Constant current power supply

[Lango1818]

Hi everyone.
I purchased a hoflungdung  (WANPTEK) generic switch mode power supply. Spec 0 -30 volt and 0 – 10 amp. This is all very good and well except if you set the current limit to say 1 amp @ 15 volts it has a capacitor on the output and shorting this unit across a .22 ohm resistor yields a current spike of 55 amps before settling down to 1 amp. Very good at sending devices to silicon heaven, if you are trying to test your new circuit. I have used good quality power supplies before and if you set the current to 1 amp max that is exactly what the supply gives you no nasty spikes. I know it is cheap and nasty and for the price I am happy enough with it. I was wondering if anyone could point me to a circuit I could put in the front end of the supply to truly limit the current. :-//Thanks in advance for any help.

[Vovk_Z]

Large current spike is a usual thing for most power supplies (with CC mode).
If I want to power something sensible like LED I use resistor in series.
If you really need a bench power supply without current spike/overshoot you have to investigate which one really fulfills this requirement.

[Benta]

Large current spike is a .usual thing for most power supplies.

Only if they are designed by an idiot.

@Lango1818: I've no solution except: stop being a cheapskate. If you want a lab supply, get one and pay what it costs. Anything else is landfill waste.

[2N3055]

Large current spike is a .usual thing for most power supplies.

Only if they are designed by an idiot.

@Lango1818: I've no solution except: stop being a cheapskate. If you want a lab supply, get one and pay what it costs. Anything else is landfill waste.

Yep, R&S and Keysight are then idiots..

Most general purpose Lab PSU today have several hundred uF elkos on their outputs.

If you want a PSU with small output capacitance you'll have to make it yourself.

If you know about comercial one please let us know.

[PCB.Wiz]

Hi everyone.
I purchased a hoflungdung  (WANPTEK) generic switch mode power supply. Spec 0 -30 volt and 0 – 10 amp. This is all very good and well except if you set the current limit to say 1 amp @ 15 volts it has a capacitor on the output and shorting this unit across a .22 ohm resistor yields a current spike of 55 amps before settling down to 1 amp. Very good at sending devices to silicon heaven, if you are trying to test your new circuit. I have used good quality power supplies before and if you set the current to 1 amp max that is exactly what the supply gives you no nasty spikes. I know it is cheap and nasty and for the price I am happy enough with it.

That spike is purely from the output capacitor.
The variable is the size of that output capacitor, and all power supplies will have some inrush spike, if you insist on attaching charged leads to a load.

I was wondering if anyone could point me to a circuit I could put in the front end of the supply to truly limit the current. :-//Thanks in advance for any help.

It's non-trivial to make a wide dynamic range, lowest voltage overhead, current limiter to 0-10 Amps.

The simplest is a resistor 

Next simplest is a classic emitter follower + emitter sense resistor feedback transistor.
Easy to do, but has minimum current and adjustment is by varying a high current resistor.

More complex is to use a current sense amplifier, with feedback, but getting that stable is not easy.

HFE is one param that varies least with power transistors heating, so a practical manual-adjust current limit, would be a user-current injection into a 'good' transistor base.
The transistor needs to be very well cooled. The manual-(re)adjust takes care of initial HFE and and changes with temperature.

[macboy]

Large current spike is a .usual thing for most power supplies.

Only if they are designed by an idiot.

@Lango1818: I've no solution except: stop being a cheapskate. If you want a lab supply, get one and pay what it costs. Anything else is landfill waste.

All CV/CC lab supplies will have this issue, no matter how expensive. The problem is caused by the need for very low - ideally zero - output impedance in CV mode, which is the normal mode of operation. An output capacitor is always used. This works along with (and part of) the CV control loop to provide a very low effective output impedance.

For CC mode, we want ideally infinite output impedance. This implies zero output capacitance. Of course, neither infinite nor zero are realizable. Unfortunately, due to the presence of the output capacitor, the CC mode of typical lab supplies is very much impaired. The supply can only truly regulate the average current, not the peak, because the output capacitor is not within the CC control loop, and it can supply large peaks.

This problem is often worse with switch mode power supplies than with linear ones. The switchers will usually have more output capacitance as a consequence of their design. In a linear power supply, most of the bulk capacitance is prior to the regulator, and the output capacitor can be designed to be quite small, and often specifically has some high-ish ESR (up to a few Ohms), since very low ESR can disturb the regulation loop. On the switcher, the capacitance will often be 100s to 1000s of uF, and always low ESR. The best switch mode lab supplies follow the switcher with a linear regulator. This linear regulator can and will have a smaller output capacitor, but it will be be present.

Some true CC lab supplies exist, e.g. HP 6186C. These are generally a true CC source with very low (10s to 100s of pF) output capacitance, combined with a shunt regulator for voltage limiting. A supply designed this way is always sourcing the programmed current. If the load is too high resistance (or disconnected) then some or all of the current will be shunted by the CV voltage compliance limit portion. This makes them very inefficient, and they actually run hottest when no load is connected. But you can do things like set it to 5 mA @ 100 V, then connect the output across an unknown zener diode and measure the breakdown voltage of it. You can't do that with a typical lab supply, because the output capacitor charged to 100 V will dump into the zener, destroying it instantly/violently.

[MegaVolt]

Many sources have this problem

The solution is to connect the load to a switched off source and only then turn it on. The current limit will work as expected.

Connecting a load to an on-line source is bad practice in this application.

[Sacodepatatas]

I was wondering if anyone could point me to a circuit I could put in the front end of the supply to truly limit the current. :-//Thanks in advance for any help.

A proper choke in series.

[ArdWar]

The only test equipment I have that get anywhere close to being actual CC source is a SMU.

I don't think I ever came across a "PSU" that's free from surge discharge problem no matter how expensive (or cheap) it is. Some models do have low capacitance mode however, with couple tens of nF effective output cap.

[dietert1]

One could also try and use an inductor to reduce the spike, but there are risks with this approach as well. We have some large inductors e.g. a FANUC 70 mH 20 A choke. One would need a pretty strong zener to absorb stored energy during turn-off.
Another approach would be a programmable supply that can used with a PC to decently ramp up output.

Regards, Dieter

[MathWizard]

Large current spike is a usual thing for most power supplies (with CC mode).
If I want to power something sensible like LED I use resistor in series.
If you really need a bench power supply without current spike/overshoot you have to investigate which one really fulfills this requirement.

I seem to remember someone testing CC mode by seeing if an LED would survive some initial voltage without a resistor.

I have enough LED's, I should see what my Siglent PSU does to an LED. I always set the CC limit on my PSU, but I don't run things without resistors. But I wonder what the max currents spikes from this could be.

I have 75mA set as max limit right now at 5V. And I've noticed that hooking up to a circuit with ~100uF across the rails, I see sparks when I hook up the alligator clip. I don't remember 5V and such a low-ish capacitance doing that.