ULN2003A capability - but current limited?

[MartinM57]

Open collector Darlington drivers are great, but if some clown:
- connects a 10R load connected to 12v to them
- or directly connects 12v to them
...due to an installation error, then either or both:
- the ULN emits it's magic smoke
- PCB tracks start to vaporise.

So what's your favourite,(and cheap) way of current limiting, to say 0.5A, 7 or 8 open collector type switched outputs? An inline 21R or so ((12-1.2)/0.5) resistor is not the right answer!

I stumbled on MAX890L, but:
- it's expensive at 8 off per circuit
- it's high side switching
- it's Vcc max is 5v
...so no good really

EDIT: ...and polyswitch fuses look far to slow to me e.g. http://www.rapidonline.com/netalogue/specs/26-0695e.pdf

 

[hippy]

One possibility may be to use a 12V regulator with a suitable current rating and rely on it's over-current / thermal shutdown to mitigate against faults, not perfect though.

Some devices like LM317T can be operated in current limiting mode but I don't know which, if any, would be suitable.

Both would be high-side limiting.

Nothing beats "operator competence" though.

 

[Andrew Cowan]

I like polyfuses for electrics work - not electronics work. For complex wiring setups (such as bulbs or motors), I often stick some polyfuses around to stop wires melting if something shorts. However, I'd agree they don't have much of a place in protecting semiconductors.

Other than that, I just use a current limited PSU until I have finished constructing and testing!

A

 

[MartinM57]

Trouble is that the 12v supply is off board and outside of my control.

It's...errr...how do I dare say it...ummm...maybe in small writing...perhaps...the...err, I am really trying to be responsible about this...but people are going to shake their heads and mutter....but it's the car battery

 

[BeanieBots]

Oh, it's a car battery.

Something like this:-
http://www.irf.com/product-info/datasheets/data/ips1011.pdf
but a bit expensive for eight-off.

Naturally, a fuse in the high side to protect cables goes without saying.
If your loads can tollerate a bit of volt drop, then fuseable resistors can be quite effective and fast.

A 'tacky' cheapskate yet still effective method is to put a single fuseable resistor in the ground path to the ULN2003. This is actually even more effective than on the switch side because the volts developed across it subtract from the logic drive and turn off the transistor while it thinks about popping. They are available in many values so a little thought should find a value which will both protect yet not effect performance.

 

[Dippy]

Do you mean actual current limiting or circuit breaking?

 

[premelec]

Per Hippy's suggestion LM317s can act as series current limiters by just putting a resistor from output to control pin and then take current to load at _control pin_ - the current will be limited at 1.2/R [1.2 being LM317 ref more or less]. There IS a voltage drop across this assembly but it often works quite well... cheap cheap... 2 parts!

 

[Andrew Cowan]

...a single fuseable resistor...

Do you mean actual current limiting or circuit breaking?

I'd imagine circuit breaking would be wanted - we are talking about a large overload (100% extra (1A), say) - a current limitting resistor would probably not be fast enough.

Any help?:
A fusible resistor device is made up of an elongated resistor film, a portion of which is disposed on a film of an organic substance attached to a base plate. The organic film is adapted to be melted by an excessive current flowing through the circuit including the elongated resistor film, whereby the organic film will decay to thereby open the circuit. The organic film is preferably produced by the polymerization of a monomer by a glow discharge within a gas atmosphere of the monomer.

A

 

[BeanieBots]

a current limitting resistor would probably not be fast enough.

You can't get any quicker. It's INSTANT.

However, a fuseable resistor will take time just like a fuse does.
A fuseable resistor is just like any other resistor except it goes open circuit once a certain power dissipation is reached instead of catching fire or giving off lots of smoke and heat.

When placed in the ground side of a NPN type transistor used as a low side switch, as soon as 5'ish volts are developed across it, no more current flows because the transistor is biased off. (basic constant current circuit).

The compromise is the volt-drop of whatever the load current is compared to the chosen resistor value to ensure adequate protection.

Only Martin can say if this or the LM317 (a constant volt loss of ~4v) will be OK for his application.

 

[Andrew Cowan]

I was thinking of a thermal current limitting resistor (as used in lots of electronics). Basically a PTC therminstor with a low resistance at 25 degrees C and a high resistance at a bit hotter.

The problem with a straight resistor finding one with a low voltage drop at normal current (500mA), but with a high voltage drop at high current (1/2A). That's why thermal limitting is used more often.

A

 

[bluejets]

Festoon light bulb. (interior light)

 

[lanternfish]

Pedant Alert! - Me

... The problem with a straight resistor finding one with a low voltage drop at normal current (500mA), but with a high voltage drop at high current (1/2A). That's why thermal limitting is used more often.

A

500mA = 1/2A I take it you meant 50mA?

 

[MartinM57]

OK...just got back from skittles. We lost . Heyho...

Take the worst case scenario of +12v straight to the unprotected darlington collector. Darlington switches on, an almost short circuit of +12v to ground (well, say to 1.2v), the PCB traces act like fuses and vaporise before the ULN2003 gives up the ghost.

Don't really mind if it current limits at 500mA or something breaks the current - as long as it doesn't need component or board replacement due to a moment of user carelessness/incompetence.

I'll go dig out a LM317 datasheet -but I'm after a clever design to give this facility on every darlington in the ULN2x03 at minimum cost. Ta!

 

[lanternfish]

How about a 1N914 Diode in the darlington common line to 0V. It goes pop at 300mA (or thereabouts). Of course there is the 1.6V drop to contend with.

 

[MartinM57]

According to the datasheet I can have 7 outputs conducting about 160mA each at 100% duty cycle (i.e. continuously on) - so over 1100mA down the ground pin...

 

[moxhamj]

Hmm - you need a cheap device that has a low resistance normally but rises to a high resistance when more than a certain current flows. And a visual indication this is happening is a bonus. My favourite device is what bluejets suggested - a lightbulb. And for automotive use there are a range of different bulbs. If the bulb starts lighting up during normal use then it is a bit underrated so go for a bigger one. Start with a festoon light for driving relays, but if that is too low then a brake light (5W) is a good replacement.

1100mA might end up being a bit much for a brake lightbulb. If that does start glowing when all 7 outputs are on, you can move up to a headlight bulb (60W).

 

[Dippy]

Haha, I just knew DrAc would mention his light bulb. It's a nice idea in many apps, but is it faster than a fuse?

Is it faster than a fast blow fuse?

Is it possible to put a shunt R in the gate connected to some threshold device which forces the 2003 input pin high?
I can't think of anything non-electronic that will be fast, and I don't know how much stick the 2003 will take before popping.
Without potentially destructive tests I wouldn't be able to see if a fast blow fuse is adequate.
Do you know for sure Martin?

 

[hippy]

Take the worst case scenario of +12v straight to the unprotected darlington collector. Darlington switches on, an almost short circuit of +12v to ground (well, say to 1.2v), the PCB traces act like fuses and vaporise before the ULN2003 gives up the ghost.

Perhaps you need to take an alternative approach, of which two come to mind -

Design PCB tracks / wiring and use output switches which can take maximum current without damage until the fuse blows out.

Have two drive switches; the first turns on and limits current using a resistor for example, checks for a fault condition and, if none, then turns on the drive switch which doesn't have a resistor.

 

[MartinM57]

OK - some good ideas there thanks. Glad to see I didn't miss any blindingly obvious soution.

Maybe time to wheel out the current-limited bench PSU and scope and try some destructive testing

Ta