Would this be a good DC inrush limiting circuit?

So I tried powering my little 48 VDC 10-20A electronic experiments and gadgets with a 48V LiFePO4 battery pack.

I initially got giant sparks when connecting the battery. I avoid the sparks by connecting a resistor in series, then manually bridging the resistor with a wire with some crimped-on connectors after a few seconds. That works, but it is of course rather cumbersome.

So I thought of building a simple relay circuit that would do this automatically.

Here’s the circuit:

Battery 48VDC (actually 55V) comes in. Inrush limiter resistor in series with the plus lead. Relay contact across the resistor. Relay contact rated 24V 60A. Relay coil with resistor in series, connected across supply after the limiter resistor. The relay coil resistor will be calculated such as to drop the voltage to the correct value (coil voltage and current TBD, but rather lower than the 48/55V supply). An electrolytic cap across the relay coil, calculated to give an appropriate relay closing delay, say 1 or 2 seconds. (I know how to calculate the resistors and caps.)

Would this be a good way to do it? Any tips? Improvements? Simpler ways to do it? Safety considerations? 50VDC won’t kill me but what are the risks of something blowing up?

Alternatively, is anybody aware of any ready made Chinese little gizmo on Amazon that I could use? So many great little Chinese gizmos on Amazon these days but I couldn’t find anything suited for this exact application.

That’s your basic inrush current limiter. It was used in my Phase Linear 200 power amplifier.

Thanks for the tip!

I searched for schematics of that amplifier to see if I could learn something. I came across a related article at this link

Apart from the differences due to the scenario, I noticed they used a Zener diode for the coil voltage and another diode, I guess for the back EMF.

Is it safe to use just a dropping resistor without the zener? Is the back EMF diode (if that’s what it is) needed?

There’s no main power switch? If not, why not?

I didn’t put one in for the time being. Will make one later.

It wouldn’t have helped either, unless I’m misunderstanding something? The sparks would have been inside the switch, would have worn them out or maybe even welded them eventually.

So these are just a bare gadgets that have 48VDC input terminals. I have a battery pack that has 48VDC output terminals. I connect them together. Tried two different gadgets. Exciting sparks both times. As opposed to when I use a 48VDC mains power supply instead of the battery. In which case of course I connect it first and then turn it on (with the mains switch).

I might add, the gadgets in question can actually start up with a minimal current and very low voltage too. Apparently there are just some large input filter caps. The inrush limiting resistor can actually be quite large to get a nice gentle startup.

<Checks OP’s handle>

Electronic gadgets.

Right.

Haha, that’s funny!

Gotta change my handle to Gyro Gearloose.

A good design will have a power switch. (Pay attention to the DC current rating of the switch, not the AC rating.) An even better design will take measures to minimize arcing between the contacts, of course.

Shunting a series resistor with contacts from a time-delay relay is one of my favorite solutions for inrush current, so I am glad to see it works for you. :slight_smile:

Instead of shunting the resistor with relay contacts, another approach is to shunt the resistor with an SCR. It would likely be a smaller, lighter, and cheaper approach, but the SCR could be dissipating around 20 W for a 20 A load.

Yet another idea is to simply install a series inductor. (No relay is necessary.) But there are a couple potential problems with using an inductor: a) It’s not a good solution for pulse loads, and b) there is a potential for LC “ringing,” but this can be minimized by choosing the right inductor.

For a sophisticated approach, google “soft starting circuits”.

Yes, I’m sure I got the idea from your post to my earlier thread. Even though at that time the need hadn’t occurred yet (I got the battery later). So thank you :slight_smile:

It seems indeed the ideal solution for my scenario. Easy to build up out of those few cheap components. Foolproof. Should work perfectly.

Well, a better solution might be to use a NTC thermistor instead of a resistor. The thermistor will start off at high resistance, go to low resistance after a couple seconds, and then the relay contacts will close across it to take it out of the circuit. I like this approach because the system will still function if the relay fails. Just something to consider.

You can also just use an NTC resistor directly. You can find them made for this purpose. Here’s one brand, for example. The B57127P0509M301 model has a current rating of 20 A; it starts at 5 ohms at room temperature but dips to 0.033 ohms at peak load. Not as efficient as a relay, but simple and effective.

Oh, definitely; that’s what’s usually done. But some people are uncomfortable with the running temperature of the thermistor, hence the additional relay.

The relay contacts across the resistor (or thermistor) is a nice solution, but not without it’s drawbacks, of course: more expense, more weight, more volume, and the relay coil dissipates some power. It’s a fine solution for home projects and industrial devices, but not for mass-produced commercial items.

