How to not blow up switching regulators?

[oldfortune_]

Hey guys,

i'm attempting to  design some circuitry that powers 12V and 5V from 24V using some of the RECOM drop in replacement switching regulators. I seem to have managed to blow a bunch of the regulators under a certain startup condition, and I was wondering if you would have some insight as to why this happened, and potentially some circuit to protect from this critical failure.

Failure:
-the regulators work fine at a stable 24V from a current limited bench top power supply, but when I disconnect and reconnect the power lead they fail? failure results in 24V being pumped into the rest of the working circuitry, which is catastrophic in this use case... The failure does not occur when you keep the power leads connected, and switch on and off the power supply, so I guess it must be something to do with a voltage spike.

I have posted the schematic of the regulator setup, I would be very thankful to hear some insight.

Thanks in advance,
Sean.

[Rerouter]

My main thought would be, what type of capacitors are you using, they may not like a 10uF ceramic sitting directly on the output,

[Monkeh]

The ceramic on the output is as specified and fine - the ceramic on the input is possibly where the problem lies.

The high inrush current of extremely low ESR caps will cause quite a voltage spike due to the inductance of the supply lead - basically works like a boost converter. Putting a moderate size electrolytic capacitor on the input is often a good solution.

This application note is a good read: http://cds.linear.com/docs/en/application-note/an88f.pdf

[oldfortune_]

My main thought would be, what type of capacitors are you using, they may not like a 10uF ceramic sitting directly on the output,

hmmm ok, in the current design i'm using 0805 ceramic caps. How would different caps behave differently? the application circuit suggested two 10uf capacitors, but now I've found another data sheet and it isn't consistent...

https://www.recom-power.com/pdf/Innoline/R-78xx-0.5.pdf

It does mention reverse current protection though, maybe that could have caused the failure at startup?

[oldfortune_]

The ceramic on the output is as specified and fine - the ceramic on the input is possibly where the problem lies.

The high inrush current of extremely low ESR caps will cause quite a voltage spike due to the inductance of the supply lead - basically works like a boost converter. Putting a moderate size electrolytic capacitor on the input is often a good solution.

This application note is a good read: http://cds.linear.com/docs/en/application-note/an88f.pdf

I'll give that a read, Thanks!

do you have any suggestion regarding methods of protection valuable 12V devices from 24V if failure was to occur?

Best,
Sean

[Monkeh]

The ceramic on the output is as specified and fine - the ceramic on the input is possibly where the problem lies.

The high inrush current of extremely low ESR caps will cause quite a voltage spike due to the inductance of the supply lead - basically works like a boost converter. Putting a moderate size electrolytic capacitor on the input is often a good solution.

This application note is a good read: http://cds.linear.com/docs/en/application-note/an88f.pdf

I'll give that a read, Thanks!

do you have any suggestion regarding methods of protection valuable 12V devices from 24V if failure was to occur?

Best,
Sean

A simple crowbar might work. Fuse and a very chunky zener diode. The diode would need to survive the 24V supply long enough for the fuse to clear.

Failing that there are a number of overvoltage protective devices available with a little searching.

[Rerouter]

For your 2 capacitors, electrolytics will be fine, i can actually vouch for your 5V regulator working fine with no input capacitor, and 2.2uF  electrolytic on the output in an Truck / Bus application of slightly more than 100 vehicles. (was powered through a switched mosfet, and fused, but no other protections)

As for protection, use a zener triggered crowbar, you can tweak them with a npn stage to up the gain, then slam a mosfet shut once it exceeds you spec, the goal is to protect the device at all costs.

[T3sl4co1l]

The power-up surge, disconnecting and reconnecting a power supply, is massive.

Easily a surge of 100A.

With low-ESR capacitors on the input side, expect massive voltage overshoot.  This is exacerbated further by low-voltage-rated ceramic capacitors, which lose a lot of capacitance at high voltage, therefore producing an even higher overshoot.

You can check if this is the problem by measuring inrush voltage with an oscilloscope.

Tim

[ajawamnet]

What Monkeh mentioned about that Linear app note is probably the issue. I design and use a lot os SMPS stuff and having a "dirty" electrolytic is usually the fix.

Really read that app note. They give a few solutions, one an in rush current limit is OK but if you look at the graphs the electro exhibits the best result

I've actually benched this with a design where I used one of their parts (we were having failures on turn on) and really does work. My scope shots looked identical to the ones on the attached page.

Just adding an electrolytic at the input fixed it. We had no further failures in MIL FAT and FQT, the units have been in production for quite some time.

[T3sl4co1l]

Ewww, they tried a tantalum... subject to inrush surge... 

The electrolytic is definitely a safer option, as long as you don't need long life or low-temperature operation.  If you do, the 22u ceramic + 0.5 is better (with some optimization for ESR, it would seem).

Tim