Purpose of diode in Linear Regulator circuit

No, the diode is reverse biased so it doesn't normally conduct.

Where it is useful is as stated, when the output voltage exceeds the input voltage. That may at first seem crazy because the regulator can only drop voltage but consider what happens when the input voltage is removed. If there is sufficient capacitance across the output or you are using it as a battery charger, when the input drops to zero, the output might still have voltage on it. In that circumstance, the diode forward conducts and keeps the input pin at least as high (minus the diode drop) as the output, preventing damage to the regulator.

Brian.

Hi,

Hawaslsh said:

I'm not sure what situations would cause the potential to be reversed

Click to expand...

In the moment you switch off the power supply ... the input voltage may drop faster than the capacitor buffered Vo.
Then the voltage is reversed.

There are many other possible situations: charger removed from a battery operated device...

Klaus

Thanks all. Pretty obvious its in reverse bias under normal operation.

Final question before making as solved. It seems that diode can be any general purpose Si diode? We have some 1N4148s around the board in other places, i assume that would suffice. The particuar one we are using has a non-repetitive peak forward current limit of 2A for us.

Thanks again for the help.

Bipolar linear regulators used to be largely NPN emitter follower
outputs. The emitter-base breakdown there is about 7V. With a fat
output capacitor and a VOUT > 7V, you could avalanche the output
transistor with various results, some of them not good. Like, the
same junction can be made into a zener-zap trim element, or just
see its Vbe drift (causing the part to deviate from its original
tested-good output qualities, a little or a lot).

Normal operation is only the first veil. A successful product must
do more - it must prevent transitional and "common abnormal
conditions" from damaging it, and hence your reputation for using
a doomed component in your design.

Hi,

Hawaslsh said:

It seems that diode can be any general purpose Si diode? We have some 1N4148s around

Click to expand...

1N4148 is advertised as SIGNALdiode...for a reason.

But a powersupply is a POWERsupply.
Thus better use diode for POWERsupplies:

1N400x datasheet tells:
"TYPICAL APPLICATIONS
For use in general purpose rectification of power supplies,
inverters, converters, and freewheeling diodes application."

Klaus

The fact that this is a power supply does not mean the diode in question has to carry much current.
The diode just has to keep the input from dropping if the input is disconnected, so the diode doesn't have to carry any significant current (unless the input were shorted to ground, which would be an usual occurrence), so a 1N4148 should be fine.

Hi,

crutschow said:

so the diode doesn't have to carry any significant current (unless the input were shorted to ground, which would be an usual occurrence), so a 1N4148 should be fine.

Click to expand...

I generally agree with your post.
But: the point is "unless the input is shorted...:

So IF the input is "not shorted" --> then you don't need the diode at all. The regulator is happy.

You need the diode to protect against reverse current.
We don't know the OP's application, the voltages, the size of capacitiors, the expectable reverse current, the timing.
We also don't know the critical reverse voltage.

So when I did my recommendation above .. I did it on the "unknown" information, this means for me "the worst case".
Maybe in 95% of all cases you don't need the diode at all. But in the 5% of cases it should protect the regulator for the unknown.
An 1N4002 is rugged, designed for high peak currents and low voltage drop at these current peaks (compared to 1N4148) and can stand a lot higher pulse energy ..... and all this at almost no additional cost (0.00 ... 0.03 Euro at Farnell at 100pcs).

I don't see a valid reason to recommend a questionable protection level.
Size and cost are not.

The whole situation changes as soon as the OP can give the "specifications" of the reverse voltage condition.
Then we can do calculations ... and there may be a good chance that a 1N4148 can be a suitable choice with the ability to safe 30€ per 1000 pieces sold power supplies.

Klaus

The diode drawing is a little hard to follow as the positive bits are at the bottom, the diode is a common addition to lin reg ckts to prevent Vout causing damage if it becomes larger than Vin.