Big crowbar module for <4VDC?

Indeed.

Thank you! Finally someone understands!

My approach had the storage cap, the fuse, and the SCR in a _very_ tight loop, with wide traces. I have even been known to solder bus bar on top of this path to ensure that high currents are reached quickly ;-)

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
 I love to cook with wine     Sometimes I even put it in the food

A lot of people have learnt a lot of very useful things from AoE, and the world is significantly better off for its existence.

I do agree that the fuse and crowbar SCR should preferably go before the regulator, especially if the transformer is not protected by a fuse in the secondary that will always blow before the transformer overheats (which is difficult unless the transformer is grossly oversized), and/or a thermal cutout in the windings.

It can be even more difficult to choose the rating of the fuse in the primary of the transformer such that it will always blow when a steady overload appears at the regulator output, yet does not blow in normal circumstances due to inrush current.

Unless there is a thermal cutout in the transformer itself, it may be possible in some cases to set fire to the transformer without first blowing the primary fuse. That could damage the expensive load by purely thermal means (flames, etc.), even if no electrical damage is caused.

It is probably worthwhile working out exactly how the current is going to be cut off after the crowbar fires. Putting the fuse and crowbar switching device before the regulator (obviously with the voltage sensing connection after the regulator) sounds like one good way whereby the likelihood of blowing the fuse cleanly could be predicted, with a suitably large capacitor and SCR. Other arrangements might require more analysis and/or testing.

Chris

A transformer that overheats before the primary fuse goes is a serious design flaw. Either the thermoswitch must open or the fuse.

What else would protect upon overload caused be a faulty electrolytic? Unless you pepper the whole unit with fuses.

Just received another schematic, for da big mother of all linear supplies. The one that needs modding. The SCR is, tada, you guessed it, on the output. Rock-solid American manufacturer, in business since 1973. Good products, top notch customer service (I had the schematic within the hour).

The SCR is a serious edition, size of a silver Dollar.

If you do the super high current spike thing (I wouldn't, but you know that by now) make sure they never put in a plain old glass fuse. Exploding glass is no fun.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Most of my designs require a thermoswitch to be embedded in the transformer. Usually self-reset types are not allowed. IOW once such a transformer has overheated it's toast. But the building didn't burn down.

Once the power went and I heard a loud boom. Then screams. Looked out the window and the slope across the valley was ablaze. Hot flames licking up the hill, and fast. Luckily the fire department was there in minutes. A major utility transformer had decided to go lalaland.

The primary fuse, usually. Sometimes the current sense resistor which just opens. However, it must be designed and rated for such an event. Those are the minor things often overlooked by younger designers. Back in Europe they used to have versions with a spring-loaded solder joint but those could be outlawed by now. The buck grazing on the hilltop on the other side of town might get lead poisoning from the miniscule plume ...

20mm or 1-1/4" fuses can also be had with sand-fill. But you know how it goes. A nuisance trip, no spare, quick dash to Home Depot. Only glass fuses to be had and, voila, it works again and everybody forgets about it. Until the day Jim's crowbar decides it's time for da big one.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

[sniped Joerg's smell of burning transformer :-]

Wait for Obama's "green team" :-(

[snip]

Funny. In mucho testing I've never fractured the glass. That would only happen if enough time elapsed to produce heating... like the way you do it ;-)

(My method hardly disturbs the voltage on the cap ;-)

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
 I love to cook with wine     Sometimes I even put it in the food

Yes, but it is not at all trivial to get it to work without the thermoswitch. If I were to hang my crowbar on someone else's PSU, then I guess I might feel duty bound to know the likely and less likely outcomes of the crowbar being activated, in the presence of whatever fault inside the PSU was the original cause of the overvoltage (so maybe the current limit doesn't work right anymore, etc.). Perhaps that is doing the PSU designer's job for them.

I agree that the transformer should ideally be protected from fire in the case of a long term short circuit appearing at its output, as well as any current between normal and the short circuit current. The worst case is probably not a short circuit (if this blows the fuse) but rather whatever current nearly but not quite blows the fuse (and there is some manufacturing variation in fuses too, even the non-user-replaceable ones). I am wary of the possibility that the original design did not necessarily achieve this degree of protection. I am not sure what is legally required but I have seen some commercial designs that worry me. Admittedly it is mostly smaller transformers than the one you are looking at, where I worry that the primary current with a nearly-shorted secondary is not quite enough to blow a primary fuse. In things like wall warts I am much more reassured when there is a thermal cutout embedded in the primary winding as well. Maybe I'm being paranoid and the flame retardants in the modern insulation are possibly better than I have assumed. I have had an old waxed-paper transformer catch fire in some audio equipment (without blowing the fuse), fortunately whilst I was there in the room to unplug it, but that experience did make me look at transformers with more suspicion.

How do they cut off the current after the SCR fires? Is it the primary fuse, or maybe the pass transistor emitter bondwire? I guess they tested it at some stage.

The fuses that they use in mains plugs in the UK are surprisingly robust. I'm not sure what the official rating is but they are made of ceramic not glass and they are filled with sand. They are used in such large numbers that the price is very very low.

Chris

I don't think a TUEV medical device inspector would let that fly.

