy

Hi Emrys,
thank you for the reply. Please close this case.

regards,
Guy