Yeti – Electrical Build Part 8

I’ve got to the point where I have a stepper motor turning but I’m hesitant to move forward until I have the inrush issue fixed with the power supply.

I asked around to find out how other people have solved the inrush problem and got no end of different opinions. It seems inrush is a big problem but the solutions are either cheap and only adequate or expensive and complex. As it’s such a common problem I expected to find numerous off the shelf DIN mountable solutions that I could just buy but surprisingly there aren’t really any. Ok, that’s not strictly fair, there are off the shelf solutions but I wasn’t about to pay the sort of money they were asking – the off the shelf solutions were expensive because they were designed for much larger transformers that I have.

First off let me give a very quick explanation of the problem. When the transformer coil is first powered up the primary coil looks, to the incoming power, very much like a short circuit and therefore draws an enormous current. This shouldn’t really come as a  surprise if you think about it, without power a transformer is nothing more than a few long wires that happen to be physically close to each other. Fortunately for us the arrangement of those long wires in the coil causes an inductive resistance to build up in a very short space of time after power on. Typically a dozen cycles or so (lets say 0.2 seconds) is all that is needed for the inductive resistance to build to the point where the coil will start drawing close to it’s rated current. In fact the first cycle (1/50th of a second) is enough to build up over half the final impedance so this really is a fast action. As always reality is a little more complex than this explanation because the size of the inrush is dependent on the exact point in the AC cycle that power is applied. If you get lucky and apply power at the peak there is almost no inrush but if you are at the zero crossing point the inrush is limited only by the circuit resistance.

Ah ha, you think. I’ll arrange it so that I power on at the peak of a cycle. Unfortunately this won’t work for a couple of reasons. Firstly there’s no simple circuit that can find the peak of a cycle and secondly if the output of the transformer is rectified to produce DC the smoothing capacitors after the rectifier will draw a large current from the secondary windings as they first charge up. Ironically, from the capacitor charging point of view switching on at the zero crossing point in the AC cycle would minimise the inrush – sometimes you just can’t win. You maybe wondering if this is really a problem and the answer is “it depends” if you have a small transformer (<250VA) then no it’s not. The inrush will be small enough that your circuit should just be able to live with it. Up to about 750VA you’re probably better off just using a slow blowing fuse. Over 750VA though you are going to blow fuses or trip breakers on a fairly regular basis unless you do something about the inrush. A slow blow fuse will help but if you get unlucky and switch on at the zero crossing even a slow blow will blow. Even if you find a fuse that doesn’t blow though do you really want to be subjecting your £150 transformer coil to that sort of abuse at every switch on?

The simplest solution to inrush is just to place and NTC (Negative Thermal Coefficient) thermistor in series with the transformer. The thermistor is made of a special material such as a sintered metal oxide which has the unusual property of reducing in electrical resistance as the temperature increases (completely the opposite to most materials).

The basic idea behind this solution is that at start up the thermistor will provide a little bit of resistance to the circuit  and prevent the worst of the inrush from happening. As the thermistor to warms up due to current flow the resistance will lower until, in theory, it’s almost as if it’s not there. With careful selection it is possible to have the thermistor drop to maybe 5% of it’s initial resistance. Note that the resistance couldn’t ever drop to zero since some resistance is needed to heat the thermistor to lower the resistance.

The shiny starts to wear off this solution when you realize that to achieve that 5% of initial resistance you’ve got to get the thermistor up to a component scorching 130°C (working temperatures vary but they tend to be around there). Assuming you can find somewhere to put such a hot component you then have to deal with the heat it’s producing. While the resistance might only be 5% of it’s starting resistance that will only occur if you are drawing full power and that means a significant amount of current. You would certainly be looking at getting rid of several watts of heat at least which means active cooling. And then then there’s the problem of variable loads on your power supply. The thermistor is great if the load on the power supply is constant. The thermistor will warm up to a constant temperature and therefore give a constant voltage drop. If the load fluctuates though, as it may well do with a CNC machine, the thermistor will heat up and cool down causing a change in the voltage drop it causes. And finally there’s the problem of short power interruptions. If the incoming power supply goes off for a fraction of a second the magnetic field in the coil is lost but it’s not enough time for the thermistor to cool down and provide a resistance that will prevent a large inrush current – typically the cool down time for a thermistor is 90 seconds. This last issue may or may not be a for Yeti; the power for the power supply comes via the contactor which is latched closed by the 24V safety supply. A power cut would power down the 24V supply and therefore release the latch. The uncertainty about whether this would be a problem comes from the fact that the 24V power supply includes capacitors and until they have discharged it’s not clear that latch would be released.

A much better approach is to use a relay to take the resistive load out of the circuit once the inrush has passed. Notice I say resistive load rather than thermistor, there is a holy war between people who feel resistors are better and those who feel thermistors are better. Personally, I fall into the thermistor camp. The way I see it if the relay fails to close I’d like a fighting chance of having a component that won’t blow up or catch fire which is a distinct possibility with a resistor. Additionally, thermistors are designed to take the inrush shock which can be many joules whereas resistors generally aren’t – a large transformer can cause resistors to literally explode. The upside of a resistor is that if the relay doesn’t close it should burn out and fail open but you can’t be sure that it will fail open, worse you can’t be sure that it won’t get so hot it melts the solder which flows all over the board shorting things at random.

I had planned on having this article cover both the problem and the solution but this article is running a bit long already so the solution will be in the next part.