Blowing fuses fast and with extreme current is dangerous. You need a special high energy fuse and preferably sand-filled:

The other reason why I do not like fuses on DC circuits is arcing. Given a sufficient voltage the fuse could just let St.Elmo dance for a while. Phssssssssssssssss ... dazzling blue light, a whiff of ozone, then a stench ...

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Didn't we start this discussion protecting a 3.3V linear regulator?

You keep changing the game to protect your argument ;-)

Besides, I should have pointed out, I use 220V fuses for this scheme.

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
 I love to cook with wine     Sometimes I even put it in the food

No, just wanted to point out another risk for folks who adopt this design strategy and then use it on a 48V telco supply.

I still wouldn't do it, unless it's sand-filled fuses. Anyhow, solution is found. Got schematic, will use existing SCR and install a nice TL431-based trigger when at that client next time. Can be done without disassembly.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

You can put the crowbar SCR across the input to the regulator and the over-voltage detector across the output - a kind of feedback. I suspect that is how Jim designs his.

On a sunny day (20 Dec 2008 10:06:18 GMT) it happened Jasen Betts wrote in :

Yes you can do that, and he even showed a small kid diagram with a 7805 IIRC. But that will _not_ protect what happens on the supply line. Jimmy has the typical rightist republican thinking: His method protects his own design, not the load :-)

Yes. Blows fuse and it leaves no charred remains like certain other suggestions must do ;-)

Linear regulator can be tested for failure mechanism with no worry that crowbar destroyed evidence.

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
 I love to cook with wine     Sometimes I even put it in the food

OUTPUT.

Jerk, Learn to read. AND: I DIDN'T post any diagram. AND: It DOES PROTECT the load... some people here are just TOO STUPID to understand simple solutions.

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

                  Why are Europeons so ignorant?
           They think they know it all about the U.S.A.
                 But never have bothered to visit

On a sunny day (Sat, 20 Dec 2008 08:31:28 -0700) it happened Jim Thompson wrote in :

OUTPUT.

Yes, you.

OUTPUT.

I suppose it would not protect the load against externally- or load- generated overvoltages.

--

John Devereux

OUTPUT.

The "load" generated its own over-voltage ?:-)

This thread started as protecting the load from failure of a simple

3.3V linear regulator.

Exotic systems would require case-by-case analysis... for example...

After the Challenger disaster a complete system-by-system, component-by-component analysis was ordered to determine any possible fault mechanisms.

It was yours truly (consulting at Sperry Space, later Honeywell Space) who found the power supply redundancy fault... one supply down, they all go down.

It needed a fix that _wasn't_ a complete redesign (requiring all kinds of compliance/qualification testing).

So I came up with a fix using HexFETs... all we had to qualify was the HexFET itself (they had no prior "S" rating).

So I have the honor of getting the first HexFET into space ;-)

==

Simple crow-barring of a power supply output, without any well-defined relief/release mechanism seems down-right stupid to me. Citing that everyone is doing it doesn't make it good... it only means it's "cheap" ;-)

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

            If I\'m talking, you should be taking notes.

[...]

Sure, I was thinking of something like a motor controller, or perhaps JL's NMR machine.

Sounds like the "oring controllers" some of the semi manufacturers seem to be peddling.

No argument here.

--

John Devereux

On a sunny day (Sat, 20 Dec 2008 15:49:59 +0000) it happened John Devereux wrote in :

This is exactly one of the important points. I know at least about one case where a guy connected the 24 V PLC power to the 5 V logic. If you have a crowbar at the output, passively measuring the output (thyristor + zener), then you have a better chance of keeping the replacement cost low. In this case they had to order new boards for a huge amount of money.

There are many ways, with a probability much higher then a series regulator failing, how a high voltage can land on a low voltage line, circuit failures in other electronics that use both supplies could happen too.

The 'evidence' case the republican Tommy refers to, does not hold if for example 24 V is applied to the output of a 3.3V regulator with crowbar before the regulator, as it may destroy the regulator anyways.

Anyways, just short the output, there may even be fold back still working, I have seen drifting supplies too, bit high voltage, crowbar active. The old Philips K9 chassis had a crowbar in the series switcher 160V (IIRC) output, to protect the horizontal output and every other thing in the set getting supply from that. It has a primary fuse too, but those switch mode would just go tick tick tick if the transistors were not kaput. Protect the load.

[...]

The relief/release mechanism _has_ been shown in this thread. There is:

a. Primary fuse

b. Current sense resistor

c. Transformer thermofuse

Of course, all (not just one alone) need to be calculated, tested and qualified to do a graceful job, meaning no loud bang or serious plume of smoke or whatever else the client requires. The goal must be that only the primary fuse goes as that is panel accessible by field service folks. They often aren't allowed or equipped to open unit panels and dive into the circuit board level.

Nope. If all the major manufacturers do it one should at least consider it and find out why they do it that way. I know for a fact that they also cater hi-rel markets.

Just one more reason although I am sure your opinion is cast in concrete here: If a sense line is shorted or one of the supply lines comes loose a bit for some reason the supply will think it needs to shovel some more coals ... crowbar comes. Both your crowbar and mine will tackle that. But in my case the field service guy can just re-affix the wires and swap a panel fuse. In your case the power supply box must be opened. This is mostly not desired and field personnel is often not allowed to do that -> PSU swap -> send back -> $$$

Aside from the fact that high-amperage fuses in DC paths are very frowned upon by agency compliance guys and by me (arcing).

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.