The datasheet of the LM317 adjustable regulator says that the max voltage rating of 40V is for the difference between input and output, instead of between + and - supply as for 78xx regulators. That's clear enough.
But it's usually desireable to filter the output of any regulator with a capacitor, and the output terminal will be momentarily shorted to ground when it's first switched on. This will cause the input-output voltage rating to be exceeded in cases where the absolute input voltage is higher than 40V. The datasheet doesn't say anything about this.
If this were a matter of dissipation, it would be of no consequence, but it concerns voltage breakdown. Can anyone clarify this point ? Can the LM317 be safely used with an input of, say 45V and a filtered output of 12V ?
That's obvious. But that's a sustained fault condition. I'm talking about a momentary initial spike, under which many devices can withstand levels that exceed their continuous max ratings.
When what's switched on? If you mean the power source that feeds it then just make sure that the source has a longer time constant.
: This will cause the input-output voltage rating to be : exceeded in cases where the absolute input voltage is higher : than 40V. The datasheet doesn't say anything about this. : : If this were a matter of dissipation, it would be of no : consequence, but it concerns voltage breakdown. Can anyone : clarify this point ? Can the LM317 be safely used with an : input of, say 45V and a filtered output of 12V ? :
If it's a power-line transformer and rectifier setup, the input is going to take at least a quarter cycle to rise. That's your salvation, as you only need to filter the output side for high frequencies, much higher than 240hz. The regulator's frequency response will take care of the lower frequencies.
If it's a switching-mode source, you do have a problem.
In either case I wouldnt push the regulator's limits. Capacitors. especially tantalum and deposited-film ones, have been known to flash-over and short for brief moments. You probably want your design to survive these conditions.
Where I have, I've always had my butt saved by the HV versions of the regulator :-).
At some point you have to draw a line in the sand and say you aren't going to protect component X if component Y if component Z fails. Failed PS actives take out capacitors and transformers, failed PS passives take out PS actives. Sometimes the circuit under supply will let its smoke out too, as long as the whole gadget doesn't catch on fire I don't see what the big deal is :-).
In time-proven commercial designs the manufacturer has good stats about warranty costs for the various failure modes and can make a sound economic decision about what to protect and what not to, with the goal of making a whole working device that is reliable and cost-effective in the real world. Those working outside the commercial world often have some truly ridiculous design constraints on their power supplies, many of them completely unrealistic.
The example you point out (switching supply with fast uncontrolled rise time driving a regulator with too much capacitance on its output) is one of those unrealistic examples. In the real world nobody would hang big capacitors that are likely to fail (probably because of their low ESR's) on the output of a regulator, and they would avoid putting a linear regulator on a switcher output with a HV output drop. But it happens way too often. Somebody somewhere should've pushed back and said how foolish it is, and it's their fault for not doing so!
Classic example of a design decision that is remarkably similar to the OP's question: transformerless 5-tube tabletop or clock radio. Turning it on makes a high-current surge through the cold filaments that wiggles the filaments. Every so often a filament will touch a cathode. If it tocuhes a cathode at the high end of the filament series-string, like the rectifier or output tube, the radio's PS's filter condenser is asked to filter 117VAC, and blows up, probably taking the failed tube, and maybe some outher tubes with it. This failure scenario is classic and common, but not common enough to have stopped billions of 5-tube transformerless radios being made over half the 20th century (most of which would still work today!)
On a sunny day (8 Nov 2006 03:38:32 -0800) it happened snipped-for-privacy@rediffmail.com wrote in :
There is also the reverse scenario, power off, where the output electrolytic is still charged while the input drops faster. IIRC long time ago I have seen an application note that took care of that with a reverse diode over the LM317. In the forward (first) case, you would need a say 30V high power zener (it would work as normal diode the other way) across the device. The zener would have to be able to handle the peak charge current. Maybe a transistor / zener combination.
On Wed, 08 Nov 2006 13:49:47 GMT, Jan Panteltje wrote in Msg.
This is a good idea for any linear regulator. I once blew out about 20 78L15's by acciedntally shorting out the (otherwise short-circuit-protected) +18V rail of a $50000 piece of lab equipment. Most of these were on a single letter-sized
6HU board densely populated with tons of discretes. The hardest thing was finding every single last one of them damn' things.
At first, of course, my heart sank when right after the short the room filled with burnt-epoxy stench and everybody looked at me...
It's gratifying to see so many replies to my question while I was out for the rest of the day. Too many to reply to one by one, I'm afraid.
