Minimum Data Frequency

...

For a display refresh frequency of , I must scan through all rows times a second. This gives me a minimum row frequency of

This tells me that I must display a new row at least times a second, or, taking the reciprocal for the maximum period that each row may be on

I am interested in finding both a reasonable minimum data drive frequency for my LED drivers, as well as a reasonable minimum PWM greyscale frequency. I will first focus on the data drive frequency, by making the assumption that, within the I have for each row before progressing to the next, the time taken to configure the output for that row should be significantly less than the on time for the pixels—ie . Making the assumption that this 'significantly less than' is an order of magnitude (acknowledging that this setup can be done whilst the previous row is still being shown), we find that

This tells us that all serial data to configure each row before it can be displayed must be completed in less than µ. Consulting the datasheet for the TLC5951DAP, we see in the timing diagram below that, for five drivers, it takes

This gives the inequality

This tells me that, in order to achieve a refresh rate of with my scan rate, I will need a GSSCK data frequency of at least . This is encouraging, as the selected TLC5951DAP driver supports up to , affording me roughly a factor of within which to find an optimal balance between refresh rate/visual flicker and signal integrity/computational demand.

Taking this maximum data drive frequency to find the maximum achievable refresh rate , we find

This is encouraging, as it tells me that my matrix topology/scan regime should be practical to produce a flicker-free display. I should also remember that I can make my data transmission into the common shift register as soon as I have latched the previous data. Assuming that the data transmission can instead take half of the time that each row is displayed instead of , we find the corresponding maximum refresh rate to be

with a corresponding data drive frequency of

For completeness, to find a reasonable minimum PWM Pixel Brightness greyscale frequency, I will make the assumption that I want at least an 8-bit (counts) resolution to my PWM counter. This requires intervals over the period for which each row is on, ie

This is again encouraging, and suggests to me that I am likely to be able to make use of the auto display repeat mode as I am likely to be driving the PWM greyscale clock at a significantly faster rate.

For the full 12-bit resolution on my greyscale control, I will need to fit at least counts within each row-on period, giving