Surge current generator to start up SMPS

Question

I want to power supply a board (variable load 1.6 W at 24 V) from a 150 VDC source (rectified AC line) through a 2 kΩ, 20 W series resistor.

Due to mechanical arrangement of the equipment the only available access is the second terminal of the 2 kΩ resistor and GND. No access to the main supply is possible. All needed power must be drawn from the supply through the 2 kΩ resistor (max 75 mA). The board can be simplified as three bucks (24 V to +10 V/+5 V/-10 V) each supply many devices.

Since the current is limited by the 2 kΩ source resistor, the board won't start operating because it needs a surge current at start-up (due to the large input capacitance). One could say a soft-start should be considered after reaching the desired voltage bucks can be enabled to operate. Good point, but the board could not be redesigned.

So, I came up with external circuitry to store sufficient energy in a large capacitor. After reaching the target voltage the energy should be released fast enough to the load. I tested the concept by connecting the board manually when the voltage across C1 is regulated through Q1D1 LDO. It works!

Where I got stuck is with making a fast switch to release the stored energy. I have tested the following circuit, but the board won't start; I think because it turns on slowly.

The second solution I considered is to use an SCR and a DIAC to connect the load after reaching the voltage threshold. The problem with this configuration is that if the source drops enough to power off the board, but is still delivering a small amount of current, the SCR stays ON and after the source reaches the nominal voltage again the boards won't restart operating (voltage divider between source resistor and load).

Maybe the fast solution is to use a comparator with a push-pull output and fast slew rate to drive the PMOS (1st solution) fast enough. It will be also easier also to define thresholds voltages by monitoring the source voltage to open and close the P-MOS switch.

I am looking for a simple solution with hand wired elements (1206, 0805, SOT23, SOT228, SOD80...). Any idea to suggest, even if it is for a complete solution, not only the switch?

Answer 1

All components in the circuit below have 2 or 3 terminals. Hopefully that's discrete enough for your taste :)

The 1st part of the circuit is the voltage limiter. The circuit you proposed has a fatal flaw: it will not work properly when the input voltage is exactly 24V - and that's the steady state of the overall circuit. When 150V via 2k is loaded with a 1.6W load, the voltage drops to 24V. Instead of an N-type pass element, we need a P-type pass element:

^{simulate this circuit – Schematic created using CircuitLab}

This is the 2nd part of the circuit: the load switch with hysteresis. The load is turned on when the C1 voltage raises above 25V. The load is turned off when that voltage goes below 23V.

CLIM must be about 25x larger than the load capacitance CL.

^{simulate this circuit}

Answer 2

Without defining your entire circuit energy, power, voltage and thermal expectations you may learn something from your attempts, but this topology is unwise due to resonant condition with LC stored energy transfer without a controlled damping R.

I think that you would be far better off using a 2W offline AC/DC regulator topology such as the MP150.

https://www.monolithicpower.com/en/mp150.html

Edit1:

Ok back to your 1st option , replace the FET resistor with a comparator at sufficient threshold and add hysteresis. The input will sag due to input loading and you do not want it to oscillate on startup. Consider a 90% threshold and 20% hysteresis for a starting point.

You will have to monitor start using any switch at different Vin thresholds somehow to characterize performance to optimize these the parameters to ensure it will work. The Comparator and FET switch ought to be fast enough but will a few (?) milliseconds of storage time be enough to start it. Some time constants ought to be defined for each condition.