How to limit output current on switch mode power supply?

[carl_lab]

I have some small notebook-power-supply-like SMPS I want to use to charge 48V LiIon batteries.
These batteries have internal  battery managment systems for safety.
I need a voltage of 13*4.2V=54.6V and limited current of about 1A (constant current mode).
I know how to change output voltage of SMTP (by changing a voltage divider in feedback loop) to fit the 54.6V I need.

But:
When I connect the power supply to battery, current is about 2A for about a second, then power supply switches off (because 2A is over spec.).
I need to add some circuit, that limit output current to about 1A.

Can someone show me a circuit diagram of a current limiting (constant current) SMPS?

I think LED light power supplies may be suitable, but I don't want to buy new power supplies, I just want to use these ones I already have.

The additional circuit maybe a shunt resistor, OP amp, reference Z diode,... something similar this way, that controlls the opto coupler in feedback loop of SMPS.

Could this work?

[Zucca]

Can you tell us which IC is using this SPS? I would look into the datasheet of that IC a as a starting point, normaly the key note or the circuit examples are very useful. If you are lucky there is an example on how to limit the current...

[bktemp]

Most notebook power supplies output around 20V. Going to >50V does not work very well, because of either the fixed turns ratio of the transformer (for forward converters) or the switching transistor will also see a much higher voltage and fail (for flyback converters).
You circuit could work, but may require a compensation network for loop stability. Instead of connecting R3 to Q1 you could also connect it directly to TL431.

[carl_lab]

Can you tell us which IC is using this SPS? I would look into the datasheet of that IC a as a starting point, normaly the key note or the circuit examples are very useful. If you are lucky there is an example on how to limit the current...

I did not open the power supplies yet.
But yes, you're right, datasheets are helpful mostly.
Problem is: I need a constant current limiter but the power supply shuts down if current exceeds limit (foldback mode):

Most notebook power supplies output around 20V. Going to >50V does not work very well, because of either the fixed turns ratio of the transformer (for forward converters) or the switching transistor will also see a much higher voltage and fail (for flyback converters).

It's actually 48V power supplies. So additional 4-5V should be not the problem.
"notebook-like" refers to formfactor/case.

Instead of connecting R3 to Q1 you could also connect it directly to TL431.

No. I don't think so.
Q1 works as an inverter.
If current exceeds limit, opto's LED will shine to stop oscillation of the primary oscillator (same as in over voltage case).

[bktemp]

Instead of connecting R3 to Q1 you could also connect it directly to TL431.

No. I don't think so.
Q1 works as an inverter.
If current exceeds limit, opto's LED will shine to stop oscillation of the primary oscillator (same as in over voltage case).

TL431 also works as an inverter:
When the voltage at its input goes over 2.5V it turns on. Q2 simply simulates overvoltage at the feedback point when the current is too high.
Since there is already a compensation network around TL431, it may be easier than adding a completely seperate control loop.

[T3sl4co1l]

This may be of interest?



It's made for CC operation primarily, with CV as a protection feature (hence, poor regulation in CV).  But the lesson should be easily applied either way.

You definitely don't want two transistors cascading, collector to base, I mean, unless "astable multivibrator" is what you're going for there!

Tim

[carl_lab]

TL431 also works as an inverter:
When the voltage at its input goes over 2.5V it turns on. Q2 simply simulates overvoltage at the feedback point when the current is too high.

You definitely don't want two transistors cascading, collector to base, I mean, unless "astable multivibrator" is what you're going for there!

OK, then this way should work:

[carl_lab]

This may be of interest?
It's made for CC operation primarily, with CV as a protection feature (hence, poor regulation in CV).  But the lesson should be easily applied either way.

I think for Li-Ion battery charging stability of max. voltage is much more important than stability of charging current.
Limiting the current is primary for overload protection of charger and battery.

[T3sl4co1l]

OK, then this way should work:

Not quite. Actually, that may be even worse, in a sense.  The TL431 acts like a transistor with a low fT, high hFE, and very stable Vbe (~2.50V).  The low fT means, not only will it still oscillate -- it will do so at a frequency where the power supply is hammering on and off, hard!

Managing gain and phase shift is the important part here.

