Transformer inrush limiting

[wobblycogs]

Hi all,

For my first electronics project I've built a simple linear power supply to provide 70V to run steppers on a CNC machine I'm building. The circuit I've built is this:



The transformer is a 1000VA with 230V primary and twin 24V secondary in series. The bridge is a 35A, 400V part. Smoothing is provided by 40,000uF for electrolytic capacitors.

The problem I'm having is with the fuse. I started off with an 8A cartridge fuse but it blew about 50% of the time on start up so I switch over to a 6A Type D breaker. The breaker stays in most of the time but I get the occasional nuisance trip. From what I've read I think the problem is inrush current while the magnetic fields establish in the coil. If I understand it correctly the current can be huge if the AC voltage is near 0V.

Someone has suggested using slow blow or time delay fuses (I believe this http://uk.rs-online.com/web/p/cartridge-fuses/5371313/ would be an appropriate part) but looking at the spec I can't see how that would be any better as it lets go under 50A surges. I'm guessing the coil must be pulling 100A+ on start up under the worst case conditions.

So I did a bit of searching and I came across this page http://www.ametherm.com/inrush-current/transformer-inrush-current.html which discusses putting an NTC thermistor in series with the power supply for the coil like this:



Is this an appropriate part? http://uk.rs-online.com/web/p/thermistors/2118002/

I'm not too worried about wasting a little power but I'd rather not dump loads of heat into my power supply case if I can avoid it. Using that 10 ohm part as an example though it looks like it has the potential to dump a lot of heat. Assuming it self heats to 85degC then the resistance drops to 1.5 ohms giving a 6.45V drop in the supply. At the maximum current  for the coil of 4.3A that equates to 27W of heat! In reality the coil will never draw 4.3A as I've over spec'ed it but I'd like to make sure it's safe if it does.

I'm open to suggestions for other solutions too :-), cheers.

[SeanB]

Use the NTC inrush limiter and use a contactor to close a contact that bypasses it , fed from the secondary rail via a RC curcuit to give a small delay of around 2 mains cycles. Capacitor of around 2200uF fed from another bridge rectifier directly across the coil on one of the 24V rails with a series resistor of around 20% of the coil DC resistance will give a delay enough to not trip the breaker. 24VDC contactor with a single contact of 15A will do, you could easily find a plug in octal relay that will be adequate. A single shot thermal fuse in series with the NTC and touching it, in a woven glass fibre sleeve, opening at 100c will provide protection if the relay fails.

[digsys]

Done a lot of work with this problem - and there are 2 issues conditions to be aware of:
1/ IF you ONLY want it for a single clean Turn-on no less than every 60 secs say, then the solution is simple -
Put a series resistor in-line with the mains, app the same resistance as the Primary. Then wire a Relay across it.
Have any sort of simple timer to switch ON the Relay after a few cycles. It's a common approach.
The wattage doesn't need to match the load, as it will EVER be on for a few cycles.
Edit:
2/ IF you want to INCLUDE brown-outs and drop-outs, ie 1-2mS > 20mS example, it becomes a LOT more difficult !
NOW, you need to make sure you do NOT switch back IN again, on the wrong state of the transformers last
magnetisation. ie was it a +ve cycle going -ve through 0V or vice-versa, or a +ve cycle going over the crest, both
+ve and -ve. This is trickier, and you need to use FETS / SCRS instead of a relay and track the AC waveform.

This condition is the far worst of the 2. If you want to pursue it, I can give you some pointers.

[IanB]

A secondary thought. If you pass 35 A through a diode with a drop that could be as high as 1 V at high currents, then it will be dissipating about 30 W, reducing to 15 W when considering it only conducts each half cycle. For four diodes in a bridge that will be 60 W total.

So I dare say you have already considered this, but some good heat sinking will be essential on that big rectifier...

[Jon86]

A secondary thought. If you pass 35 A through a diode with a drop that could be as high as 1 V at high currents, then it will be dissipating about 30 W, reducing to 15 W when considering it only conducts each half cycle. For four diodes in a bridge that will be 60 W total.

So I dare say you have already considered this, but some good heat sinking will be essential on that big rectifier...

Well I hope he doesn't pull 35A through it, otherwise he's going to blow the arse out of that 1KVA transformer! and the fuse!

[BravoV]

Too bad the working voltage is too high, otherwise the Linear Technology LT4320 is almost a perfect solution for you. 

[AlanR]

I have this belief that if some engineer has done all the hard work that I do not have to do, then I don't try to reinvent the wheel so to speak. Just integrate their ideas into your designs!! Any engineer who says they did not get some ideas from other people's designs are liars or must like reinventing the wheel as a hobby.


