Hi,
Erik, I see.
But in my humble opinion, with C51 the maximum limit of refresh rate is defined not by physical factors but on MCU productivity.
For example, there are LED display drivers from Allegro micro (shift/parallel registers with some additional features) which may work even at 10MHz. But which C51 may provide such rates?
I try explain a problem. My current project uses C51 working at about 37MHz clock (18.432 in X2 mode). It has alot of interrupts:
- T0 for refresh LEDs and lamps;
- T1 for time-sensetive inputs;
- T2 for control PWM of PCA (sound);
- PCA OVR for input/output ports;
- CAN OVR for network communication ticks.
As well there are others (non-periodic) interrupts (INT0/1, four PCA inputs, UART and CAN macro).
Here is the table which indicates main "time gluttons":

INT     Freq,  Period  ISR          Average
        Hz     us      duration,us  MCU load
T0      1500   667     242          36,3%
T1      400    2500    35           1,4%
T2      18000  55,6    13           23,4%

Rest ones take all about 1%.

Now it means that above 60% of MCU time is paid for interrupts and only 40% is kept for main stream of program. Indeed, I can increase refresh rate twice up to 3kHz but then program will have only 1,5% of time to main execution.
So this example shows that the upper limit of refresh rate is defined with MCU productivity and has nothing with limits which come from physical realization of LED device. It is true at least for most C51s (if only you do not use 100MIPS devices of course :)

Regards,
Oleg