Pulse-by-Pulse Current Limiting, when implemented properly, provides a rugged and reliable power supply which will not fail. To achieve effective current limiting, you must have a fast current sensor which will respond in nanoseconds, going directly to your controller with minimum propagation delay time to the gate drive.

In order to prevent false tripping of the current sense circuit due to capacitive turn-on spikes, you will need to have an RC filter on the current sense waveform. There is then a fine line you must walk between preventing overcurrent failures and false tripping.

Be very careful if you are using a controller with Leading-Edge Pulse Blanking. You will usually need all the speed you can get to protect your power supply, and the 100 ns or so blanking time in the control may be long enough to destroy the switches.

Sometimes it will be necessary to deliberately slow down the switching of the
power FET in order to get completely effective current limiting.

 

Testing Your Current Limit Circuit

How good is your current limit implementation?
You should be able to conduct the following tests on the circuit:

• Put a hard Short Circuit on the power supply output.
  (Note: don't put the short ahead of the output rectifier diode, it is unreasonable to expect to protect your power switches in the event of a diode failure without some very special precautions. If the diode dies, the FETs will probably go too.)
• Put a wide-bandwidth current probe on the switch lead (or a voltage probe on a current sense resistor) . Gradually increase the input voltage until the power supply starts switching, monitoring the current in the switch on the oscilloscope. You should see very narrow pulses with their peak determined by the current limit threshold.
• Increase the input voltage to maximum line (plus a safety margin).
  The current pulses should get narrower and narrower, but the peaks will not run out of control, although they may increase a little from that at low line due to propagation delays. If the peaks start increasing considerably, your current sensing network is too slow.
• At high line, current pulses may be as narrow as 50 ns or less. You have to be able to control this situation reliably.

Repeat this test at the minimum and maximum temperature of power supply operation.

© copyright Ridley Engineering, Inc. 2007