I love the smps bricks that can be found at flea markets, dumpsters, etc.

These old laptop/printer/game console supplies are small, light, and powerful.

Yet, if you ask them to deliver their power supply rating all at once, say, a bank
of halogens, or a foam cutter wire, the low initial resistance causes them to go into fault mode,
and shut down.

Is there a way around this?

 

Maybe a slow-start circuit for the loads you mentioned?

 

low initial resistance causes them to go into fault mode,
and shut down. Is there a way around this?

Several ways. An NTC thermistor comes to mind. The hotter it gets, the lower its ohms.

 

Maybe a slow-start circuit for the loads you mentioned?

I would go at that with a resistor to a capacitor and them controlling the gate of a mosfet. Three parts and you have a slow start circuit worth several amps.

 

If you're really good at electronics, you might read the part number off the controller chip, trace out the circuit board, and add some capacitance to delay the shut-down detector.

Emphasis on, "might".

 

Yes, a "slow start circuit" sounds like a one size fits all. I have several different mfg. models.

I guess I can search "slow start circuit" and get plenty of ideas. Many of the bricks that I have
are 6.5, 14, 18 amps. I just ordered a bunch of 25A TIP35c NPN. Maybe I can use those.

 

The idea of defeating the internal safety circuitry of the brick just doesn't sound good. You don't have access to the original design issues that led to the circuitry. Better to fix the problem with the load and let the brick do its thing.

 

I just ordered a bunch of 25A TIP35c NPN.

You would be much better off with a mosfet. Those puppies can virtually disappear (in the voltage department) if you get one the right size.

 

Here's a nice one for $1.39, but I must confess, I didn't try very hard searching for the best transistor or the best price, just one that would do the job and be nearly idiot proof.

 

IRLB8314 MOSFET, Wow, 170A! That might do nicely...

I found this, specifically for lamps on smps:

​

http://www.discovercircuits.com/DJ-Circuits/lampcurrentlimit.htm

And I also found a paper from Motorola on the subject of mosfet/inrush, but it was a little over my head,
and no component values to get me started:



http://www.bonavolta.ch/hobby/files/MotorolaAN1542.pdf

Which scheme should I focus on developing? I've already created a 10A
LM723 power supply, with a 250va transformer, & it works great. Now I want to
to do something with these SMPS that I have been collecting. Initially, I was going
to use one of my SMPS for low voltage halogen lights above my workbench, and save my expensive
transformers for other projects, but I ran into a snag. If I can make a versatil soft start circuit,
maybe I can use it also for building a hybrid SMPS/Linear adj. power supply... That
would be cool!

 

Your first circuit is a constant current sink. If the load doesn't start you might have to set the current to 2 or 3 amps. After the load gets warmed up, it will limit itself to 1 amp.

I simplified the second circuit for you. It should work for any power supply equal to or greater than 4.5 volts.

 

You can override the current sense pin on the smps chip, but then its open to disaster when it tries to exceed its own rating, and then blows the mosfet and takes out the pulse transformer.....

 

...................
I simplified the second circuit for you. It should work for any power supply equal to or greater than 4.5 volts.

The problem with the simplified circuit is that the gate voltage delta between OFF and ON is small and thus the turn on will be relatively fast even for a large RC time constant.
It needs the added Miller capacitance from gate to drain to provide a decent start-up ramp.
Below is a simulation with an added Miller capacitance of 0.1pF (essentially no capacitor) and 75nF to show the difference.

The actual capacitance values needed depend upon how slow the ramp needs to be, the supply voltage, and the characteristics of the particular MOSFET used.

Edit: R2 is the load, if that's not apparent, and is just an arbitrary value for the simulation.

 

That's a great plan, @crutschow ! So...why is your brain still working at 5 AM when mine went mushy before 4 AM?
Decades of healthy living? Dancing with the California Raisins? You had already slept and it was morning for you? What registers as 5 AM in Florida is 2 AM in California so it was only a little past your normal bed time?

You don't have to answer that. It's an excellent circuit and completely corrects what I did after a 21 hour day of mostly physical work. I was so focused on douching the dog smell out of a used car that I was merely killing time at AAC during rest periods. Not my best work.

 

That's a great plan, @crutschow ! So...why is your brain still working at 5 AM when mine went mushy before 4 AM?
Decades of healthy living? Dancing with the California Raisins? You had already slept and it was morning for you? What registers as 5 AM in Florida is 2 AM in California so it was only a little past your normal bed time?
.......................

I'm rather a night owl and seldom get to bed before 2 AM.
But it looks like you went to bed at a much later (local) time than myself.

 

Is {c} or C2 a component, or the miller cap. of the mosfet?

I'm going to try this on my workbench right now.

The problem with the simplified circuit is that the gate voltage delta between OFF and ON is small and thus the turn on will be relatively fast even for a large RC time constant.
It needs the Miller capacitance from gate to drain to provide a decent start-up ramp.
Below is a simulation with a Miller capacitance of 0.1pF (essentially no capacitor) and 75nF to show the difference.

The actual capacitances needed depend upon how slow the ramp needs to be, the supply voltage, and the characteristics of the particular MOSFET used.

 

Is {c} or C2 a component, or the miller cap. of the mosfet?
..................

It's an added discrete capacitor.
The MOSFET Miller capacitance is already included in the MOSFET model.

 

I built #12's circuit first and only,
because when I sat down to do it,
I realized I did not have values for
C2 in crutschow's circuit (sorry!)

Anyway, I tried a 50N06 & a P24NF10 mosfet.
I used a 1M trimmer for the resistor.

And... It works! I have an Microsoft 360 SMPS of 14A, and it
worked with a 12A halogen load, and I have a Dell
SMPS rated at 18A, and it also worked with the same
load (three 50w halogens, 12v, in parallel).
Either mosfet worked fine, but the P24NF10 had
better pins, and fit better in my breadboard.
Both ran slightly warm, without heatsink. The one
meg pot made no difference when I rotated it between
10% & 90%.

Thank you both, this is so cool!

What should we try next?

 

Sorry, crutschow, I now see the 75nF value was in your notes. SORRY!

 

Sorry, crutschow, I now see the 75nF value was in your notes. SORRY!

No problem.
If you get the turn-on limiting you need without that miller capacitor, then you obviously don't need it.