High initial output current prevents smps to start-up and reach to stable condition

[seyyah]

I've designed a 24V 2.5A smps using powerint's top266eg chip using their reference designs. It works good normally. But if I connect a high capacity load that draws 5A at start-up but below 0.5A in normal condition, it enters short circuit protection and keeps restarting but cannot reach to stable condition. It has a soft start function that takes 17ms for this kind of situations but it seems not to be enough. I tried a commercial power supply with similar specs and with the same load and it works without problem. It draws a peak current of 3.8A or so at start-up. So I think I should slow down start-up process but I don't have an idea how to do it. What can I do for this problem? Any ideas? Thanks...

 

[chuckey]

How about a circuit?
Frank

 

[BradtheRad]

Here are two variations on a delayed startup.

Circuit #1:
While a capacitor charges, the transistor passes less current. It takes a second to ramp up to full current. The darlington arrangement permits the capacitor to be a very small value.

Circuit #2:
Using a relay. A series resistor provides partial power to the load for the first second. When the relay activates, the load receives full power.

**broken link removed**

 

[dick_freebird]

A foldback limiter can be a problem for exactly this reason.
A straight constant current limiter is preferable in my book,
but this may not protect the product as well under a long
term output short circuit scenario. And you've already gone
and picked your pony.

You might like a designed load switch between the DC-DC
and the troublesome load; there are integrated ones as
well as the discrete concepts shown. Have to wonder
why the driven load has not, itself, an inrush current limit
or a local soft-start function - why the high amperage at
low delivered voltage?

Is there an option to change the foldback current level
or the timeout, on the part you have picked?

 

[seyyah]

Circuit is attached

 

[chuckey]

Try making C5 bigger. The opto coupler conducts more if the voltage or current increase.
Frank

 

[seyyah]

The load may vary. So some may include inrush current limit some may not. I'm using CV mode. There's also CV/CC mode(which I tried but not inspected in detail and did not help) it may help but I need to investigate it further and deeply. There's an option to limit the short circuit current. It's at max now it can be lowered but I don't think it will help. Also restart time can be altered, again I don't think it can help, but I'll try that as chuckey suggested... On the other hand do you believe modifying control loop helps this situation?

 

[schmitt trigger]

If everything else fails, you may want to consider something as simple as adding a PTC thermistor in series with the output.

 

[seyyah]

Wouldn't an ntc be better?

 

[schmitt trigger]

Wouldn't an ntc be better?

YES!!
I made a typo error. The NTC thermistor is the one you need.
Thanks for pointing out.

EPCOS has a very extensive line of NTC inrush limiter thermistors.
https://www.epcos.com/epcos-en/5300...urrent-protection/ntc-inrush-current-limiters

 

[mtwieg]

Is the startup current a function of the rate of rise of the supply voltage (i.e. a large cap bank) or will it always draw more current regardless of the startup profile of the supply voltage? How long does the high current condition last? Depending on the answer, it may be feasible to handle the transient current with a large capacitor bank. Otherwise you will either have to raise the current capacity of the supply or reduce the startup load current by modifying the load itself.

 

[seyyah]

I tried ametek's ntc with 3.5A rating. It worked but it gets real hot @ 1.5A load current.

- - - Updated - - -

The waveform is attached. The blue one is current(measured on a 0.1 ohm ressistor and sees 5A or so top) and the yellow one is output voltage.

 

[FvM]

I tried ametek's ntc with 3.5A rating. It worked but it gets real hot @ 1.5A load current.

An inrush current limiting NTC must be hot in operation. That's how it works. You may want to use an electronic (transistor) current limiter instead.

You didn't yet tell if the load will be able to start with 2.5 A constant current (mtwiegs question in post #11). The test can be easily performed with a current limited lab power supply.

 

[BradtheRad]

From your scope trace it looks as though the startup surge lasts 45 mSec.

A choke is a simple way to stifle sudden current surges.
Your schematic has a small value choke (L2). By increasing it to 400 mH, that may be sufficient, according to this simulation.

The load would draw 4.8 A normally. However it is held below 2.3 A during the first 50 mSec of power-up.

 

[Orson Cart]

The TOP switch has a finite soft start time - as you say, if the output C is too large (and/or the load), the circuit will not reach a steady state inside the startup time and the chip will assume it has lost control and do its shutdown & auto restart thing. At very light loads it will eventually get there if more and more energy remains on the caps each time. The only solution is less o/p C or less load or a bigger TOPswitch (and matching transformer) that allows more power at startup.

 

[seyyah]

2.5A start-up current is not a problem, I've tested it for a long period of time and many times.

- - - Updated - - -

From your scope trace it looks as though the startup surge lasts 45 mSec.

A choke is a simple way to stifle sudden current surges.
Your schematic has a small value choke (L2). By increasing it to 400 mH, that may be sufficient, according to this simulation.

The load would draw 4.8 A normally. However it is held below 2.3 A during the first 50 mSec of power-up.

I think that value is not practically applicable...

 

[FvM]

2.5A start-up current is not a problem.

O.K., use a transistor current limiter circuit.