Should I add a reverse current protection diode across this L78L12 regulator?

[matthuszagh]

Is the presence of D6 important in the attached circuit? The L78 datasheet recommends its presence whenever the output is > 6 V and there is significant output capacitance. The L78L datasheet doesn't mention this diode, but I would expect it to still be applicable. I'm not sure if this classifies as a case with significant output capacitance. I do have a 100u cap at the output, but there's 10R between it and the regulator. Is there a more quantitative way to know when this is needed?

I already ordered PCBs for this circuit before thinking of adding this diode, so I'd have to redo the layout and get it refabbed. That's ok, but I'd like to be sure it's needed before doing that.

FYI in case anyone isn't familiar with the L78 or L78L, this has a 12V output.

[David Hess]

In the past that diode was added in cases where a crowbar circuit was present at the input.  If the crowbar fires, the diode prevents the reverse voltage across the regulator from destroying it.

The crowbar has to be across the input because if it was across the output, then the regulator would limit the current preventing the input fuse from blowing.

[matthuszagh]

In the past that diode was added in cases where a crowbar circuit was present at the input.  If the crowbar fires, the diode prevents the reverse voltage across the regulator from destroying it.

The crowbar has to be across the input because if it was across the output, then the regulator would limit the current preventing the input fuse from blowing.

Yeah that situation I was aware of. But I read somewhere that just powering down the circuit in the case that there is significantly more output capacitance than input capacitance could blow the regulator series pass transistor too.

[SiliconWizard]

In the past that diode was added in cases where a crowbar circuit was present at the input.  If the crowbar fires, the diode prevents the reverse voltage across the regulator from destroying it.

The crowbar has to be across the input because if it was across the output, then the regulator would limit the current preventing the input fuse from blowing.

Yeah that situation I was aware of. But I read somewhere that just powering down the circuit in the case that there is significantly more output capacitance than input capacitance could blow the regulator series pass transistor too.

Yes. I've seen, and done that typically with LM317 regulators (I think it was in the datasheets) but rarely if ever with 78xx ones. I don't recall if the 78xx had protection that the LM317 did not.

[Ian.M]

Not just crowbar circuits.  Any circuit that could pull down the input voltage faster that the output capacitor normally discharges can cause excessive reverse current flow in some types of linear regulator. (e.g. high current loads fed from the unregulated supply, on loss of supply). If nothing else can pull down the unregulated rail (i.e. there is no risk of a short to ground on disconnection), and nothing can back-feed the output,  reverse current protection is not needed.

Read the specific manufacturer's specific datasheet carefully for any hints if such a diode is or may be required.  Older datasheets often proudly claim a shunt reverse current protection diode is not needed or is internal, and give a peak current rating for it.

As you've got plenty of headroom, if reverse current protection is needed, you could always cut the input track and patch in a series diode.

[matthuszagh]

Not just crowbar circuits.  Any circuit that could pull down the input voltage faster that the output capacitor normally discharges can cause excessive reverse current flow in some types of linear regulator. (e.g. high current loads fed from the unregulated supply, on loss of supply). If nothing else can pull down the unregulated rail (i.e. there is no risk of a short to ground on disconnection), and nothing can back-feed the output,  reverse current protection is not needed.

Read the specific manufacturer's specific datasheet carefully for any hints if such a diode is or may be required.  Older datasheets often proudly claim a shunt reverse current protection diode is not needed or is internal, and give a peak current rating for it.

As you've got plenty of headroom, if reverse current protection is needed, you could always cut the input track and patch in a series diode.

Ok, I think I'm safe then. I have a reverse polarity protection diode at the DC input connection, which is connected to the power input of these regulators (4 in parallel).