I'm not convinced that it matters one bit whether the driver is driving 8, 16, 32, or 128 bits. What matters is how well the designer handles the system requirements and how well he selects suitable components.


Oh, but it is! If you know you have, at most, a 1/32 duty cycle when multiplexing yet can't exceed the maximal power dissipated in the driver when the current through the LED peaks at 1.5x the DC rating, well, you have to find another multiplexing scheme. If you have a registered SIPO register, then it remains static as long as you allow it to do so, and you can then draw current from each of its outputs for a period of time proportional to the rate at which you have to scan in order to meet timing constraints. If you have 16 outputs, there's no reason you can't draw current from each of them in pairs, or quads, if it meets power dissipation limits and provides the necessary brightness at that timing ratio.


Yes, you have to consider all of those, which is why you can't limit yourself to one or two multiplexing schemes, nor can you limit yourself only to use highly integrated constant-current drivers.


True enough, but you CAN gain lots of options by using discrete components to provide the current and use mass sinks, e.g. ULN2803, which are capable of handling considerable currents. As an option, you could use n-channel MOSFET's as low-side drivers to sink your column-current if you mux column by column, or two columns, etc. I've seen several versions of small-package MOSFET pairs or arrays that would be very effective. Moreover, since they're small (SOT-23 size or the like) and have an on-resistance well under 1 ohm, often under 0.1 ohm), they won't be heating up much, even when sinking well over an ampere.


There is no guarantee that a device for a given application can and will be "cheap", nor can it be guaranteed that such a display can be made "small". If you absolutely have to have a sign that will be readable at 50 meters in direct sunlight, you'll have to spend whatever is necessary, and tolerate otherwise objectionable size, else you'll have to accept compromised performance.



Sadly, that's probably true in the eyes of most designers, though the parts probably won't fail if used as specified. It's only when one allows violations of electrical, thermal, and other environmental specs that well-designed components will routinely fail.

If the display won't work with DC drive, then attempting to make it work under multiplexing is silly. Given that the sign can be useably operated with DC drive, it's likely that 2:1 multiplexing will work. Whether the anodes are driven in sets of 4 or sets of 256 shouldn't matter, if the LED current the anode drivers provide is constant, and the temperature rise/watt-second is within limits. If you have to have that sign under conditions that won't allow a "normal" design to remain within specificatios for the components you choose, perhaps you need liquid nitrogen cooling, or perhaps you need another approach. Your client, or whoever is going to pay the bill, will have to decide that.

If the ambient lighting is so strong that it must be "on" all the time, it's probably the wrong technology. Traffic signals have shown that LED's can be made to work under most lighting conditions with proper shading, but there are conditions under which, though they operate properly, they aren't very visible.

RE