Recently.. one of my LM317 got shot and fried my arduino circuit. I was using LM317 to convert 16V to 5V. This time I am thinking of adding a zener protector to safeguard my circuit in such senarios , my circuit current consumption is max 50ma @ 5V. I am planning to use attached circuit. My understanding is even if the LM317 gets short and voltage output is 16V, current will be ~ 160 ma after the resistor R3. So my question is what should be the wattage resistor(100R) R3 and wattage zener diode ? and how to calculate the same?
Power is calculated as Voltage X Current, so 5V x 0.16A = 0.8W. ...A one or two watt Zener should work.
It's more common to use a fuse instead of a resistor. You don't get a voltage loss across the fuse (better voltage regulation) and if the regulator shorts-out and the fuse blows, you'll know it.
If you only need 50mA a 100mA slow fuse should be fine. It's by the way more common to place the fuse near the input of a circuit.
Also, it's recommended to place a diode across the input and output of the LM317 (of course with the cathode connected to input, not the other way around )
Zeners are generally bad at overvoltage protection.
TVS diodes (super zeners) are commonly used for that.
A normal zener will blow before the fuse does.
I doubt that a fuse/zener/TVS is fast enough.
Google "crowbar circuits" (zener/thyristor).
Better still is a good supply.
For <$2.00 you can get 5volt switching (micro) buck converters, free shipping.
Leo..
Why don't you use a LM7805, and forget about the resistors, just use the appropriate bypass caps.
The tag of the LM7805 is gnd, so you won't need to insulate the tag from the heatsink.
Tom....
If you use the circuit you have posted, how will you know if the 317 has failed, the zener will just clamp the output.
Use a 1W zener and put a 250mA fuse in place of the 100R resistor, now if 317 fails and output goes to 16V, for an instant more than 250mA flows and blows the fuse, output goes to zero, device is saved.
What you want is a "crowbar". The 5.1 V Zener will trip the SCR if the voltage exceeds about 5.7 V, completely shutting down the regulator. If the regulator has failed, damage to it by this action is entirely irrelevant. Do note however, that the actual design problem in the circuit above, is the use of a variable resistor.
Paul__B:
Do note however, that the actual design problem in the circuit above, is the use of a variable resistor.
+1
A trimpot it not perfect. If you try to adjust the pot while the Arduino is connected, contact bounce will create 15volt spikes on the output.
Leo..
That circuit is very sensitive to contact bounce in the trimmer - the recommended circuit
has a capacitor across the trimmer which both prevents this and reduces ripple amplification,
go look at the datasheet "Typical Application" circuit (Figure 9.2 in the T.I. datasheet).
Watcher:
Any recommended part values for the "crowbar" above?
I'm not sure the R or C are needed - just make sure the zener leakage at the
working voltage is well below the minimum trigger current of the SCR. Note that
SCR gate will be a volt or so above ground so the zener voltage will be a bit less
than you might think - check the datasheet for the SCR in question (it just has to
be big enough to handle the current, so most devices will be fine).
Actually I found the site that crowbar circuit came from and it admits the C prevents the crowbar
from firing with short pulses, which is simply wrong - the whole point of the circuit is to prevent
the output voltage exceeding the breakdown voltage of the load, short spikes are just as able
to destroy as long spikes.
Just the zener and SCR are needed I conclude. In fact I'd add a small resistor in series with the
SCR gate to limit the current through it, perhaps 100 ohms or so. However I'd also then check
the circuit functions as expected with a 'scope to back up my intuition in case I've missed something.
For really demanding applications I would suggest a high speed comparator, flip-flop and MOSFET would
be better to respond really fast (SCRs are not exactly snappy switches).
Remember that there is a cap across the supply line (C2), and maybe more downstream. They should kill spikes.
A small cap on the gate might be needed if the crowbar circuit is used in a switching supply (HF more hash).
I would replace the zener/gate resistor for a 100ohm trimpot (gate to wiper).
And pick the zener 1-1.5volt less than supply.
Then you can trim the crowbar firing voltage, and have some gate series resistance.
Leo..
Capacitor C is there to stop triggering when the supply powers up.
Going from 0 to Vreg has in some instances the same gate trigger effect as Vreg to Vunreg when failure occurs.
Thanks all for your responses. Good to know about crowbar circuit protection. Also there was a recommendation to use 7805 instead of LM317, is that means 7805 is short circuit protected? i.e., even if its failed, there are no chances of input = output?
would replace the zener/gate resistor for a 100ohm trimpot (gate to wiper).
In the context, that suggestion is an absolute scream!
@Paul,
Forgive my ignorance but I'm not seeing the humor in this. What is your point ?
Is it that less than 100 ohms might not be enough current limiting for the zener ?
Or is it that there is no need for adjustment in this application ?
)
