Soft starting larger transformers (<1 kVA) and power supplies

[nctnico]

Now we just need to source a +200 W NTC 
Do you really think I didn't consider that? Or did you miss the little "k" before "VA"?

Using an NTC is the standard solution.

Farnell even has a dedicated section for inrush limiting NTCs:
https://nl.farnell.com/c/circuit-protection/thermistors/inrush-current-limiting-icl-ntc-thermistors

Following free_electron's suggestion to use a 40 Ohm one:
https://nl.farnell.com/ametherm/sl22-40005/ntc-thermistor/dp/1703898

[free_electron]

Now we just need to source a +200 W NTC 
Do you really think I didn't consider that? Or did you miss the little "k" before "VA"?

why ? this has been done for ages this way. even the computer power supplies do this !. you don't need a kilowatt ntc. that's the whole point.
It is only to dampen the inrush current. Even if that transformer behaves as a dead short a 40 ohm ntc at 220 volt is a maximum of 5.5 ampere. easy peasy. the NTC's easily handle 30 ampere.

https://www.digikey.com/en/product-highlight/e/epcos/p27-series-ntc-inrush-current-limiters?utm_adgroup=xGeneral&utm_source=google&utm_medium=cpc&utm_campaign=Dynamic%20Search_EN_Product&utm_term=&utm_content=xGeneral&gclid=CjwKCAjw3cSSBhBGEiwAVII0Z0WGZvb49hPQjjfHICyj6N9YaShSxeOYK7_AET0BLNhrmd6_rRNIIhoCZIQQAvD_BwE

So your fuse will not pop, the ntc will heat up , lower its resistance and provide a smooth start.
The relay kicks in a after a few seconds bridging the NTC and switching on the load. the NTC is out of the loop.

So next time before blowing your fuse think about what i wrote. Yes i noted the small K. You don't need kilowatt NTC's. you have 5 ampere tops.

[Benta]

@OP I would add a bleeder resistor across the timing capacitor for repeatability of delay between short power off-on intervals.

Yes, @jbb had the same comment, so I went back and looked at this point.
I analyzed again and have to agree it's an issue. Unfortunately it's not solved with a simple bleed resistor: an active "C2-reset" is necessary.

I attach the "Rev1" version of the circuit which implements this. Additionally the simulations.
Sim1.png:
Red trace: C1 voltage.
Blue trace: C2 voltage.
Green trace: BRT22 LED current.

Sim2.png:
Purple trace: applied mains voltage. The interruption in the middle is for provoking problems none happen.

The Triac/Diac circuit can be replaced with a resistor or self-heating power NTC or whatever according to your needs. I'm agnostic there. But the delay circuit works well and is built from jellybean parts, except for BRT22 and relay.

[nctnico]

@OP I would add a bleeder resistor across the timing capacitor for repeatability of delay between short power off-on intervals.

Yes, @jbb had the same comment, so I went back and looked at this point.
I analyzed again and have to agree it's an issue. Unfortunately it's not solved with a simple bleed resistor: an active "C2-reset" is necessary.

Which I why I typically use a dedicated chip (voltage supervisor) for these kind of designs instead of a discrete solution. Less parts and better behaviour.

[Benta]

Which I why I typically use a dedicated chip (voltage supervisor) for these kind of designs instead of a discrete solution. Less parts and better behaviour.

Like an MC33164? Nice idea, I like it.
Why don't you publish anything, being so smart?

[floobydust]

I measured primary inductance about 60mH on a 1,000VA EI core 115VAC primary I have lying around.

[Benta]

I measured primary inductance about 60mH on a 1,000VA EI core 115VAC primary I have lying around.

I find that implausible. It would imply a magnetizing current of around 5 A.

[nctnico]

Which I why I typically use a dedicated chip (voltage supervisor) for these kind of designs instead of a discrete solution. Less parts and better behaviour.

Like an MC33164? Nice idea, I like it.
Why don't you publish anything, being so smart?

As I wrote: I'd use an NTC for a transformer application. Nice and simple. I suspect anyone can wire up a component with 2 leads.

However, if you really want a circuit with a predictable startup sequency, then a voltage supervisor chip is a good option. Something with an adjustable delay can be handy if the secondary circuit needs time to discharge itself. The LM8365 for example.

[Benta]

@nctnico:
Since you're the super-expert here on this subject, why don't you just present your own super-circuit?
Just commenting and critizing others in a 360-degree-circle isn't very helpful.
I'm certain that with all your experience, you can post functioning circuits from your library within minutes. That would be helpful to everyone.

[nctnico]

Feel free to PM me for a quotation for my design services. Other than that I have to refer you to the datasheet of the chip I mentioned so you have a good starting point. IOW: I can tell you which ingredients to use to bake a good bread but I'm not going to hand out bread for free.

[free_electron]

t I'm not going to hand out bread for free.

he want you to put it in his mouth , move his jaw and clean up the day after as well...

[NiHaoMike]

If the efficiency loss of the NTC is not acceptable, an easy fix is to bypass it with a relay connected to the main capacitors. It will close once there's a decent amount of time to allow them to precharge.