Okay. Accepted that it's not adviseable to use the LM317 with an input that may exceed 40V, no matter what the input-output difference is. But for the sake of argument, what about NatSem's statement that the device can be used to regulate hundreds of volts as long as the input-output differential does not exceed 40V ? This would then assume that the load has 0 capacitance. Can such a load even exist in practice, without even a pF or so of stray capacitance ? If not, then that brings us back to the matter of time-limited surge and rise time.
The regulator is to be used with a transformer-rectifier input of 27 Vrms at normal line voltage, with load current ranging from a very few mA to about 0.6A. The transformer also has another load ranging from 0-3A. That's an input of nearly 37V to the regulator at light load. That leaves too little margin for mains fluctuations that may rise to more than +10%.
Circumstances permit the use of only the most widely available parts. I hadn't thought of the LM317HV. It's a good choice. I don't have it in stock and will have to send for it from another city - which is not as simple a matter in certain parts of the world as it is in some others. Otherwise I'll have to use another transformer for the 12V supply, or take a different approach to the rest of the design.
Well, I vote for the Zener across it - but one thing I haven't seen mentioned - what about a cap between the output and the ref. terminal, then a cap from ref. to ground? I'd be bypassing ref. anyway.
And a long rise time on the input sounds good too.
Why would you want to push your luck? LM317HV: 60 volt rating. If you want reliability, de-rate parts. Do not operate them at their maximum rating, and for voltage ratings, certainly not beyond the maximum even for a little while. OTOH, if you are putting in 45V and expect the output to run at, say, 30V, then you could add some protection (zener across the regulator; crowbar to kill the input if the output gets shorted...) and maybe be OK, at a cost well above just using the LM317HV or similar in the first place.
You can go for as much belt and suspenders as you want. I would. But with a small cap on the output and a big filter cap across the rectifier, your initial turn on should not cause grief - it is a non-issue.
As to the belt and suspenders, the zener has been recommended. In addition to that, is there any reason you can't stuff some R - say 33 ohms - in the input to drop the voltage a bit? It would help in the 317 dissipation, too. Without the R you're looking at ~ 15 watts. With it the 317 gets ~ 3 watts.
On Thu, 09 Nov 2006 09:59:01 +0800, budgie wrote in Msg.
This was worse because those people depended on that practically irreplaceable unit, and they had been not so enthusiastic about my sticking soldering irons into expensive things in the first place.
I was VERY VERY happy when I discovered that all that stench and smoke was only from a few cents' worth of TO92 jellybean parts. I immediately added a handful of protection diodes.
As a non-native user of English, I'm not familiar with the term "belt and suspenders". But the way you use it, it seems to mean things that are not strictly essential, but are added anyway to ensure satisfactory performance and/or extra security and safety. Correct ?
If that's correct, then I'm also a belt and suspenders man, but not in the way some designers like to keep on adding more and more stuff to compensate for a problem that could be solved more elegantly by revising a flawed approach.
Back to the topic. The load imposed on the transformer can be divided into two main groups, the one requiring a +12V supply being the lighter load. Instead of using separate rectifiers for the two sections, I'm thinking of using a common FW rectifier, heavily filtered, and insert a 1A diode in series with the lighter load. That section will have another, lighter, filter cap and the diode will act as a buffer against the effects of fluctuations caused by the heavier load.
The main filter will be 10000uF, and the secondary one 470 or 1000uF. The filter at the output of the regulator will be about 47uF Al electrolytic, paralelled by a .22uF ceramic disc. Other filters of the same order will be placed at entry points to various sub-sections. The total capacitance seen by the regulator output will be no more than 200uF. I think that will take care of the rise times.
The belt and suspenders would be to use the HV version, while the slow input rise time would allow using a plain
317 should the need ever arise.
You might have missed the part where I said that the regulator load varies from a few mA to about 0.6A. That pretty much rules out using a series resistor. A value high enough for a significant voltage drop at light load will be much too high when the load rises. One possible alternative would be to place a dozen or so diodes in series with the regulator input.
So the options are : multiple series diodes, a different transformer, switched-mode regulation, or a heavy heatsink. Comments please ?
Uh-oh. I was careless with my wording there. I considered the voltage differential issue already solved or clarified. That is, the heavy filtering at input and the resultant slow rise time would permit the use of an LM317, and the problem doesn't arise anyway if I use the HV version.
This is a one-off project, true, but under rather unusual circumstances. I live in a remote region where my clients bought an expensive piece of equipment made by a giant multinational company. In equivalent economic terms, it would be something like an investment of half a million dollars in the US. Unfortunately, it was heavily damaged by a power surge during a storm.
The company engineer (who flew in from a distant city) saw some of the work I'd done earlier for the same clients, and told them that it would be better for them in the long run to let me redesign and build every non-mechanical system from scratch. The power supply is only a tiny portion of the whole thing.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.