Note that Vin can't be pulled any lower than 2.5V (if R3 = 0), or in general, whatever the (R3 ||  R5) + R6 voltage divider comes out as.  If you don't need the current limit to work effectively down to low output voltages, then no problem.  But then, if such a crude limit is acceptable -- would it be acceptable to implement this on the primary side instead?

Tim

[carl_lab]

I want to charge 13-cell-Li-Ion-battery-packs, so 13*3V = 39V is the lowest battery voltage if fully discharged.

[oldway]

Why not only change the output voltage of the SMPS (that's easy) and add an external linear current limiter (also very easy, 2 pnp's transistors with heat sink, a 0R6 resistor more two other resistors).
Power dissipation will be max. 15W, that's acceptable.

[carl_lab]

Why not only change the output voltage of the SMPS (that's easy) and add an external linear current limiter (also very easy, 2 pnp's transistors with heat sink, a 0R6 resistor more two other resistors).
Power dissipation will be max. 15W, that's acceptable.

15W of heat extra is 15W too much. The power supplies is fanless design.

Tomorrow I'll break the case open of one of the power supplies, we'll see...

[carl_lab]

Can you tell us which IC is using this SPS? I would look into the datasheet of that IC a as a starting point...

DAP008
http://fa.itservice-bg.net/baza/ic_power/sm_data/DAP008_data.pdf



I fear, there is no TL431, so voltage regulation will be not that good...
But there are two opto couplers in my device, I'll try to retrace the circuit diagram.

[Zucca]

Just skimmed the Datasheet.

The DAP008 features two efficient protective circuitries:
1. In presence of an overcurrent condition, the output
pulses are disabled and the device enters a safe
burst mode, trying to restart. Once the default has
gone, the device auto?recovers.

that's your problem.

The DAP008 implements a standard current mode
architecture where the switch?off time is dictated by the
peak current setpoint.

With an internal structure operating at a fixed 65 kHz, the
controller supplies itself from the high?voltage rail,
avoiding the need of an auxiliary winding. This feature
naturally eases the designer task in battery charger
applications. Finally, current?mode control provides an
excellent audio?susceptibility and inherent pulse?by?pulse
control.
When the current setpoint falls below a given value, e.g.
the output power demand diminishes, the IC automatically
enters the so?called skip cycle mode and provides excellent
efficiency at light loads.

Not bad for charging batteries. On the other side it looks to my poor eyes that the output voltage is set by using the trafo on the right side. The rest is current business.
Interesting.

Can you give us some board pics?

[xavier60]

TL431 also works as an inverter:
When the voltage at its input goes over 2.5V it turns on. Q2 simply simulates overvoltage at the feedback point when the current is too high.

You definitely don't want two transistors cascading, collector to base, I mean, unless "astable multivibrator" is what you're going for there!

It should work. The current regulation loop needs to be kept slow to avoid stability problems. Picking a high value for R3 will help to make the current loop response slow.

[xavier60]

placing a capacitor between B and E of Q2 will slow the current loop response more. R3 needs to be kept high to limit the voltage range the current loop has control over.

[T3sl4co1l]

Even better is an R+C from base to collector.  The series collector resistor R3 generates a proportional voltage drop, so that there is an "output" voltage from the transistor to draw feedback from (otherwise, the REF pin will have little voltage change on it, being centered at 2.50V for obvious reasons).  The series base resistor R2 allows feedback to be applied; otherwise the "input" (base) impedance would be shorted out by R1, making feedback impossible to apply.

Likewise, C1 should be an R+C.  In both cases, the R+C is a pole-zero network, that significantly increases the phase margin when dimensioned properly.

The C1 response will stack with the Q2 response, making phase margin worse for current regulation than for voltage regulation.  You'll need to test actual values in circuit, and play around until both operating modes behave well (i.e., step load transient testing).

Tim

[carl_lab]

... there are two opto couplers in my device, I'll try to retrace the circuit diagram.

Here is it:

[carl_lab]

Even better is an R+C from base to collector.

So, this is what you mean?



BTW:
What is the 2nd feedback loop for (ZD54, ZD55, IC3 -> ADJ pin)?
Is it just a redundant over voltage protection?

Which resistor has to be changed for voltage adjustment? R55+R58 (one of these)?

What is D51 for?
For temperature compensation of Q51?