Power supply protection networks

s a

he

=A0 =A0 =A0 =A0 =A0 =A0 =A0 ... The TL431 is nice

in

I there a chip that drives a P-fet pass device to perform overvoltage protection?

"Phil Allison" "Phil Hobbs"

There is the idiot customer that plugs in the wrong wall wart.

** And YOU are an IDIOT poster who does not even bother to read the words he criticises.

.... Phil

That's quite easy, compared to repairing equipment that survived a house fire. Especially when it was from a friend's home.

--
You can\'t have a sense of humor, if you have no sense!

Then you're ready to meet dimbulb. The smell will remind you of that course.

--
You can\'t have a sense of humor, if you have no sense!

Like a radio shack protoboard in a control system at a nuclear power plant?

--
You can\'t have a sense of humor, if you have no sense!

[...]

They are usually under 50c but in qties. ON Semi has quite favorable pricing, also when it comes to POR/BOR chips.

[...]

I still use vellum as well, on occasion. I guess the next generation won't even know what vellum is.

Didn't see it was a depletion mode device. But at 500mA this one is very close to being tapped out. One over-current situation on your load and it might develop a spontaneous crater.

FETs aren't good cap multipliers, that would be a BJT domain. And yes, if it was my circuit I'd do the vanilla way. Either a nice fat P-channel or a chip.

--
Regards, Joerg

http://www.analogconsultants.com/

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

[snip]

Wall warts are getting out-of-hand. I have so many now that I've taken to labeling each new arrival with a silver-colored Sharpie. Even with that I have a couple of orphans that I have no clue as to what they go with :-(

...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

| 1962 | Obama... Recklessness Cloaked in Righteousness

[...]

Not that I know of, but maybe. I never use chips when I have to employ an external pass device anyhow since that gets too expensive. LMV431 plus BJT usually does the trick. Of course this can be different in very high cost countries like in Europe where there are painful SMT placement costs. But a chip that costs several bucks is hardly warranted even there.

--
Regards, Joerg

http://www.analogconsultants.com/

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

[...]

Or the polyfuse smokes out. I am not a great believer in those things except for protecting control lines that can't realistically be protected any other way.

Why are so many companies still do kitting? It never made sense to me.

I had a hard time convincing companies to quit doing that. Even when I ran a division and was da boss this proposition was met with great skepticism. Then I lucked out, found and hired a production manager who believed in Kanban and now there was more "convincing power". After we ditched kitting, costs dropped quite noticeably and the skeptics became less skeptic ...

--
Regards, Joerg

http://www.analogconsultants.com/

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

The surface-mount polyfuses have low voltage ratings and they are serious... a few volts over rating and they evaporate. The bigger leaded parts are better.

The trip current of a surfmount polyfuse is also very layout dependant; pcb traces/pours/vias dominate heat transfer to ambient.

2:1 at least.

How can you stuff a board without pulling the parts?

We have two semi-auto p-n-p machines with automated carousels that present the correct parts to the operator. Each carousel bin has to be filled with the correct parts. That's "kitting" to us. A limited number of reel feeders are also available, for bypass caps and such. For those, we can just pull the reel from stock, use it, and maybe put it back.

If we send boards out to a contractor to stuff, again we have to furnish the correct parts in about the correct quantities, in labeled bagggies. Kits again.

John

I have three treatments to the anonymous wallwart problem: first, the 'unknown' ones go into a duffel bag under the spare bed... Second, the multipin types (mini DIN-5 for USB hard disks) get a quick analysis, pinout into my little 'pinouts.txt' file, and (if I can locate the silver paint marker) a label. Third, for my own use, I try to use 24 VAC wallwarts, and put DC/DC converters and V-doubler rectifiers at the point-of-load. There's not much in the way of wrong connection that can threaten this kind of system; it's either the right power, or not enough to start up the converter.

The duffel bag is getting a bit heavy, though.

Hey, I resemble that comment!

Blew up an MP3 player once, around a decade ago (back when they were still kinda spendy), by plugging in the wrong plug. The MP3 player was sitting along lots of other electronics, and I think it was a 36V? printer's wall wart that had the same connector as the MP3's charger. Zaappp.... phffft!

Ok, that's not the typical meaning of kitting. The typical scenario that I found is this:

Production wants to build 40 machines of type XYZ. So they start pulling

40 transformers, 40 fuse assemblies, 160 casters, and so on. When there are only 38 transformers in the stock room a mid-size problem situation arises because the kit order can't be completed. A gigantic waste of time and inventory Dollars. What we changed that to is this:

Whenever a production guy feels his floor stock for certain parts is uncomfortably low he or she is at liberty to get as many more from the stock room as desired. The next step in our change process was that many parts were no longer stored in the stock room, they went right from loading dock to floor. To my surprise a lot of higher level folks felt a bit uncomfortable with this idea. Until they saw that it worked.

