01-03-2021 05:38 PM
Can TMDS_33 signals be driven through regular LVC33 buffers like the 74LVC245? What would be the issues with doing this? This is for a point-to-point FPGA connect via a backplane. Is it possible to omit the 50ohm pullups on the inputs?
01-04-2021 11:46 AM
I'm not familiar with this level shifter, but it seems to be for single ended signals whereas tmds is differential.
TMDS also relies on a pull up at the inputs to function. So I'm not sure how this should work with a signal you're trying to level shift.
You should try sketch a diagram of this circuit you are proposing, try to understand the level shifter and how it might treat a differential signal like this. My guess is that this level n shift is not suitable here since its intended for use with single ended signals
01-04-2021 12:47 PM
I think all the differential signaling means is that the ‘P’ and ‘N’ signals are at opposite logic levels. When one of the pair is at a logic 1 the other of the pair is at a logic 0. Since this is just 3v3 logic I think a pair of single-ended drivers may be used. Since this is for distances measured in inches and I am thinking ordinary buffers may be adequate. Since the drivers max out about 200 MHz a somewhat slower interface will be used. Normally TMDS input buffers have open collector outputs meaning a pull-up resistor is required. But since the ‘245s are actively driving the bus I think the resistors may be omitted. I am not sure about actively driving an open collector input. Why the ‘245’s? It is easier to wire the signals for prototyping and buffers can be made bi-directional. I would like to have the bus driven using the ‘245 buffers as that may enable multiple modes of operation for the bus just by reprogramming the FPGA. If I cannot get the differential serial transceiver interface working I would still be able to use the board as a single-ended parallel bus without having to scrap the board.