I will add that you will need to have a strong background in circuit analysis. This will allow you to adjust components to meet your specific needs of your design if you have a fundamental understanding of the circuit that you are analyzing.


http://www.ti.com/lit/an/snaa057b/snaa057b.pdf


Analyze their example circuit. In particular, look at how they solve the current inrush issue. 

[megajocke]

I'm not too worried about wasting a little power but I'd rather not dump loads of heat into my power supply case if I can avoid it. Using that 10 ohm part as an example though it looks like it has the potential to dump a lot of heat. Assuming it self heats to 85degC then the resistance drops to 1.5 ohms giving a 6.45V drop in the supply. At the maximum current  for the coil of 4.3A that equates to 27W of heat! In reality the coil will never draw 4.3A as I've over spec'ed it but I'd like to make sure it's safe if it does.

I'm open to suggestions for other solutions too :-), cheers.

If I read the table in the datasheet correctly you are off by a factor of 10 on the resistance there. You can also use the formula given in the datasheet to calculate the expected resistance at a certain current (using the k and n parameters) and it's around 0.1 - 0.2 ohms. That part should be OK.

[megajocke]

A secondary thought. If you pass 35 A through a diode with a drop that could be as high as 1 V at high currents, then it will be dissipating about 30 W, reducing to 15 W when considering it only conducts each half cycle. For four diodes in a bridge that will be 60 W total.

So I dare say you have already considered this, but some good heat sinking will be essential on that big rectifier...

The transformer (48 V 1000 VA) can only stand a DC output current of 12 A before overheating assuming a current form factor of 1.8. But that's still 25 W to get rid of, so good point.

[Lurch]

I'd be using a thermal overload for the transformer protection. The fuse would be for backup/cable protection. No idea where you so can't be more specific about parts or devices. Over here 13A fuse in plug and thermal overload in the supply line for the transformer would suffice.

[wobblycogs]

Thanks for all the help, I've got a lot of reading to do now but I'm sure I'll be back with some questions 

As for parts over heating I think I'll be ok as the system won't be drawing anywhere near it's maximum current. Even with all four steppers at full load it only comes in at 11A and it's highly unlikely that would happen for more than a few seconds at a time. The bridge is bolted to a large sheet of aluminium and there will be a fan blowing air over the coil and bridge. Anecdotally this seems to be fine looking at other similar builds.

While I'm not averse to building a little circuit to control the inrush I'm surprised I've not found a simple pre-built module to do it. Everything in my control case is DIN mounted and I'd sort of expected to find a DIN mountable inrush control. Do such things exist?

[calexanian]

Been down this road. switching on the primary with a healthy switch and you really will have no problem if you are using good caps. Otherwise the best thing to do is feed the caps with a small series resistor with a NO relay contact across it controlled by a short time delay. Make sure there is no load on the system during the charge up period. The whole thing can be a relay, c cap, a couple resistors, and you are in business.

[tszaboo]

I can't see how that would be any better as it lets go under 50A surges

Fuses have more characteristics than only current rating. You need to look at the I²T rating also, and understand how a fuse works. The current heats up the device, while there is a constant cooling of the ambient. If you have 50A surge for 20ms than that is 50 A²s. Your fast fuse is long gone because of this, while your slow easily survives. Slow fuses have more mass, so it takes londer to heat it up.
After this you only need other 5-6 parameter to size a fuse.

[Rufus]

Been down this road. switching on the primary with a healthy switch and you really will have no problem if you are using good caps. Otherwise the best thing to do is feed the caps with a small series resistor with a NO relay contact across it controlled by a short time delay. Make sure there is no load on the system during the charge up period. The whole thing can be a relay, c cap, a couple resistors, and you are in business.

Charging reservoir capacitors adds to the problem but it is mostly transformer and can't be fixed with anything on the secondary side.

The OP didn't say but I wouldn't be surprised if he has a toroidal transformer and they are especially wicked for the problem.

There is a reasonably good discussion and solution concerning big audio amp supplies here http://sound.westhost.com/project39.htm

[kxenos]

A simple technique that is used in many designs is that you have a resistor to limit the inrush current and a N.O. relay that bypasses the resistor after a delay (e.g. 0.5sec) has passed.

[wobblycogs]

Yes, it's a toroidal transformer, this one to be precise http://www.airlinktransformers.com/chassis_mounting_toroidal_transformers/chassis_mounting_toroidal_transformers_standard_range/CM1000224/.

Funnily enough I already have that west house project39 page open to read. I got to it from this http://sound.westhost.com/articles/inrush.htm page which also discusses the problem.

The control cabinet I'm building includes a 24V supply for the various safety systems and I have a spare 24V N.O. relay good for 10A @ 230V so I could do something with that. I wanted to avoid putting 24V into the main power supply though as it might one day end up in it's own box.

