Line conditioner feeding Switch mode PSU

[Skogsgurra]

I had an interesting case this week: A gas analyzer was damaged by mains transients and a line conditioner (magnet stabilizer) had been installed to protect it.

After the installation, the analyser failed very often - and it happened when it was switched off or on. I was asked to find out what the problem was and hooked up a recorder to observe voltages and currents. It was quite clear what happened: switching off the primary voltage to the stabilizer caused the output to decay from nominal voltage to zero volts over about ten cycles. This is a consequence of the secondary parallel resonance circuit that is part of the stabilizer. So far all seemed to be OK.

The gas analyzer has a switch mode power supply. It delivers nominal voltages to the internal electronics even if the mains voltage varies over a wide range. Normally one of these supplies can work with an input voltage between 100 and 260 V. The internal electronics consumes constant power regardless of input voltage which means that the input power is constant.

Constant input power translates into increasing line current when mains voltage drops. Half voltage is equal to double current, which is easy to understand.

But, when the line conditioner output came down to about 20 V (from nominal 230 V) the input current increased to about ten times nominal current for about ten milliseconds. Such high a current will heat the switches (Mosfet, resistive channel) proportional to current squared and we were lucky that the analyser survived this time. It usually doesn't.

I told the customer to use a line filter and varistors to suppress the line transients instead of using the stabilizer and also not to switch off the stabilizer primary before they have chenged the installation. I hope that it will solve the problem.

My question: Is this something that everybody knows about? Are Magnetic stabilizers known to kill switch mode power supplies? Do modern PSUs have a voltage monitor that switches off when input voltage gets too low? This PSU obviously does not, it is 10+ years old.

Comments?

 

[logbook]

An interesting application.

Any decent off-line switched-mode power supply would have a "brown-out" detector and would just switch off gracefully when the mains dropped outside of the specified working voltage range. Alternatively a current trip should operate inside the supply. It is not reasonable to have to protect the power supply again this drooping input voltage. On the other hand this sort of brown-out condition is not normal, so the power supply manufacturer probably wouldn’t have tested for it.

… but then again, the new EMC dips and transients standards encourage the supply manufacturer to keep going during massive dips. Thus the brown-out detectors would need to be disabled for optimum performance against this criterion. You then get into a time constant issue of how long a brown-out is tolerable before you trip the supply

 

[IRstuff]

While it's not clear that everyone knows about it, the designer of the PSU should be designing against such conditions, or whoever did the design specification was remiss in thinking that such conditions never exist.

TTFN

 

[busbar]

 
Anecdotally, many switchmode power supplies seem to be designed to work best against stiff (low-impedance) sources, and ferroresonant transformers like to see ‘passive’ or linear load. [There is a classical similarity to interaction between earlier gensets and UPSs.] This makes for quite a zoo of aftermarket fixes, sometimes with no guarantees while further compounding problems.

This guy’s work in is a step in the right direction

http://www.powerstandards.com