Simple power supply overheating components, shorting zener

What are the temperature limits on the various components and what temperature are they experiencing?

The ceramic power resistor got hot enough to discolor, and it melted a plastic IC socket next to it. I ran the circuit for about an hour, and the resistor was just a bit warm. I did not notice any other components with elevated temperatures. There is only an 8 pin PIC, and two dual solid state MOSFET relays in DIP packages. Ambient conditions are normal industrial/commercial temperature ranges, about 0 to 40 C. The inside of the test set might get hotter than that, but it is forced air cooled. I plan to run the test set at maximum output and try to see if the area near this component gets unusually hot.

Thanks for the quick reply.

Paul

Dan Thomas

That is a good point. It might be as high as 25 amps at 280 VAC input. The time constant should be only about 15 uSec, however, so I don't think that would cause the resistor to overheat. It might explain the shorted zener, however, although the electrolytic capacitor across it should absorb the spike. Maybe the customer has a very noisy connector or switch that causes the voltage to turn on and off rapidly.

Paul

I would doubt that if 7805 turns on and off for some line conditions and draw more current and power when it restarts its operation than in steady state. Did you try to increase the value of C1 and use 105C capacitor. You may also decrease the value to see if the problem occurs on your side. Why do you use 12 V zener by the way? did you try 15V?

Paul E. Schoen wrote:

tested

has

mW.

drop

Spikes on the AC line, or ACLO having a substantial voltage with repsect to gnd ... if this is connected ot the mains it can see harsh voltage spikes from other eqpmnt turning on/off etc.. I would suspect a basic generator is just as bad if not worse unless there is some sort of filter/protection.

Any spikes would cuase large currents to flow through the capacitors etc, maybe it might be worth using a capcitor from the end of c2 down to aclo to form a capacitor divider.

Colin =^.^=

I cant read the circuit too well but it looks like C!/2 are failing. Why do you have 2 caps in series? Are they rated for direct mains connection (X2) ?

There might be something else on the circuit that is injecting nasty high-frequency crap into the line, and series line inductance is allowing a lot of energy into your unit. For example, a noisy AC-motor inverter drive or something like that. Short of a proper transformer-isolated design, for a fix, my suggestion is an external power resistor of at least 100-200R (maybe a 5W-10W ceramic type). Check your zener surge rating and substitute a TVS if necessary to stay within the maximum ratings under all conditions (including transient). For future designs, I suggest you use a bigger octal case and put a suitable transformer in there if this is an industrial application. That kind of capacitive dropper 'stuff' is really only (barely) suitable for low-end consumer applications.

Best regards, Spehro Pefhany

Several potential problems:

(1) What is the voltage rating of those 2uF capacitors? It better be at least 600V each. Power lines have spikes on them! As a real-world example, when I open up old tube radios, about 20% of them have exploded capacitors across the 120V power line. Those caps are sometimes 200VDC rated paper capacitors, sometimes even 600VDC.

(2) Did you know that many types of metallized film capacitors are unabashedly designed to short out several times a second? The metal film is very thin, so the short clears itself as the film vaporizes, but you do get a heck of a current pulse.

(3) Imagine if the guy turns off the power switch just as the line voltage is at its negative peak. At 280 VAC that's 396 volts. Then he waits a second plus 1/120th of a second and flips the switch back on again. Now you have 392 volts positive from the line going into the lead of a capacitor that is charges to minus 392. That feeds your diode, the zener, and the 15 ohm resistor with SEVEN HUNDRED NINETY TWO volts. Current could be as high as SEVENTY EIGHT amps. Waay too much current for all the components involved.

(4) Even if the components can take the current, imagine what happens to that poor resistor. It has 700+ volts across its terminals. If it's one of those spiral-cut resistors, it could arc over from one turn to the next, likely leading to the same thing happening to the next turn, and the next, and the next.... leading to really high currents there for a few milliseconds. You need a resistor there with a continuous voltage distribuution insiude, like a carbon comp type. Even then I'd worry about it blowing apart.

Methinks you're in a no-win situation. How's about you buy a tiny wall-wart transformer, or such? See

oops, a few afterthoughts:

(5) I hope you're not making a 1uF non-polar capacitor out of two 2uF polarized reverse connected electrolytics! Many electrolytics are designed for smoothing applications where the ripple voltage is a small fraction of the applied voltage. In the two-capacitor case, that's not true. Plus since electrolytics have an up to 200% tolerance, the smaller one may end up with up to 50% reverse voltage applied across it. That's not a good thing.

He says they're film. You can get film caps rated for 250VAC line voltage, but of course they are larger and more expensive than the ones that will work for a while then short.

Best regards, Spehro Pefhany

Spehro Pefhany a écrit :

But these are still not guaranteed to not show momentarily shorts.

This circuit cannot work. It is a charge pump. It will kill the zener with the inrush current and then will pump up to 600V into the cap(without a load), yes it is a voltage doubler. It cannot supply more than 30mA(on 230V) either with a horrible ripple. Or are these 2u caps parallel? Why don't you use a bridge rectifier instead?

oops, right, they're film, the question is what is their voltage rating, and is it enough to guarantee them not shorting under all conditions?

Maybe you could fit a transformer in there-- a 1watt one at digi-key is only 1x1.3x0.8 inches. Downside: $12 !!

Or a complete 5VDC wall-wart from the same place is only $3.95! I'd seriously consider this unless it is awfully uinconvenient for the user.

I believe the caps are 250 VAC polypropylene, although possibly 400 VDC. If a transformer were used, it would have to be rated at 250 or even 300 VAC primary, if the secondary is to be accurate enough for measurement (it cannot saturate). There is limited space in the portable test set. Cost is not a major factor.

Paul

Alternatively you can do the whole shebang of voltage detection with a LP339 quad comparator and a snootful of resistors all which can run off microamps.

Or a CMOS quad schmitt NAND trigger with even fewer resistors.

Probably huge line transients. There's a low-impedance path through the input caps and the 15 ohm thing, and it's all silicon after that.

Try replacing the zener with a transzorb, a zener rated for very high peak power. Better yet, use a small transformer supply or buy a tiny universal-input switcher.

John

The circuit *does* work. The capacitors (in series) act as a voltage doubler, but the load of the zener diode and the 100uF filter cap limit the voltage, and absorb the energy on the forward conduction cycle. A bridge rectifier cannot be used because the measuring circuit must be referenced to the ACLO point. The 15 ohm resistor does add some error (about 1.5 VAC) to the AC voltage reading, however, so it might be better to move the reference GND to the ACLO line. I am very limited in space, so I cannot add more (or larger) components, unless I go to a SMT design.

Thanks,

Paul

The GND in the schematic is an internal reference ground for the rest of the circuit. The capacitive divider might be worthwhile to consider, but I'd rather pump mostr of the energy into the filter capacitor across the zener. I will look at the waveforms with a scope to see if there is anything to be concerned about, but this circuit has only failed at the customer's site. I will see if I can talk to him and find out how he is powering the unit. I don't think random voltage spikes would cause the resistor to overheat, but possibly some sort of SCR phase fired regulator could be a serious problem with a repetitive sharp rise time waveform.

Thanks,

Paul