Fault analysis on cpacitive power supply

[perky]

Hi All,

I'm designing a capacitive mains power supply. I need to do single component fault analysis and have a few questions:

The circuit is a standard capacitor with inrush limiting series resistor affair, and uses bridge rectifier. The capicitor has a parallel bleed capacitor. The requirement is that the circuit should not provide any dangerous voltages to the user should a single component fail short or open circuit so:

1) Does this mean that I should double up on the bleed resistor (i.e. put 2 in parallel)? AFAICS if a single bleed resistor were to fail O/C then the capacitor could remain charged. Putting 2 in parallel would seem to pass the single component fail test.

2) Do I have to consider what happens if the inrush limiting resistor were to fail S/C? This could lead to excessive inrush currents through the rest of the circuit (for a very short time), potentially causing cascade multi-component failure.

3) I have to consider what would happen if any of the bridge rectifying diodes were to fail O/C or S/C. I've noticed in Spice simulation that I could get full mains AC voltage swing with double the AC peak voltage value on internal circuitry. Should I use an MOV on this? The reason for this is the capacitor charges to peak voltage and is unable to discharge due to the diode failure, and this voltage is 'added' to the internal swing. Adding an MOV effectievly prevents the capacitor from being charged so high under that fault condition.

Has anyone out there done full single fault analysis on capacitive supplies, and what steps did you take?

Thanks!

Mark.

 

[duffy]

I've built a couple of these - the inrush limiting thermistor is to stop the fuse blowing when the caps charge. If it fails, it blows the fuse (or doesn't charge the caps) so not an issue.

If the bridge fails, the fuse blows - should not be an issue.

The single bleeder resistor is an interesting point, if one fails and you have a thousand microfarads sitting at +180V for hours that could be an issue.

There's a time-limit constraint, I usually pick a value that's pretty high for the bleeder in this kind of supply - 68k at 180V is almost half a watt of power, but it's also over a minute for one RC time constant (1000µf), and still over 50V (which UL considers dangerous) at the minute mark.

At some point you need to factor in the idea that the tech isn't a complete idiot, and will check those big dangerous caps before getting his fingers in there.

 

[perky]

Hi duffy,

I've built a couple of these - the inrush limiting thermistor is to stop the fuse blowing when the caps charge. If it fails, it blows the fuse (or doesn't charge the caps) so not an issue.
.

Well if the inrush limiter fails short there could be many amps flowing through the circuit on power up (depending on voltage phase). Since this charges up the capacitor very quickly indeed I suspect even a fast acting fuse won't get hot enough to blow. Question would be what is a reasonable max inrush current to choose. I'm thinking of putting an additional resistor in series with the first one - this would seem to pass the single component failure test.

If the bridge fails, the fuse blows - should not be an issue.

Actually if certain diodes in the bridge fail open circuit then this won't happen. Imagine if the positive half of the cycle was OK (both diodes conduct), but curent couldn't flow in the negative half due to one of the diodes blown open circuit - this would charge up the capacitor to the point where the voltages internally can become twice the peak voltage of the mains. This means creepage and clearance distances need to be doubled.

The single bleeder resistor is an interesting point, if one fails and you have a thousand microfarads sitting at +180V for hours that could be an issue.
There's a time-limit constraint, I usually pick a value that's pretty high for the bleeder in this kind of supply - 68k at 180V is almost half a watt of power, but it's also over a minute for one RC time constant (1000µf), and still over 50V (which UL considers dangerous) at the minute mark.
At some point you need to factor in the idea that the tech isn't a complete idiot, and will check those big dangerous caps before getting his fingers in there.

This circuit could be unplugged from the mains leaving the live pin on the plug exposed to an end user, not a techie. My cap though is only 680nF so time constant should be well under 2 seconds.

As an aside, I also think that the zener used for regulation also has to be doubled up. If the zener should fail O/C and my circuit happened not to be sinking current at the time the voltage could rise to well above the capacitor rating, so causing tha capacitor to blow. There has to be a backup current path to prevent this.

Mark.

 

[duffy]

The requirement is that the circuit should not provide any dangerous voltages to the user should a single component fail short or open circuit

Well if the inrush limiter fails short there could be many amps flowing through the circuit on power up

Inrush surge current is not voltage. This does not appear to be an issue.

this would charge up the capacitor to the point where the voltages internally can become twice the peak voltage of the mains.

Don't think so. I've built this kind of a supply with a voltage doubler off a switch for 110/220v operation - you have to have a pair of caps in series, then switch one of the mains leads off the bridge to the tap point. No single failure, short or open, is going to suddenly make a voltage doubler out of it.

 

[perky]

Inrush surge current is not voltage. This does not appear to be an issue.
.

Not sure what you mean. If the inrush suppressor is shorted, you'd get very large inrush currents (maybe 100's of amps) - this would last for a very short time while the capacitor charges. That inrush current has to flow through the circuit somehow.

this would charge up the capacitor to the point where the voltages internally can become twice the peak voltage of the mains.

Don't think so. I've built this kind of a supply with a voltage doubler off a switch for 110/220v operation - you have to have a pair of caps in series, then switch one of the mains leads off the bridge to the tap point. No single failure, short or open, is going to suddenly make a voltage doubler out of it.

OK, are you actually using a bridge rectifier, or a pair of half wave rectifiers? Are you using something like Figure 3 of **broken link removed**? The bridge rectifier technique gives twice the current per uF.

Any chance in a circuit example or link for your circuit? I'm thinking of using something similar to this