[SArepairman]

You can consider using an active rectifier, as your voltage is low enough I believe:


http://www.electronicsweekly.com/news/design/power/in-depth-linear-technologys-active-bridge-rectifier-2013-06/

this way you can give your project a modern spin.


or you can use a schottky diode., it will have less losses, and less heat dissipation, if you give a shit.

I would use a 555 timer and a relay/power resistor for inrush current limiting, I think its way more professional then those stupid NTC things that will probably break.

[SeanB]

You already have a 24VAC supply from the transformer, use it. Extra 1A bridge rectifier that only does the limiting relay will be easy, and all the parts aside from the bridge and resistors are DIN mountable. You can just use a DIN mount box to hold those if you want it to look nice.

[wobblycogs]

Thanks for all the ideas for ways to solve this problem. I've had a think about it and I'm hoping that this solution will work and meet with at least some approval :-)

I noticed that it's possible to buy relays with timers built in so I figured rather than build a timer circuit to control a relay I'd just use a pre-built one, as an added bonus the one I've selected runs off 230V AC so there's no need for an additional rectifier and smoothing capacitor to generate 24V DC. As a safety measure I've put in a thermal fuse. To be honest if the relay failed to close I'm not actually sure the fuse would blow as I'm not loading the power supply very heavily.

I'd really appreciate it if someone could take a quick look at the components I've chosen and tell me if this will work. I realize this isn't the cheapest or most elegant solution but it fits in the space I have and I feel comfortable building it.

• Time Delay Relay - http://uk.rs-online.com/web/p/time-delay-relays/7740304/ - Supply 24VDC or 24 to 240V AC, Output 8A max
• Thermistor - http://uk.rs-online.com/web/p/products/211-8002/ - 10Ohm, 5.1W
• Thermal Fuse - http://www.ebay.co.uk/itm/10-Pcs-Circuit-Cutoff-TF-Aluminum-Thermal-Fuse-110-Degree-Celsius-250V-10A-/121184286421 - 100 deg C, 10A

I'd guess at 1.5 seconds for and initial inrush protection time and then tweak it from there if the breaker trips.

[SeanB]

You could drop inrush time down to 0.5 seconds with no problem. Set it to mode A and set the time to 0.5 seconds and it will still work, that is a 20 cycle period of AC mains to charge the capacitors, which will charge in the first 5 cycles to close to full voltage.

[NiHaoMike]

You could try a circuit breaker designed for compressor motors.

[Richard Head]

Wobblycogs

The cause of this problem is probably the fact that the transformer you are using is toroidal and not so much the discharged electrolytic capacitors.
Because toroidal transformers are tape wound there is no inherent airgap in their structure. This results in the remenant flux in the core remaining firmly in one quadrant of the B-H curve when switched off.
Now if you switch off the power to the transformer when the core flux is near a peak in one direction it remains there (This is how ferrite core memories work).
If, when you switch the mains back on, the core is magnetically excited in the same direction as it was when switched off then there is a very high likelyhood that the core will saturate as there is insufficient flux headroom for twice the volt-second product across the primary.
The same effect occurs with Variacs when powered up. Since a Variac uses a toroidal core it too is subject to this phenomenon and often trips the upstream breaker.
The solution can be to have the transformer re-designed with more primary turns to give more flux headroom, implement in-rush current limiting or use an E-I transformer core configuration.
I hope this helps you a little.

Rich 

[Kremmen]

In case it is relevant to the OP here is a pdf attachment of a limiter circuit that i have used instead of an NTC resistor. NTCs have the downside that when hot they do not do what they are supposed to. So repeated restarts can cause high inrush when the NTC is still hot.

The circuit is meant to control large contactors but is easily adaptable to smaller targets. It is not the simplest possible circuit because there is the safety feature that it is separately supplied and should the supply fail it will fail the entire function safe.
If there is anything interesting, feel free to use as you please.

[wobblycogs]

Thanks to everyone's help I've got the inrush limiter built and tested - it works perfectly . I thought some people might like to see the final result so here it is in all it's spaghetti wiring glory. The lump up the top left is a VFD and the blue things top right are steppers drives if anyone was wondering. The transformer coil is hidden under the  frame that the capacitors and inrush circuit are mounted on and there's a cover that goes over the top to prevent (any more) accidents with the capacitors.

Actually, thinking about it, I have another quick question. Should the dc common on the secondary side of the transformer be wired to earth as shown in the circuit diagram attached? I've lost count of the number of transformer circuits I've looked at, some wire to earth and some don't. Clearly it works without this connection but as I understand it the reason for wiring it to earth is to stop the secondary floating up to a high potential relative to earth which seems like a good thing. I just don't want to accidentally let the blue smoke out now that I've  got this far!

Next up is finding some decent connectors for drive --> steppers and spindle --> VFD at a price that won't cause the bank manager to pass out.

[electronics technician]

enjoy cheers
schematic is in the video