Partially riffing off of Crafter Man, for loads that are not heavily pulsed, I am fond of inductors (in particular ferrites, which have a steeper impedance versus frequency curve than an ideal RL inductor like an dielectric core coil) in parallel with diodes.

Laird and Murata both make cheap low DCR (1-2 milliohm) ferrites with 10-12A ratings - you could parallel a couple of those to get your 20A. A diode in parallel with the inductor(s) both gives you a soft start and tends to damp LC ringing. If your load is moderately pulsed, you probably want an electrolytic cap downstream of the parallel ferrite/diode combination to reduce the pulsed component of the low down to no more than equal to the quasi-DC current of 10-20A.

I don’t really like relays, although you can’t beat their DC resistance: moving (especially bouncing) parts are a Bad Thing, and need their own snubbers, etc.

Thanks for the comments everyone!

The NTC idea is interesting. The off-the-shelf power supply that I also use has that solution on the AC side. Those do run surprisingly hot under load.

My relay solution will not actually switch current in practice. The load starts at only milliamps, apart from that turn-on zap. By the time the relay closes, there will be only millivolts across the resistor and milliamps through the circuit. Also the current will have long ceased by the time the relay opens (because of that cap I have across the coil).

I did wonder about snubbers. Do I need a snubber diode across the coil? RC snubber across the contacts?

ISTM the relay coil current in my circuit never changes abruptly. It rises slowly and drops slowly as that electrolytic cap charges and discharges. Or is THAT bad for a relay?

The current should be higher than that before the relay contacts close. What value resistor are you using? I’m thinking it might be too high. (Instead of trying to figure out the “ideal” resistor value, use a NTC thermistor across the relay contacts as mentioned previously.)

If the relay coil is energized with DC (which I believe is the case), then you want to put a reverse-biased “freewheeling diode” across the coil.

For your application I would not put a RC snubber circuit across the relay contacts. You don’t need it with the resistor (or thermistor) in parallel with the contacts. Reason… when the contacts close, the voltage across the contacts will be small, and when the contacts open there won’t be much of a spark due to the load being capacitive.

Thanks, will do that!

You know, I was wondering about that.

I haven’t built the auto limiter circuit yet. Just getting ready to order the parts. Currently I use the manual resistor shunt method described in the OP whenever I connect the battery.

One of the loads I am powering is this programmable buck converter (also representative of my other gadgets).

It powers up in an “off” state. I then set voltage and current (real neat, can do that with bluetooth on my PC!) and “turn it on”. I can turn it back off as well. So this off state represents the minimum circuit load, which is also the starting load.

I just measured the minimum load to be 12.5mA.

As for the resistor, my intial thought process did call for going with the biggest value consistent with the circuit starting up, possibly hundreds of ohms or even a kilo-ohm. (Calculated upper bound with that current would be 3k5.)

When the sparks happened, I didn’t have any such resistors lying around. But I did happen to have some low value resistors. I wired up a couple in series to get 5 ohm, which I am now using.

I hoped it would at least reduce the sparks.

To my surprise, no sparks whatsoever with only 5 ohm!

So I did conclude hundreds of ohms was not necessary. I was about to settle on 50 ohms. (At 55V, that’s about 1 Amp into a short circuit.)

If you don’t mind my asking, how would an inrush limiting resistor be too big?

Why should there be a minimum current before the relay closes?

The 12.5mA is the given minimun load of the circuit, although once the relay is in, its coil current will add to this (24V coil, some tens of mA I guess?)

If you want to see another reference design, something engineered by Toyota for the Prius, check this one out.

You just need some relays, a big resistor, and a microcontroller to do the sequencing. As well as whatever circuitry is needed to drive the relays - packaged SSRs have this in them. As he explains, you technically need just 2 relays, 1 resistor, and of course the microcontroller. (a few arduino pins is all you need) I would use perf board for the backplane to mount your circuitry.

If the resistor has too high of a value, there will be arcing across the relay contacts. If the resistor has too low of a value, you’ll get arcing at the battery terminal or power switch.

Instead of trying to determine the optimum value for the resistor, simply put the relay contacts across the NTC thermistor referenced by Dr. Strangelove. Close the contacts a few seconds after power up.

OK, that makes sense.

Thanks everyone! Great info and tips. I may go with the NTC (great to have the datasheet, lots of info there! And found a place to buy too) or perhaps a 10 or 50 ohm resistor.