First Proper CNC Router

[Neale]

As I slowly edge closer to the electronics step, I was wondering why we have to over spec a switch mode power supply.

The stepper drivers we use limit the current to the motors, so why do we need to make sure we choose a power supply that can supply say +20% more current? Surely a driver set to say 2A will only draw 2A from the PSU?

I'm sure there is a very logic explanation.

Well, there's a very technical explanation! Actually, the stepper driver does not quite limit current to the nominal value. In fact, it's often unclear what that nominal value means anyway - peak current? Average current? So the value that you set the stepper driver to will be somewhere near the maximum current drawn from the PSU but not necessarily exactly the value you expect. The other problem is that the stepper driver doesn't actually supply a steady current over the whole pulse, but chops it alternately on/off so that the average is around the desired value. So peak current draw might be higher than expected during the pulses, even if the average is around what you would expect. The SMPS must be able to cope with these short-duration pulses without going into "overload" when it might shut down altogether for a brief period, until the "overload" has gone away. This can be pretty catastrophic for the machine overall and losing steps might be the least of your problems. Hence the need to over-spec a SMPS just to make sure that it never goes into current-limiting or overload mode. I would be worrying if the SMPS is only capable of supplying 20% over nominal current - doesn't seem like much headroom to me. That's why the usual advice (even from people like Leadshine who should understand these things) is to use a "linear" power supply which is very tolerant of pulse current overloads.

This is a bit of an over-simplification, but the fact is that the current draw is very "peaky" and not steady, you are adding together the loads of multiple drivers without any idea of whether their pulse loads will occur simultaneously or not, and you have a power supply that is not capable of handling overloads gracefully. A linear PSU will be heavier, probably larger, more expensive - but more reliable. And they're easy to build - there are several "how-to" threads on the forum. If you go SMPS, over-spec the rating to make sure!

[Voicecoil]

The other option might be to add a decent size capacitor (maybe 4700...15,000uF) on the output of the SMPS to deliver short term current pulses without tripping the protection. I say "might" because some wimpy SMPS's are just so wimpy that they find it difficult to start up into a large capacitive load, so you'd be b******d from the start. However some do have a "soft start" provision that avoids this and such an approach has got me out of a hole once or twice before.

[diycnc]

Thanks for the explanation.

If you were going to use a SMPS, how much extra over the nominal current would be recommended?

[m_c]

In terms of current draw, there is no real reason to overspec a SMPS. Once it's overcome the initial start-up surge and charged up it's various internal capacitors, and the stepper driver capacitors, the average current draw is minimal at idle/slow speeds. At idle, with say a 5V 2A stepper motor, it should be seeing a combined total of 2A at 5V through either or both of it's windings, so 10W. Now say you've got a 2A 50V SMPS, that's a 100W source. The stepper driver (which is essentially a fancy SMPS), although switching currents might peak at 2A, the various capacitors will smooth that somewhat before it reaches the SMPS circuitry.
Even at higher speed/load where you happen to be in the optimum point to be applying 50V to get the full 2A through the motors, due to the switching nature of the stepper drivers you're not likely to draw the full 100W capacity. Plus once you get beyond the optimum voltage/current point, back emf will limit the amount of current the available voltage can push through the stepper motor windings, and power consumption will actually reduce. This is why it's recommend for a typical 3 axis machine, you only need a supply capable of delivering 2/3 the maximum combined current.

The real big problem with SMPS's is there inability to handle the regenerated energy from stepper motors/drivers as axes are decelerated. During deceleration all that kinetic energy gets dumped back into the power supply, which can cause it to go overvoltage. SMPS rarely have the ability to handle that, with the better ones simply shutting down (often until they get power cycled), and the cheaper ones releasing their magic smoke. Bigger SMPS will normally have bigger capacitors so they'll absorb a bit more energy before going over voltage.
You can get around this by installing a reverse energy dump circuit which will dump any regenerated energy to a big resistor rather than into the supply, but for the cost, you should be able to build a suitable linear supply that will avoid all these problems, be far simpler, and have less bits to potentially go wrong.

[cropwell]

dump any regenerated energy to a big resistor

Resistor ?