TO-220 exploded...estimated overload?

[eccerr0r]

Re: TO-220 exploded...estimated overload?

Probably burned rosin, though I'm not sure what the grayed stuff on the metal tab is. The good BYQ28E is much more shiny.

Right now my power supply is not supplying enough watts to see how well the F5KF20 is doing under load, though so far the diode is supplying 100mA or so at 120V okay...

[redwire]

Re: TO-220 exploded...estimated overload?

You might be over-torturing things using light bulbs as load? Their resistance is non-linear and much lower when cold.
A 100W 120VAC bulb is 144 ohms lit but at 1/2 voltage drops to 124 ohms, and 1/4 voltage is 90 ohms.
I found sometimes an SMPS can't "get it up" and stays stuck outputting low voltage and high heat. Would have to graph power vs voltage.

[eccerr0r]

Re: TO-220 exploded...estimated overload?

Well, with lower voltage, even with lower ohmmage the amount of current is lower.

100W (RMS)

120V 144Ω = 100W = 0.83A
60V 124Ω = 29W = 0.48A
30V 90Ω = 10W = 0.33A

So the current is still going down as voltage goes down, and diode loss is (V=0.7)*I (though V goes up slightly as I goes up...) which is mostly linear.

The thing I was hoping for is that I would expect the above to be "average" current. Peak current flow is what I'm not totally accounting for, and this would be much higher than 0.83A. Then again most diodes have higher peak current limits.

Anyway I was hoping to get 300W from this supply for proof of concept. That would be less than 3A average, so these diodes should be able to handle it -- as long as they stay cool.

[eccerr0r]

Re: TO-220 exploded...estimated overload?

So far so good, the F5FK20 has survived many hours of powering the 100W bulb.
The switching transistors do not like it over 150W draw however. MOSFET driving is a PITA...

[eccerr0r]

Re: TO-220 exploded...estimated overload?

One thing that really sucks about shoulder washers... if they melt, it will end up protecting the heatsink...

[stj]

Re: TO-220 exploded...estimated overload?

try using a fiber washer on the plastic one, it will stop the screw touching the tab if the plastic melts

[eccerr0r]

Re: TO-220 exploded...estimated overload?

well that's not exactly the problem, when the plastic melts, the whole thing goes loose, and the transistor/diode goes into runaway as the heatsink is no longer making contact... I guess those springs have an added advantage.

[eccerr0r]

Re: TO-220 exploded...estimated overload?

BTW, how do people feel about those plastic covered TO-220s, they clearly have higher thermal resistance, but have people been swapping them with regular TO-220s with insulation readily and still generally satisfying thermal requirements?

[stj]

Re: TO-220 exploded...estimated overload?

some of those have very low thermal resistance, i'v done it a couple of times simply because of parts availability.
that said, i wasnt cooking the old parts either.

[eccerr0r]

Re: TO-220 exploded...estimated overload?

Yeah, that's exactly why I'm using it now, parts availability

When you say you weren't cooking the old part, are you sure?

I guess if the old part failed from ESD or something like that, then it wasn't being cooked. But perhaps the reason why it did fail is because it cooked itself inside the plastic? Heh...

(Anyone have any photos of dead fried plastic covered TO-220F(?)s? Seems that the thermal resistance is quite a bit different. Suspecting that one shouldn't be running TO-220F's anywhere near the limits of the device...)

[stj]

Re: TO-220 exploded...estimated overload?

When you say you weren't cooking the old part, are you sure?

when your repairing industrial boards that arent cased, getting ones with parts broken off by clumsy idiots is not uncommon!

[eccerr0r]

Re: TO-220 exploded...estimated overload?

Ahh yes, forgot about mechanical damage, that indeed is a failure mechanism Actually the thermal resistance of unheatsinked TO-220 and TO-220F's are about the same, so if no heatsink, not much difference. It seems that both are around 62°C/W.

Difference starts when there's a heatsink attached or at least the "infinite heatsink" case (which of course doesn't exist). I'm reading anywhere from 0.7-2.5°C/W for TO-220 and 3.5-6°C/W for TO-220F.

[redwire]

Re: TO-220 exploded...estimated overload?

TO-220F has quite poor heat transfer to a heatink, around half that of the TO-220.
I was going to build an audio power amp using them but they can't get rid of heat:
FJPF5200 (TO-220F) 2.5°C/W
FJP5200 (TO-220) 1.25°C/W
2SC5200 (TO-264) 0.83°C/W

What frequency are you running, perhaps your heating is aggravated by switching losses? The transformer secondary ringing can cause rectifiers diodes to switch on/off many times and see high peak currents in reverse recovery.
It might be worthwhile to lower it and see what happens.

[eccerr0r]

Re: TO-220 exploded...estimated overload?

So you actually tried subbing a TO-220F into a TO-220 solution and had it fail? Or is it just theoretical based on the numbers?

The switcher should be running around 40KHz, and as far as I can see, the "new" diode is working fine now.

My initial design the switching MOSFET and the diode were on the same heatsink, and both went really hot. The MOSFET died, and I suspect the diode was given a fatal blow though barely holding on - at least the diode tester says it was still OK.

Then when I moved it to its own heatsink, it got hot because it was barely holding on...and then killed itself.

The new TO-220F diode seems to be staying fairly cool, though it does have a slightly better heatsink now. Without air blowing on the heatsink, I can't really feel it getting warm at all really, though the added thermal resistance may have some impact. But on the other hand, no thermal pad, and no shoulder washer, so I could torque it down a bit more...

[redwire]

Re: TO-220 exploded...estimated overload?

yes TO-220 fullpacks ran really hot, I tried them(MOSFETs) in mains SMPS instead of all the insulator/hardware to keep it simple. Thermal imaging cam showed over double the operating temp.
MOSFET's have thicker insulating plastic to meet safety standards for high voltage. One Vishay part was 3.8°C/W so at 10W almost 65°C.
40kHz seems fine, if the mosfet has good gate drive and snubber + deadtime, that is most of the battle.

[eccerr0r]

Re: TO-220 exploded...estimated overload?

The switching diode has a different challenge...