What are the major variables in determining savings?
The primary approaches to driving large arrays are series strings, parallel connections, and matrix addressing (multiplexing).
Series Strings of LEDs: When driving a large array of LEDs, you need to be concerned about current distribution and power efficiency. Series strings of LEDs are often used to improve both of these factors. Depending on the type of LED and the operating current, forward voltage can range from 1.1V (IR emitter at low current) to about 10V (some blue emitters at high current). However, when you operate LEDs in series, you can be sure that all of them in a string have the same current. Also, if you are working with a relatively high supply voltage, you can improve efficiency by connecting strings of appropriate length. For example, operating 6 IR emitters in series at 50mA DC will require about 8V, depending on temperature and the type of LED. For a 12V supply, this leaves 4V for switching and current regulating bias. Series strings have a couple of disadvantages: any LED failing open circuit will disable the entire string, 2) Any LED failing shorted will reduce the forward drop for the string, possibly affecting current regulation, 3) Compared to parallel connections, circuit board layout can be more complicated.
Connecting LEDs in Parallel: You can also connect LEDs in parallel. However, variations in the forward voltage requirements of individual LEDs will result in non-uniform current distribution. To minimize these effects, you can use any combination of several approaches: 1) Use individual current limiting resistors or regulating circuits, 2) Use LEDs chosen from the same production lot and/or matched for forward voltage, 3) Connect series strings of LEDs in parallel. This last approach has the effect of averaging out the forward voltage over several LEDs.
Matrix Addressing (Multiplexing): If you need to control the pattern of driven LEDs, as required for graphical and character displays, matrix addressing should be considered. Also, many ready-made LED arrays are designed for matrix addressing. (Lest you think that this is your only choice, be advised that may LED displays are actually driven by long shift register chains with individually current regulated parallel outputs for each LED.) Depending on the nature of your logic or microcontroller port outputs, you can choose common anode or common cathode connections. For example, common anode arrays have groups of LED anodes connected together, with each cathode available for a separate switching and/or regulating circuit. You could use NPN common emitter (open collector) switches or NFET common source (open drain) switches on each cathode. Further circuit details are beyond the scope of this topic. However, the important thing to note about multiplexing is this: Since each LED is only on part of the time, you will need to supply higher peak currents to achieve the same brightness compared to non-multiplexed DC applications. Among other things, this will mean that the forward voltage for each LED is higher, possibly much higher. For example, suppose that each LED in a multiplexed array is on 10% of the time. To achieve the same brightness as with 50mA DC, you will need to supply approxmately 500mA peak, and you can expect forward voltage to rise, perhaps as much as a volt. This may mean that the LED gets significantly hotter. For information about driving LEDs at higher currents, see How Hard Can I Drive LEDs?
Some additional details, cautions:
1. To maximize array output, obtain 'superbright' or 'extra super bright' LEDs.
2. LEDs have a negative tempco of light output and a negative tempco for forward voltage (usually ~-2mV/K). To some extent, these may be self compensating if you use fixed resistors for current limiting, but be aware that the above setup may draw more current as things warm up.