« Diode Recovery Losses | Main | How do I design ...? »

August 17, 2001

UPS Restart

Question: Can you help with an interaction between an UPS system and switching-mode power supplies?

Original Question: We service a fire alarm panel that uses a switching power supply. Every time the building experiences a power outage (which is at least twice a month) the power supply fails to restore commercial power to the panel without going through a "warm" start procedure which consists of disconnecting the backup batteries then disconnecting 110 V, waiting 30 seconds or so, then reconnecting 110 V followed by reconnecting the backup batteries. Instead the panel continues to run off the backup batteries until they either go dead or until we arrive and "warm" start the system. Since the batteries and charging circuit are obviously good since they "hold" the system up in the absence of commercial power why doesn't the system switch back to 110 V when it's restored? In addition we noticed the LED indicating the power supply is "on" was oscillating rapidly while the problem was present. After "warm" starting the panel it resumed normal operation.

Answer: This is what I think is happening. Input power is sensed at the bulk energy capacitor on the input of the switching-mode power supply. It is sensed here to minimize short-time transients that the power supply can operate through. When this voltage drops below a certain threshold for a period of time (such as the decay when ac power is lost) the battery is switched in either with an SCR, or some latching switch that acts like an SCR, which can only be reset by removing power, both ac and dc. The 30 seconds is needed to make sure all capacitors holding energy in the circuit are discharged. The oscillating LED is probably meant to indicate ac power was lost and the system in running on battery power. The circuit is probably working as designed. It could have been designed that way for two reasons. One reason is that turning off an SCR that is carrying current complicates the circuitry and it is cheaper to design it as it is now working. Another reason is that in some applications you want to know that an event took place and want to require a manual reset so the system can be checked for proper operation. Hence you might design it that way with no regard for cost. If you want automatic reset, you might open a dialog with the manufacturer. It may be as simple as changing a jumper to get the operation you want. I suspect they have a solution for automatic reset, but what is involved in swapping or modifying equipment and the costs involved is the open question.

Posted by Jerrold Foutz at August 17, 2001 10:03 AM