Electronics > Projects, Designs, and Technical Stuff
Protecting against Voltage Regulator failure?
OM222O:
--- Quote from: kizmit99 on March 05, 2019, 12:43:59 am ---
--- Quote from: Doctorandus_P on March 05, 2019, 12:33:08 am ---It is a part of the price you pay for cheaping out.
--- End quote ---
Hey! I resemble that remark! ;)
--- Quote from: Doctorandus_P on March 05, 2019, 12:33:08 am ---This is why Lab power supplies with adjustable current limit are invented.
If you set it to a voltage and current slightly above what your board needs then if it fails, on over current, the output voltage collapses and your circuit has a chance to survive. 3V3 circuits tend to be more sensitive to abuse than the older higher voltage stuff.
--- End quote ---
Yeah, just happens that my bench supplies are on a different work area (and too bulky to use where I was doing this testing). If it's of any consequence, that's what I used *after* it melted down to figure out what went bad... ::)
I guess most of what bugs me here is the regulator failing with Vin tied to Vout. I'm not too worried that it blew up ('specially if I shorted it), I just think it's a pretty damn inconsiderate way to fail... :-//
--- End quote ---
to be fair the "inconsiderate" way to fail is how any LDO will fail ... they have an output element (be it a BJT,darlington pair or a fet) which is controlled by an op amp and is connected to input on the other end. If something happens and the device fails, it usually occurs on the output element. BJTs and FETs both tend to fail short ... so there's you're answer :D you can engineer the output element to fail open but that would drastically increase the cost of the device, so they're not really used in LDOs.
kizmit99:
--- Quote from: helius on March 05, 2019, 01:21:04 am ---See https://www.analog.com/media/en/technical-documentation/application-notes/an88f.pdf
Tantalum or aluminum electrolytics can both be used as snubbers, as long as they have sufficient ESR to damp this parasitic oscillation.
In your case, if there is no ceramic bypass capacitor, I don't think that this is your problem.
--- End quote ---
Very useful app-note :-+ thanks very much for that link!
The dev board does in fact have a 22uF ceramic bypass cap on the input (although it also has a 47uF aluminum electrolytic, fed through a diode though so that might effectively isolate it), so I would say it's looking like this effect could be responsible for the failure of the ASM1085. And I guess everyone (except me :o) sees Vin to Vout shorts as an expected failure mode. I can accept that, but I have to say I've used a fair number of 7805 type regulators over the years and haven't experienced this type of failure before... Live an learn though...
Just for completeness sake I took a look at the unloaded output of both the 5V and 9V "wallwarts". Both are providing nicely regulated DC power (5.2V and 9.2V respectively) with less than 10mV ripple. The 5V one actually is labeled as a "switching power supply", so perhaps referring to these as "wallwarts" was misleading (although they definitely fit my definition). I don't think the suggestions that the actual Vin could have been significantly higher than the spec'd values apply. [Just to be clear, I'm only referring to comments like "the 9V power was probably something like 12V", not the ringing effect.]
Thanks to all for the suggestions about adding a crowbar protection circuit. I've never bothered with something like this in the past, but I'm finding that with surface mounted parts the debug and repair process is significantly more of a pain the arse than it was back in the day (sockets!). I will look into what that would take to add to my design.
Since there seems to be a decent amount of helpful suggestions being offered - I figured I'd actually draw up a schema for the complete power plan for these boards (attached below because the imgur pic probably isn't readable).
The design is actually two separate boards (Main and Mega) that connect together, plus the Cyclone dev board rides on the Main board as a daughter board. (So three boards in total, two I control the design of, the dev board I use as is). The Main board (plus the cyclone board) can be used standalone without the Mega board. The Main board runs on 3.3V and basically just provides I/O for the FPGA. The only exception being it needs to provide 5V to a PS/2 keyboard. The Mega board is a re-spun Arduino Mega2560 with some peripherals - so it runs mostly on 5V, with some 3.3V level translators. The power circuit on the Mega board is mostly lifted directly from the Arduino reference design (to support running on either external power or from a USB cable).
The idea was that I could run just the Main board by applying power to the Cyclone board. In version 1 this was just 5VDC and I used that to power the keyboard. In the next version I added the Mega board and wanted it to run on "normal" Arduino style power adapters. I also was concerned that I would accidentally plug a >5V plug into the dev board, so I added the 5V regulator to the Main board. The intent was if the Mega board was present everything would be powered from the Mega supply, but I didn't want anything burning up if both where accidentally present.
Having drawn this up on one sheet for this post, I now see several glaring problems with the overall design:
* With VDC2 being provided by Mega (and no VDC1) there is no 3.3V being generated for either the cyclone or the Main board - Doh!
* If running on USB power there is (potentially) neither VDC1 or VDC2, so again no power to Main or dev boards.
I think both of those can be worked-around by dropping D4 (diode from VDC2 to VDC_IN on Main board, and adding a diode from 5V (supplied by Mega, on Main board) to VDC1 on Main - which would feed 5V to the regulator on the dev board. Of course plugging in power on J1 would be a no-no in that mode...
Oh well, I'm just babbling now -- thanks for the suggestions :-+
Navigation
[0] Message Index
[*] Previous page
Go to full version