TO-220 exploded...estimated overload?

[eccerr0r]

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

Uhoh.

BYQ28E dual diode is getting really hot in a boost PSU with a 120V 100W lamp as a load...

Got to 60°C, stopped it before it got hotter. Weird.

BYQ28E should be a fast recovery diode at least, not sure why it's getting so hot.

[stj]

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

maybe it needs a snubber

[redwire]

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

Silicon rectifier diodes will dissipate roughly a watt per amp. This is with a center-tapped transformer, dual diodes, so one on 1/2 the time. A bridge dissipates twice as much with two diodes on at a time. Schottky rectifiers have lower Vf and make less heat.
If you want highest efficiency, synchronous rectification is the way to go.

BYQ28E with no heatsink and 1W will rise to 75°C

[eccerr0r]

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

[sarcasm] I need a 200V Schottky diode ... [/sarcasm]

Well, if you get the above sarcasm, that's why I'm using a supposedly fast regular diode. Actually I'm running both diodes in parallel so it should help, though who knows how balanced the diodes are. Would hope they're fairly well balanced.

I have a heatsink on it, though it's a fairly small one - about 1"x2" so not an "unlimited" one that would guarantee Tj=Ta. Its thermal resistance is fairly high due to size. I calculated out about 2W so the thermal resistance of the TO220 would require at least a small heatsink, but apparently once again I miscalculated. Not exactly sure how much current is going through it because it's pulsed, of course.

I will need to take a 'scope to this and see what the waveforms look like. I think my 3843 is also getting quite toasty too after some circuit mods to save the IRF640...

[eccerr0r]

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

And grr... speaking of synchronous rectification, once again I have to deal with high side control. BOOO...

After thinking about this a bit, I do wonder if it stays at 60°C or not long term, if it doesn't get hotter than 60°C it might be fine. Ambient is 25°C so it's not too horrible, just need to make sure it doesn't melt the shoulder washer (Nylon melting point at 190°C???!! Wow, I must have been running this diode way too hot when I killed the other IRF640) ...

[eccerr0r]

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

One thing I didn't have... large thermal mass ... or a fan.

So I added a fan, it stays cool. Much better.

[stj]

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

you dont need mass, you need exposed surface area - because you want to radiate the heat - not hold on to it.

[eccerr0r]

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

well, it's hard to increase aluminum mass without also increasing its surface area...
Of course just increasing its mass isn't enough but it at least will give some buffer time to radiate the heat.

The cheap crap piece of metal I was using came from an old PSU, it was a heatsink, increased surface area from the tab of the TO-220 to a fairly large area, just apparently not enough area...

[stj]

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

a sheet of copper - or a bit of flattened pipe will help

[sam_sam_sam]

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

Why not use a heat sink with a built in fan

Or

Make one with a temperature speed sensor

[eccerr0r]

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

Too much work, inverter has enough/too many parts as it is...

Dammit. The number of components needed for the microcontroller interface to the FETs is disgustingly large, though it seems to be the only way I can ensure they turn on and off cleanly...

(I was planning to perfboard this but it's coming up to too many parts to perfboard. Might have to solderless breadboard to ensure the basic functionality works as SPICE says, then I have to layout and etch a bonafide PCB for this... and hope that I don't have any mistakes that will leak more magic smoke.)

[eccerr0r]

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

MELTDOWN INITIATED.

YAY I KILLED THE BYQ28E!

Sigh. It doesn't look way too "interesting" yet, and the fan blew enough heat away from the pins to prevent solder meltdown.

I guess the fan wasn't enough, need more conduction and surface area away from the diode. Then again, perhaps the diode was already damaged from the MOSFET overheating. Or then again I was trying to power a 100W and a 60W 120V lamp in parallel. It was struggling...which is unfortunate, as I was hoping for it to do 300-500W minimum.

BTW failure mode: short.

[stj]

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

maybe you need to be more selective in relation to the fet and diode on resistance/voltdrop

[eccerr0r]

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

Maybe. FET actually stays cool now, washer is still perfectly intact, despite the diode melting down. I was using the destroyed shoulder washer so double destroy, no loss.

Since the diode was overheated from the previous incident, I don't know if it was partially damaged from then or not. But either way I need a new one now, have to see what's in my parts bin...

And no I do not have any 200V Schottky diodes.

----

Candidates:
F5KF20
Another BYQ28E-200
5THZ52
BY329-1200
MR821 (nope)
S30SC4M (hah)
D83-004 (I think not)

Hmm... looking at the specs the BYQ28E-200 is the best of them... ugh.

[eccerr0r]

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

Another candidate:

30DF2
MR851 (nope)
MR750 (no way)
1N1614 (WTF?)

heh.

[sam_sam_sam]

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

Candidates:
F5KF20
Another BYQ28E-200
5THZ52
BY329-1200
MR821 (nope)
S30SC4M (hah)
D83-004 (I think not)

Hmm... looking at the specs the BYQ28E-200 is the best of them... ugh.

Another candidate:

30DF2
MR851 (nope)
MR750 (no way)
1N1614 (WTF?)

heh.

( you have me laughing here ) Things not going so well for you with this project

[eccerr0r]

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

Yep when on a limited budget, that's what happens.

The sad thing is if you know exactly what each part is without looking them up in google or whatnot... then again if these parts are in your hand to look at, looking at them is more amusing.

Yes some of those diodes are slow recovery and reverse voltage breakdown is way too low, especially the schottky diodes. Also the current ratings are too low for many of them as well.

[stj]

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

how about using fets instead of diodes for syncronous rectification?

[eccerr0r]

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

Again, this would be high voltage, high side rectification (boost converter), and if high side control for motors is bad, this is even worse! Synchronous rectification for buck converters is much easier... (plus I need more MOSFETs and that's already in short supply )

I pulled the dead BYQ28E (surprisingly this did not explode like the two transistors earlier in this post) and one of the two diodes were shorted. The other one appears to still be "fine". On the positive note, having the diode short out saves the switching transistors, and luckily my "supply power" is "current limited" (and need MPPT to truly extract the full amount of power from it.)

I decided to torture the F5KF20 next, but it's now on a larger, finned heat sink.
Fortunately no shoulder washer needed. I don't have to worry about electrocuting myself on the diode flange.

[stj]

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

is that rust on the legs?