1. Home
  2. Electronics Design
  3. This is something one cannot simulate ;)

This is something one cannot simulate ;)

Oops. The smell, the smoke...

See

formatting link

That smell that seems to stick to the inside of one's mouth, the smoke rising after a muted pop, this was not just too much current through a resistor.

No, this was a boost converter, and the switching transistor between pins 1 and 2 forgot to turn off! See the photo, more of a pop ... phut ... smoke.

Or something. By that stage it was 'okay, lets start over, something's very wrong here...' I'd socketed the 3063, but couldn't be bothered discovering why it smoked, too many alterations on the matrix board.

So I build a new one and it works first time :)

Funny thing is, the one that smoked is only one of three or four I built without a polyswitch and 16V, 1500W TVS diode! Murphy wins another round. So the next one has the protection components...

Power is a little 12V 5AH SLA battery with 10 mOhm shunt, could'a been worse, there's a couple of 12V 100AH batteries on the floor, but they're fused 35A.

Grant.

Reply to
Grant
Loading thread data ...

sing

1

oke.

very

ring

nd. =A0

rse,

5A.

It has a thermal shutdown according to the block diagram. Clearly the thermal shutdown didn't work. I doubt many companies screen the thermal shutdown, other than QA testing. The block diagram doesn't indicate a safety timer. Sometimes you put a timer in with a setting way longer than you ever expect the switch to be on. These protection schemes are minimally useful in production, but very useful in the prototyping stage, mostly because if a part blows up in the prototyping stage, the customer never feels warm and fuzzy about the part. You must be wondering if this part is going to fail in the field since you don't know why one failed in the lab.

I've gone as far in on-demand switchers to put in a watch-dog timer. If the board gets a supply glitch, you'd hate to have the control logic wander into some stuck state.

You can test thermal shutdown pretty easily. Generally what you do is characterize a parasitic diode on the part. Forward bias the diode with a constant current. Measure the voltage of this diode at various elevated temperatures, well presuming you have a test oven. Then you short the power device and the thermal shutdown should cycle on and off. Basically a low frequency oscillator.

Reply to
miso

I was close to the 40V max output, around 36 or 37V. Yet today I saw 64V on output of another after I'd miswired the voltage feedback, no smoke! I turned it off smartly though. Wired the feedback properly and it still worked.

After a while I'll give up on minimal components and just wire in something that'll work properly, get passed this part and move on. Can always revisit later to lower parts count.

What I was doing at the time was connecting the minus end of a 36V zener to the feedback terminal, felt like I glitched it. The joys of hardware prototyping, trying to minimise external components. Though the feedback pin is next to Vcc, and rated up to Vcc -- probably something lodged under the PCB, it was getting messy, time for a new proto. anyway.

I'm not trying to run anywhere near max power of the chip, high voltage yes.

The chip gets a little warm, so does the inductor, when the LEDs are bunched up close, they get very warm at 20mA, so I may run them at 16 to 18mA. This is a lots of little LEDs driver, not using those monster LEDs (yet).

Grant.

Reply to
Grant

 

Vcc,

No, if the 36V zener was shorted from an earlier probe mishap, then I took comp input to 3 times Vcc, oops!?

Grant.

Reply to
Grant

Was debugging a board once and my first thought was "I dont remember an LED being there". The copper track must have been molten

Reply to
Dirk Bruere at NeoPax

Grant expounded in news: snipped-for-privacy@4ax.com:

..

Heh heh- reminds me of some left over/unmarked ICs I had around in the '80s. For kicks I soldered the AC power cord across both sides of the chip, and plugged it in.

A nice *pop* and a little smoke for the fireworks display.

Warren

Reply to
Warren

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.