fire! / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

fire!

1 2 next

Move Thread

LAN_403

Ben Solon

Sat Apr 28 2012, 07:21PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

i built uzzors' flyback aux supply and i have checked and rechecked my wiring. what happened was when i plugged it in, the 10 ohm gate drive resistor went up in flames and then the nearby live mosfet fed mains through the fire. the 1k low pass resistor met a similar fate maybe .2 seconds later(am i imagining things?).

after i cleaned up all the soot, i then removed the ic and scoped different points on the circuit:
-all of the mosfets pins where live
-pins 3 and 6 where live(derivative of above)
-power supply now shorts through the transformer and the shunt

i am wondering how this came about. i think i know how events unfolded after the first fire, but what caused it is a mystery to me. when building the thing i checked individual sections separately for functionality before wiring them together. maybe the resistor had to dissipate too much power driving the gate?

mister_rf

Sat Apr 28 2012, 08:38PM

Registered Member #4465 Joined: Wed Apr 18 2012, 08:37AM
Location: Bucharest, Romania
Posts: 145

You need to give more details about the transformer and the power MOSFET you have used forâ€¦ Maybe there's a problem.

dude_500

Sat Apr 28 2012, 10:37PM

Registered Member #2288 Joined: Wed Aug 12 2009, 10:42PM
Location: Cambridge, MA
Posts: 179

The gate resistor would not immediately go up in flames and exhibit a mains-fed flame if it were merely overheating from too much gate charge. This sounds like some major wiring fault.

Alex M

Sat Apr 28 2012, 10:50PM

Registered Member #3943 Joined: Sun Jun 12 2011, 05:24PM
Location: The Shire, UK
Posts: 552

This one?

We need more details.

Just a though, if it is that one can you check your mains voltage? Perhaps it is a bit higher than 230vac (mine is 253vac).

Also for mains to be on pin 1 I would say there is something touching where there shouldn't be. Triple-check everything using a continuity meter. Also perhaps you got one of the windings connected the wrong way around on the switching transformer.

Ben Solon

Sat Apr 28 2012, 11:22PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

I figured out what I did, but it still doesnâ€™t explain the massive destruction.... I swapped pins 4 and 5 thinking 4 was ground. But as I said, why would that cause a fire? It works fine after several hours of replacing parts and scraping soot (minus a small deadtime issue that is easily fixed).

For parts, I am using fqpf7n60 fets from Fairchild. Other than that all of my semiconductors are identical to Eirik's design.
As for the transformer it is a flyback from a smps which operates at 55 kHz. I have compensated the timing components to match this. And I know itâ€™s not overvoltage because my system is designed for 170vdc as I live in the u.s.(I donâ€™t think 170 could possibly ring over 600 do you?)

Note Alex1M6: your picture looks like it needs to be re-sized slightly.

dude_500

Sun Apr 29 2012, 05:43AM

Registered Member #2288 Joined: Wed Aug 12 2009, 10:42PM
Location: Cambridge, MA
Posts: 179

If you didn't properly ground the control chip, any number of things could have happened. It is an unspecified condition of the chip, so for all you know it caused the FET to go to an always-on stage which would be instant destruction for everything, including control circuits if the FET failed in a way where gate and mains were connected

mister_rf

Sun Apr 29 2012, 06:45AM

Registered Member #4465 Joined: Wed Apr 18 2012, 08:37AM
Location: Bucharest, Romania
Posts: 145

When powering up a SMPS it is desirable to minimize the chance of blowing your circuit if there still be a fault. There are two ways of doing this:
-bring up the AC line voltage gradually by using of a Variac
-and the second solution the use of a series high power load to limit current to the power MOSFET.
Known as â€˜The series light bulb trickâ€™, thatâ€™s in practice the more simple solution, just using a normal light bulb (40-60W) since it will limit current (if we are lucky) to non-destructive levels.
To provide adequate protection need to place the light bulb directly in the B+ circuit. In this way it will limit the current from the main filter capacitors to the transformer of power supply.
If we get the light bulb at full brightness, thatâ€™s for sure a short circuit or a major fault probably is still present. When the equipment is operating normally we expect initially a bright light bulb but then settles at reduced brightness.

1335681807 4465 FT137711 Light Bulb Trick

klugesmith

Sun Apr 29 2012, 03:19PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

mister_rf wrote ...
When powering up a SMPS it is desirable to minimize the chance of blowing your circuit if there still be a fault. There are two ways of doing this:
-bring up the AC line voltage gradually by using of a Variac
-and the second solution the use of a series high power load to limit current to the power MOSFET. ...

Good picture there, showing only the relevant details.
Welcome to 4hv.
-Rich

Ben Solon

Sun Apr 29 2012, 06:23PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

it probably was just a fatal mosfet error. seems most likely. but once that was fixed it works just fine(again- deadtime problems). the series resistance in the circuit is just too high to need another limiting factor anyways: i picked my fets because of the high on resistance. right now i am reverting to a low voltage circuit on a breadboard so i can learn how to properly use the ic without the danger of a mains screwup.

but just to get it straight, the chip works like this: chip turns output on for a duration equal to either clock cycle or until current turns it off, whichever happens first. then deadtime is carried out. then reset?

i am new to this whole current mode power supply thing.

mister_rf

Sun Apr 29 2012, 10:47PM

Registered Member #4465 Joined: Wed Apr 18 2012, 08:37AM
Location: Bucharest, Romania
Posts: 145

Let me clarify something, â€™'the series light bulb trickâ€™ itâ€™s only a temporary measure, so the 40W light bulb will be connected in the circuit only during the first tests, because there is no guarantee that the circuit is fault-free.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob