5V rail thermal runaway

[Gothenlor]

Hello,

I use a very simple linear +5V power supply in a tube amplifier to control relays and leds. See attached schematics.
The amp when in a hot environment, after 2-3 hour powered on, F2 fuse blows. If I replace the fuse, it works again for 2-3 hours.
I replaced the LM1117 without success.

LM1117-05 is a SOT223 package, with a top and bottom polygon plane, connected with vias for minimal thermal management. (see attached)
When I probe the temperature of the top layer on top of the regulator, it goes up to 65-70°C and then the fuse blows.

Total consumption is about 145mA and Pd of the regulator about 0.36W. SOT-223 Tja is 62°C/W so the temperature should be 22°C above ambient temperature, it matches with my temperature measurement.

So what in a so simple circuit could cause this ? There is also MMBF170 not far from the regulator but should not be influenced by this temperature.

[moffy]

The only way you are blowing the fuse is by excessive current > 0.5A, you should monitor the current and find out where it is going as the LM1117 has an 800ma current limit.

[Kim Christensen]

I use a very simple linear +5V power supply in a tube amplifier to control relays and leds. See attached schematics.

Does the 6.3VAC winding power anything else? ie: Is the 6.3VAC side "grounded" in any way or have a path to chassis ground via a filament or something else?
I assume the 145mA measured is true RMS AC current through the fuse.

[MathWizard]

What voltage do you get from the rectifier ?

[Gothenlor]

Does the 6.3VAC winding power anything else? ie: Is the 6.3VAC side "grounded" in any way or have a path to chassis ground via a filament or something else?
I assume the 145mA measured is true RMS AC current through the fuse.

The 6.3V winding is isolated and dedicated to the 5Vdc rail. I meant 145mA DC.

What voltage do you get from the rectifier ?

6.55Vdc

[Kim Christensen]

Does the 6.3VAC winding power anything else? ie: Is the 6.3VAC side "grounded" in any way or have a path to chassis ground via a filament or something else?
I assume the 145mA measured is true RMS AC current through the fuse.

The 6.3V winding is isolated and dedicated to the 5Vdc rail. I meant 145mA DC.

Ok, rough calculation says that would result in apx in 230mA of AC current in the transformer. Still not enough to pop the fuse.

What voltage do you get from the rectifier ?

6.55Vdc

With 6.3Vac, the peak voltage is 8.9V so if you subtract 2x diode drops from that, you should have around 7.5Vdc (A little less because of winding resistance). Calculating backwards, it looks like your transformer is outputting 5.6V AC when loaded.

[Stray Electron]

  Does the LM1117 have built-in overcurrent and thermal protection?  if it does then the problem is either in the 1117 itself or something ahead of it is bad since anything that follows shouldn't be able to overload F1 (the LM1117 protects F1!)  What kind of heat sink do you have on the LM1117?

  I would break the circuit at the output of the LM1117 and put an amp meter in series with it and monitor the current and see what it's doing.

[tooki]

Does the LM1117 have built-in overcurrent and thermal protection?

Yes, of course.

if it does then the problem is either in the 1117 itself

Didn't OP say it's been replaced?

or something ahead of it is bad since anything that follows shouldn't be able to overload F1 (the LM1117 protects F1!) 

Huh? F1 is 500mA, the current limit in the LM1117 is 800mA min, 1.2A typ, 1.5A max.

What kind of heat sink do you have on the LM1117?

It's shown right in the original post: it's an SMD part with copper pours for heatsinking. Based on the TI datasheet's examples, I guesstimate they've actually got an R_thJA of about 80K/W. (Looks to be about 0.5 square inches of copper on the top, 0.25 square inches on the bottom, with the regulator on the bottom.)

I would break the circuit at the output of the LM1117 and put an amp meter in series with it and monitor the current and see what it's doing.

I think we'll see that far too much current is flowing. Something else in the device is heating up and then drawing too much power.

[tooki]

Hello,

I use a very simple linear +5V power supply in a tube amplifier to control relays and leds. See attached schematics.
The amp when in a hot environment, after 2-3 hour powered on, F2 fuse blows. If I replace the fuse, it works again for 2-3 hours.
I replaced the LM1117 without success.

LM1117-05 is a SOT223 package, with a top and bottom polygon plane, connected with vias for minimal thermal management. (see attached)
When I probe the temperature of the top layer on top of the regulator, it goes up to 65-70°C and then the fuse blows.

Total consumption is about 145mA and Pd of the regulator about 0.36W. SOT-223 Tja is 62°C/W so the temperature should be 22°C above ambient temperature, it matches with my temperature measurement.

So what in a so simple circuit could cause this ? There is also MMBF170 not far from the regulator but should not be influenced by this temperature.

You seem to think there's a thermal runaway in the regulator, but this is clearly false: if it's reaching 70C*, the temperature isn't getting anywhere near the thermal cutoff. The fact that it's the fuse blowing means that there's plain and simply too much current being drawn. If the regulator's current limit or thermal cutoff were tripping instead, the fuse wouldn't blow.

So really, you need to monitor your device as it heats up to find out which part of the load is starting to draw too much as it heats up.

*R_thJA is junction to ambient, so if you're measuring 70C on the opposite side copper plane, the junction temperature will be higher, in the case of small, non-thermally-optimized SMD packages like SOT223, possibly significantly higher. But the fact the fuse is blowing tells us it's not reaching the 125C where the thermal cutout engages.

[jwet]

I suspect very high inrush current charging up that 2200 uF cap.  If you have a stout transformer winding with very low resistance and that's a high quality cap, this could be pushing a lot of instantaneous current through that fuse.  An instantaneous 10x fault will degrade if not blow a standard fuse.  I know you say that the circuit fails after a few hours, I suspect that the fuse is getting worked on power up. starts to melt but doesn't quite there just creating high resistance and reliability issue.  After a few hours, it gets fully warmed up and lets go.  Some solutions or things to try, use a T or TT (slow blow) type fuse, use a 1 amp fuse (you're protecting wires),  add a NTC startup circuit or buy your caps from a dodgy Chinese suppliers so they have higher ESR (joke).  Monolithic regulators working below their abs maxes are almost never the problem, they're close to  bullet proof.