Yes, I guess it is, in that respect. However, you can also think of it as a process which you start but which completes on its own.


well ... not exactly ... but you asked me to elaborate ... here goes ...

First, consider the individual LED in the array. Since you're multiplexing, you probably energize a row and then individually select the columns. Given, then, that you can energize a row in some way, to be considered later, let us assume, for now, that there's an ideal current source driving the anode of the LED, namely the row driver. The LED is illuminated as a consequence of current flowing from its anode through its cathode and onward. You drive this common cathode by means of an NPN (assume for the moment that it's ideal, i.e. can't be damaged) transistor with an appropriately chosen cap in parallel with it. The NPN collector is at the LED cathode, as is the positive end of the cap, and the emitter is at GND, as is the negative terminal of the cap. When the row is not energized, the only function the NPN can perform is to discharge the cap. When the row is energized, the NPN will turn the LED ON as well as providing a current path for the cap to discharge. Once the NPN is turned off, the only path for current from the LED anode to the GND is through the cap, which takes some time to charge. As the cap charges from the "ideal" curren source driving the row anodes, the voltage rises, causing a decrease in the amount of current, and, hence, a decrease in the LED brightness with increasing capacitor charge. The brightness decreases with time. At the same time, the retinal persistence on which multiplexing schemes depend, behaves as in the case where you've been "zapped" by a camera flash, in that it decreases with time as well. Hence, the cycle time has to be chosen such that this effect is minimized.

In reality, a resistor between the positive terminal of the cap and the collector of the NPN is required to protect the NPN from the current as it discharges the CAP. The fact that this resistor also can cause current to flow through the LED has to be considered when choosing it, but it's not critical, as the main function is accompished through the cap, and the "ideal" source which drives the ROW while the cap drives the COLUMN .

In keeping with the DRAM effect, I suppose one could choose a cycle time that would keep the current flowing all the time, even if not at a constant rate, in order to combat the apparent flicker that is present in most multiplexing schemes.

Consider, too, that each row has to be driven by its own NPN and capacitor. Keep in mind that most multiplexing schemes rely on the "on" time to regulate the brightness. This means that the same time is allocated to columns that contain LED's to be turned OFF as to those that have an LED that's turned ON. In this scheme, however, the brightness is controlled by the charge time of the capacitor in this circuit and not by the length of time the NPN is conducting, provided that it has been on long enough to discharge the cap. This introduces some flexibility not avaialble in most multiplexing schemes.

That "ideal" current source, however, can't really be an ideal current source. It must be an ideal voltage source, and an additional resistor in series with the LED, more specifically, with the column, must be introduced between the LED cathode and the row driver. That way, when many LED's are ON, they have adequate supply at their anodes.

Does that help at all?

RE