How much fudge factor on power for a project?

[A-sic Enginerd]

This is the first time I'll be designing the power supply circuit for a project.

I can crunch all the numbers and determine the max that the supply should ever see, but wondering what may be the norm for designing in some fudge factor?

NOT running off a batt so no need to be super critical about low power usage and all that. Planning on doing a linear supply so as to keep the supplies that much cleaner to the circuit.

So toss me some numbers. What do you typically use? 10% over, 50%, 100%, more???

I'm just barely getting started on this side of the project so I haven't at all looked on what it would take to build margin in. Guessing it shouldn't be too bad though as the design is relatively low power by nature. In fact, the hardest part (for me) is simply the number of voltages needed. (analog and digital supplies @ 3.3v, a +5v, and a split rail +- 15v)

[Zero999]

Sorry I don't understand the question.

What do you mean by fudge factor and "the max that the supply should ever see"?

[A-sic Enginerd]

If I calculate that the max current a design should draw is X amps (or mA, etc.), how much extra should I build into what the power supply is capable of delivering?

Yes, the question itself assumes some margin should be built into the supply.

[Zero999]

It depends on whether you want any room for expansion?

If the answer is no then there's no need to bother, just design the power supply for the maximum current. In the real world when cost is a factor, power supplies are often not designed to meet the peak demand continuously, for example an audio amplifier may be able to put 100W RMS into an 8 Ohm load for a second but it won't necessarily be designed to do it continuously because a normal music source won't do that.

[ejeffrey]

One important thing is to make sure the supply behaves sensibly under fault conditions, especially an output short circuit.  The easiest thing to do is make sure that the short-circuit current is enough to blow the fuse cleanly. 

[A-sic Enginerd]

One important thing is to make sure the supply behaves sensibly under fault conditions, especially an output short circuit.  The easiest thing to do is make sure that the short-circuit current is enough to blow the fuse cleanly.

Good tip. Basically - put some error handling into the power supply. Hadn't really thought about that. I'll definitely make a conscious effort to keep that in constant consideration.

[Zad]

Keep in mind that the steady state current is not the same as the transient load seen on power up. Power supplies usually have sufficient thermal mass in the heat sinks to cope so long as the transient is short enough. Of course don't forget to de-rate the power supply if you are likely to run in significantly hotter ambient temperatures. At a minimum, unless you know you are going to be able to manage power quite precisely, and can guarantee your loads and temperature control, then 100% over-capacity is not out of the way. 500% is not unheard of in systems that require long term reliability.

[DaveW]

For derating a standard linear regulator, the normal rules for mil aerospace are:
Minimum dropout voltage * 1.2 < Vout - Vmin < 0.75 * Maximum voltage drop across regulator
Junction temperature < 0.75 * Tmax.

Make sure all the calculations are done over temperature and you'll extend the life of your circuits, and no need to add on extra power capacity if you're not planning on adding more circuitry later.