The labeled baggies became history, too. That's how this company used to work but we changed that as well. We contracted out board fab and the whole chebang of parts went to that contract house. No kitting, everything went. Jelly-bean parts such as 0.1uF caps came from their own stock, were billed to us, and generally they just left those bulk feeders on the machines when changing from our job to some other client's.

--
Regards, Joerg

http://www.analogconsultants.com/

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

Phil Hobbs a écrit :

1nF vs bond wire?

OK, some figures:

Some poking around suggest 1 mil gold wire for SOT23... Let say it's 1mm long and that this wire takes all the energy.

Gold specific heat is 0.129 J/(g K) (sorry no BTU per whatever ounce Fahrenheit ... :-) Gold volumic mass is 19.3 g/cm^3

A 1 mil diameter and 1mm long wire is 5E-7 cm^3 which is 9.8ug

This gives a temperature rise rate of 15.5E6 K/J Ouch...

Well, let say your cap is 10nF, charged at 10V. That's 0.5uJ and 7.7K temp rise.

At 25V it's a 48K rise and at 40V (3904/06 max Vceo) it's a 123K rise.

You'll hardly vaporise it for a single event. More this assumes worst case, that is all the energy is dissipated into the bond wire which is plain wrong: A 1 mil 1mm long gold wire is 43mR if I got it right, so that's a 0.43ns time constant with the 10nF cap.

Noway a 3904/06 could switch that fast, not even considering the 20K base resistance, so you're totally safe...

--
Thanks,
Fred.

[...]

10nF is probably somewhat ok. But it's not the bond wire. A typical case I encountered at a client was switching diodes. They got a wee current spike, so short that it barely showed up on the scope. So everyone thought that this would never cause an issue. Well, this board developed a history of failing gradually after some time of operation and it was always those diodes. They didn't die but degraded. So they called me in. Fired up SPICE, showed the problem, people scoffed. "Nah, that's impossible, that's ridiculous, can't be, look how short it is". Change was implemented anyhow, just in case. No more field failures.

This morning I got a call from a client for whom I fixed a similar problem about 18 months ago. Asked him for the number of field failures related to the old problem since then. Zilch.

[...]

--
Regards, Joerg

http://www.analogconsultants.com/

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

We don't have "floor stock" except a carousel thing full of nuts/bolts/washers, which are still technically in the stockroom.

The stock room is open and unguarded and is along one wall of the production area. Anybody can grab a part if they need it (and fill out a line on the signout form of course.) But each production batch has a work order that says to build N assemblies, and we pull a kit for that batch, and account for the parts used. We know at all times how many parts are in each stock bin, fairly accurately, and we know what we need to buy for anticipated production needs.

I can also ask my computer to show me the status of any part, where it is used, what it costs, and usually see the datasheets. And I can examine and price any BOM.

I have worked for companies that lived in Kit Hell, where the stockroom floor was covered with partially-complete kits that had shortages, and parts were constantly being stolen from some incomplete kits to try to finish other incomplete kits.

We can't do that. We manufacture hundreds of different assemblies in various batch sizes from 1 to maybe 100, some in-house and some outside, and some of the parts are very expensive.

John

But Phil wrote the thing about bond wire, so I just put some figures on this...

Sure this happens. Do you have some more info, diode reference or its kind, spike characteristics,...

--
Thanks,
Fred.

[...]

That's what we did away with, pulling work order kits. The computer knew from the numbers received and the WIP systems going through the works how much was on hand, it didn't really matter whether the part was already installed or not.

Same with us, except that there was no datasheet link in the beginning. That came when we introduced Agile.

Same with us. Some AD converters cost as much as a very fancy meal for two with French champagne. But we had all boards made at the same assembly house. If you do in-house in parallel that can get old though.

Another thing that helps a lot is if engineers try to use "company standard" parts as much as reasonably possible, not always resorting to their own favorites. That gave purchasing more leverage.

--
Regards, Joerg

http://www.analogconsultants.com/

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

[...]

Ah, ok. yes, it takes a lot of punch to blow a bond wire.

Not anymore, this was a while ago. AFAIK it was the BAS21 or a very similar diode and it was operated as a T/R switch, two of them per transceiver. The spike was really miniscule, nsec range and back then DSOs didn't have that kind of sample rate. Several amps. Since it was irregular and low duty cycle, neither analog scopes nor sampling heads helped.

--
Regards, Joerg

http://www.analogconsultants.com/

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