[TriEmbed] How to Damage an LDO

[Pete Soper]

The question was about a 7805, which is a very, very old design. The 
datasheet I pointed to was published in 1976 and I cited it as a 
reaction to Carl's post.

But I just ran a modern 7805 backwards and at seven volts (notice that 
for *this chip* TI said the hazard is for voltages "greater than 
approximately seven volts")  it draws very little current. At 30 volts 
it draws a fraction of an amp but a quick test showed it still worked 
when connected normally before I tossed it into the trash can. So even 
if the design hasn't been updated this thing is not fragile. And no 
amount of capacitive load is going to cause more than five volts to go 
through the output to ground if the source current is turned off.

I didn't actually care about Brian's question. I wanted to make sure 
folks realized the implication of Carl's suggestion.

But I used the thread as an excuse for the sermon about reverse biasing 
with MCU GPIO pins, which is a very serious issue and the mechanism that 
some folks are inadvertently using to get "dead pins" or completely dead 
chips. But that had nothing to do with Brian's question and looking back 
I see that the example of "VCC+.3v" in the context of an MCU pin was a 
really awful confusion factor to throw in. It was a terrible mistake to 
mention this sooner than later in my posting.

-Pete
PS If anybody is interested in a step by step explanation of the MCU 
hazard I piggy-backed here's a good starting point 
<https://wisense.wordpress.com/2014/11/28/interfacing-safely-with-the-msp430/> 
(written about a TI MCU family, but *depending on the data sheet* 
applicable to many other MCUs having very vaguely defined diode current 
limits. The takeaway is to avoid accidentally powering your application 
via current through a GPIO pin!) And yes, this hazard applies to other 
kinds of chips, not just MCUs.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20151210/3a3eaa24/attachment.htm>