i'm using Diodes Inc. driver chip AL8861 to drive 2 LEDs per string.
they recommend 4.7 to 10uF cap.
per their data sheet:
"The minimum capacitance needed is determined by input power, cable‘s length and peak current. 4.7μF to 10μF is a commonly used value for most cases. A higher value will improve performance at lower input voltages, especially when the source impedance is high."

my input voltage can be as low as 10 volts. is that considered low input? LEDs are no more than 2" away from driver. peak current is 1amp.
i've been using 10uF CERAMIC caps but they have greatly increased in price.

when they say "improve performance", does that mean running at lower supply current?
1. can i safely use 4.7uF caps if i'm not concerned with maximum efficiency?
2. what happens if voltage rating is exceeded by 1 or 2 volts. a 25volt part that gets 26 volts

thanks

 

The supply capacitor is to insure that the input voltage is at least the minimum specified for the IC at all times – it could be that your circuit supply & impedance is adequate such that no capacitor is required (especially if your nominal input voltage is well over 4.5V). The application notes state that this capacitor is required because of the relatively high peak current demands, so if the supply is unable to deliver the high current peak (without excessive voltage drop) then the capacitor is a must.

It is the distance from the supply source which is important, not the distance from the output to the LEDs.

So if you are operating the IC at 5V then you could only afford a dip of 0.5V before the input voltage was under spec; but at 25V you would have a possible voltage drop of over 20V before being outside of the specification. But the performance of the device might be compromised by the input voltage rapidly changing from say 25V to 10V. – take a look at the data sheet graphical data for parameters shown for changes in input voltage.

If you exceed the capacitor voltage rating, you will reduce the expected life of the capacitor.

 

1. can i safely use 4.7uF caps if i'm not concerned with maximum efficiency?

Electrolytic capacitances over 10μF generally have wider tolerances. Hence applications will generally take this into account. You should not see any difference in performance between 4.7μF and 10μF.

2. what happens if voltage rating is exceeded by 1 or 2 volts. a 25volt part that gets 26 volts

You degrade the performance, lifespan and possibly risk having the capacitor explode.
Circuit designers will intentionally derate their components, for example, use a capacitor rated for 25V when only 16V is needed.

 

Aluminum electrolytic capacitors can be operated right at their rated voltage and will withstand short term over-voltage. I'd have no reservations about operating at 25 V device at 24 volts, but I probably wouldn't operate at 26. In most cases a part rated for a higher voltage at the same capacitance will have a higher "ripple current" rating because the "equivalent series resistance" (ESR) is lower. ERS is a big concern in switch mode converters.

Solid tantalum capacitors are the only type I know of where it is common to deliberately derate them for operating voltage, particularly at elevated temperature. The general applications info published by the big manufacturers like Kemet will cover this.

Film capacitors can generally be operated without voltage derating but the current waveform can be an important consideration. Film caps aren't suitable in this case because they are so large relative to capacitance and votlage rating ("CV product").

Ceramic capacitors are a problematic lot. Many suffer from a large negative voltage coefficient of capacitance. A 10 V part operated at 10 V might have capacitance of less than 25% of its nominal value. One rated at 25 V and operated at 10 V will still be lower than nominal, but not by as much. "X5R" and "X7R" types are generally much better in this regard than "Z5U" or "Y5V" (the difference is due to the actual ceramic material used).

I'm surprised you say ceramics are getting more expensive. The trend for a long time in surface mount parts is very significant reduction in prices. High-value through-hole types used to be and I believe still are much more expensive than the equivalent surface mount part.

If you already have a ceramic capacitor of at least a couple of microfarads in place, then adding a good-quality aluminum electroltyic cap in parallel, if you have the space, is a reasonable thing to do.

The reason efficiency drops a bit if the source impedance is higher than desirable is that a switch mode regulator has a "negative input resistance" characteristic - if the input voltage falls, the current rises to maintain the same output power. Lower input voltage almost always translates into somewhat less efficiency in the converter circuit.

"Buck" converters are fairly demanding of the input capacitor. When the switch is ON, current flows from the input, charging the inductor and delivering current to the output. When the switch turns off, the input current drops to zero but the output current continues to flow because of the energy stored in the inductor. This "chopping" of the input current makes more demands on the capacitor, mostly in terms of ESR, than some other topologies. The RMS current at the input is high relative to the average input current, and it is the capacitor in lots of circumstances that bears the brunt of this. If the input connection wires are long, their resistance makes the problem worse because the capacitor has to supply a larger part of the current when the switch is on, then charge again when the switch is off. Be sure to keep the input wires very close together - twisted together or otherwise tightly bundled. This reduces inductance which can also cause problems.