Soft Start Inrush Current Limiter- PCB design

[TStar]

Hi folks,
I hope you will be interested to take a look at my latest shared project at PCBWay. It's a Soft Start Inrush Current Limiter designed to be used with a toroidal power transformer up to 500VA or 12A. This should be sufficient to supply 3-4 NEMA 34 stepper motors.

https://www.pcbway.com/project/share..._026d2a0e.html


The original circuit design by Rod Elliot can be found here:

https://sound-au.com/project39.htm

Any feedback on the design and layout would be much appreciated



Thanks for looking!

[Muzzer]

Wow that is a massive overkill. You can simply get a timer relay and a metal clad resistor from CPC or Farnell to do the job. I used a Finder timer relay - I think it was this one last time https://cpc.farnell.com/finder/80-21...0timer%20relay

Worst thing is, Rod Elliott doesn't seem to understand what causes the periodic / random current surge when the transformer is connected. The effect is called "remanence". During normal operation, there is a fairly large AC magnetising current flowing in the primary due to the primary inductance, resulting in an alternating magnetising flux in the core. This is independent of the load on the output. At the instant when you disconnect the mains voltage, much of the magnetising flux at that instant remains (hence "remanence"). If you tried to dismantle the transformer at this point, you'd find the cores were magnetised and wanting to stay together. The magnitude and polarity of the remanence is random, as the interruption is not synchronised to the voltage waveform. When you now come along and reconnect the mains voltage to the primary, you superimpose a magnetising current / flux on top of the remanent flux. Depending on the instant at which that reconnection happens, you can now end up with a higher flux than you would ever see in normal operation.

The steady state magnetising flux of most transformers is actually very close to the saturation flux of the core, so it doesn't take much additional flux to saturate them, which is what causes the random massive inrush currents. The current during saturation is only limited by the voltage, winding resistance and the impedance of the mains network. And these large surges are a random function of the instantaneous voltages at connection and disconnection of the mains, which is why you only seem them from time to time rather than at every power on. You may have wondered why they only happen every so often.

To implement a resistive soft start on the transformer primary, you need to use a resistor that is small enough in value to drive the magnetising current close to its equilibrium value but high enough to limit the current if there is saturation during the first few cycles. This is why the soft start resistor value needs to be selected according to the specific transformer. For the range 1kW - 3kW, I have found values in the range 3 to 22 Ohms work well and the delay timer only needs to be set to a second or so.

[TStar]

Thanks for the feedback, Rod's design produces a much shorter delay, he says the optimum is around 100ms, just 5 cycles. With the 100K variable resistor my version should produce a range of 0-500ms. The combined resistance of the parallel resistor bank is 50 Ohms. This should only protect against initial inrush, and hopefully won't cause much heating of the resistors. It won't do anything about the problem of remanance though. I guess a solution to that might be a switching circuit which is synchronized to the AC? This would involve zero cross detection. I know there are relays out there which have built-in zero cross detection. It still wouldn't solve the polarity though, it could start half a wave out of sync.

[m_c]

SSRs typically switch at the zero crossing point, however that is the worst point to switch as it'll cause the biggest inrush, which is why they're a poor choice for switching transformers (plus the fact they generally fail short circuit isn't good from a safety perspective)

It only takes a couple cycles for the transformer to fully magnetise provided it receives enough power, but it usually takes longer than a few cycles for the power supply voltage to recover, which is why you want a slight delay.

Personally, provided the transformer isn't going to be switched off and on regularly, I just use a thermistor. Not technically perfect, but adequate and far less likely to go wrong if sized correctly.

[Muzzer]

One issue with a thermistor is the fairly long time constant for it to cool down again and until it does so, the resistance may be too low to prevent tripping the breaker. For an application like a welder, the transformer needs to be restarted frequently (controlled by the torch switch), so a fixed resistor works well. For my 300A single phase TIG welder with up to ~70A input current, the timer / resistor solution seemed the best solution.

For my CNC machines, the soft start event only happens when the machines power up. However, the operating load current varies significantly from idle to full chat. May not be an issue but I'd rather not have a large, continuously hot body in the cabinet.

[Muzzer]

Thanks for the feedback, Rod's design produces a much shorter delay, he says the optimum is around 100ms, just 5 cycles. With the 100K variable resistor my version should produce a range of 0-500ms. The combined resistance of the parallel resistor bank is 50 Ohms. This should only protect against initial inrush, and hopefully won't cause much heating of the resistors. It won't do anything about the problem of remanance though. I guess a solution to that might be a switching circuit which is synchronized to the AC? This would involve zero cross detection. I know there are relays out there which have built-in zero cross detection. It still wouldn't solve the polarity though, it could start half a wave out of sync.

Yes, another option might be a means of turning on and off at a defined phase angle, ideally ensuring the appropriate polarity.

A phase angle controller using a triac to gradually open up the duty cycle is the usual method used in industrial soft starts. Both methods are pretty messy / complex, whereas the timer and resistor method is a piece of wind to implement.

[the great waldo]

Hi Muzzer
Any chance of a quick schematic for timer relay and a metal clad resistor wiring ?
Thanks in advance.
Cheers
Andrew

[Neale]

It's interesting that the original design talks about using this kind of inrush limiter with toroidal transformers "up to 500VA." This surprised me as I have been cheerfully using toroidal transformers in CNC machines in the 500-650VA range for more than 10 years and never had any issues. Not so much as a single MCB trip. But then I see that the original design was for audio applications and I wonder if it is quite so relevant for stepper driver PSUs. At least, at the "home user" scale. Big commercial machines are a different kettle of fish.

There was a discussion on this forum a while back of capacitor sizing for stepper PSU use and learned discourses of ripple against capacitance, none of which took into account what level of ripple was acceptable, and this is where stepper and audio applications are very different. I put an oscilloscope on my router supply and ran some gcode to exercise all three axes. I saw around 10V ripple. And, frankly, while that sounds a lot on a nominal 68V supply, it doesn't make any practical difference to performance when the drivers are happy to accept 24V to 70V and are switching at rates way over the 100Hz ripple frequency. So, capacitors do not have to be that big. But for audio, I suspect that that would be entirely unacceptable leading to much larger smoothing caps and hence much higher inrush currents. Supply ripple may have audible consequences!

In short, I reckon that typical "domestic" stepper supplies do not need this level of sophistication. It won't hurt, but I wouldn't worry about installing such devices. Above, say, 1000VA, maybe it's a different story but I have no experience at that level. All I'm saying, I think, is that you shouldn't read across from one situation and assume that it applies in another very different one!

[Muzzer]

The issue is actually about the transformer cores periodically saturating when connected to the mains, something that happens on a fairly random basis as I explained. The tendency for them to saturate depends on how much magnetising flux has been designed in and how close that is to the saturation flux limit.

Some transformers run very close to saturation and are more prone to the effect than others. To increase the margin requires higher inductance which in turn requires more steel and copper and higher material cost. There's a commercial decision involved, which is one reason we see variation between different examples.

You may notice that apparently similar transformers from different suppliers behave differently - this may be seen as different steady state (unloaded) temperatures, for instance. I'm guessing your toroidal audio transformer doesn't run a high magnetising current / flux, which is why you don't see tripping breakers. I have 2 yellow site transformers that are both rated at 3000VA and are of similar dimensions, yet one sits at around 50C and the other barely warms up (perhaps 30C) when left powered but unloaded.

[TStar]

I am hoping to use this with a toroidal transformer to supply AC directly to the drivers, so I won't be using a capacitor. In a DC setup the capacitor provides a bit of inrush protection to the drives but without it I think this unit is probably a good idea.