Re: Zap a volt regulator w/ no cap??


Tim Wescott wrote:
John Popelish wrote:

eromlignod wrote:
(snip)

I have an application in a machine that has a 24-volt power supply. I
have a tilt sensor (analog output) that needs 12 Vdc as its input
supply. Until now this part of the machine used an embedded circuit
that the sensor mounted directly to and that provided its own 12 V.

I converted this part of the machine's control to be read from a PLC
analog input module, so I had to come up with my own 12 V source. What
I did was add an LM2940CT-12 voltage regulator (TO-220 package) to the
circuit right before the analog sensor. This provided my 12V and
worked great for about a day and a half, then my analog sensor fried.
When I measured the voltage from the regulator I found that it was now
the full 24V input...so apparently the regulator failed first and then
cooked the sensor with 24 V.

When I took a look at the spec. *** on the LM2940, I see that they
recommend putting capacitors from the input and the output to
ground...I didn't do this. Other than this mistake, I can't see
anything else wrong with the circuit. Could the absence of these caps
have caused the regulator to fry? If not, what else could have caused
it? I'd like to know exactly what my problem is before I toast another
sensor (they're $175 a pop).


Low drop out regulators (of which the LM2940 is one) are only stable
with certain values of capacitance (and sometimes only with certain
values of series resistance in that capacitance).
The graph on page 11 shows how tricky this is.
http://info.hobbyengineering.com/specs/NATSEMI-LM2940.pdf

Since you have lots of excess voltage, a follower output type linear
regulator like an LM7812 would be less problematical. It still works
best with a small capacitor from both input and output to ground, close
to the regulator (say, .1 uF) but the values and series resistance are
not at all critical, compared to the LM2940.
See page 22 of:
http://www.fairchildsemi.com/ds/LM/LM7812.pdf

"Works best" is a mild way of saying "leave them off at your peril". I
stopped in the middle of build a circuit with an embedded 78L05
regulator; when powered up without an input cap it oscillated at 80MHz.

You should also check on power dissipation. These regulators are
supposed to have thermal shutdowns, but that doesn't mean they work
right, and you don't want a shutdown event anyway. Check the input
current to your 12V module; you'll be dissipating that times 12V (24V -
12V), and more if your 24V supply goes higher.

Make sure you have enough heat sinking to keep the regulator cool. The
best rule of thumb I know is to run it for a while then put your thumb
on it. If you pull back and say "ouch" then the heat sink isn't
dissipating enough.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

I can't speak for all thermal shutdown schemes, but on the parts I
designed, the shutdown was not very intelligent. That is, it would
sense the die temperature, shut down, then cool due to the shutdown,
power up, and of course get hot again enough to trip the shutdown
circuit. You could make an oscillator using the shutdown. As part of
the characterization of the part, a diode on the chip would be
characterized over temperature, i.e. with the part in an oven and no
load so the die temp equals ambient. Then if you short the output, you
can monitor this diode to see the temperature trip points.

There may be smarter regulators that stay shut down once tripped. It is
a marketing issue since the smart design would need a power on reset
from the input voltage. Not impossible, but POR circuits can be fooled.

.

Quantcast