day shot to hell

I guess that depends on the application. I'm more worried abut fuse-blowers getting too happy than even attempting to save any downstream electronics. In my case, the large cost is the service outage, not the electronics.

I was protecting $1000's of dollars of custom-made stuff. I could make a 10A fuse element vanish is micro-seconds (filter-cap-bus-bar-SCR loop) ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    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     |

              Obama:  Master of Opaque Transparency

[snip]

I just designed a POR/Reference that does this sequence...

(1) Verify supply is at least 3 CMOS thresholds

(2) Verify Reference potential has reached regulation

(3) Verify VDD has reached minimum per specification

(4) Verify an additional input (monitoring switcher) is reporting OK

(4) Time-out for 100mS (on-chip oscillator and 14-stage counter ;-)

(5) Release RESETbar ...Jim Thompson

--
| James E.Thompson, CTO                            |    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     |

                Ability without honor is useless.
                                       
                                   - Marcus Tullius Cicero

The price may be (tens of) $1000s of dollars, but the cost certainly isn't. ;-) OTOH, a royally pissed off customer might easily be worth

40dB more. The boss doesn't like it when we're the subject of the television program.

...but transistors are free, packages are not.

If that were packaged by itself it would be only 4-pins (or 6-pin if you'd like to use the reference ;-)

In this case it's about 5% of the total chip function. ...Jim Thompson

--
| James E.Thompson, CTO                            |    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     |
             
The AMA represents only 19% of doctors, and only that high if you
count student members.  In Obama-talk that\'s considered consensus.

Would you do it at $1-$2 per supply?

and

If I made such a function as a custom part? Sure? ...Jim Thompson

--
| James E.Thompson, CTO                            |    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 you wanted a President with balls why didn\'t you elect Hillary?

and

You mean as part of the spec? Well, of course. I meant as an added expense just in case...

Let's put it that way: If the crowbar trips something else _has_ already failed. Meaning it's a service call no matter what. The difference between having a crowbar or not having one could be several thousand bucks (board repairable versus it being toastissimo). It can also prevent plumes of smoke and a rather embarrassing case where the fire engines had to come out because the automatic fire alarm and the sprinklers came on, damaging all sorts of other equipment.

In one of my recent cases the crowbar was put in there because the board could otherwise take out a $10k laser connected to it. Now that would really make for a pissed client. Not so much because of the $10k but because it would take more than a month of leadtime to get a new one.

It's like the airbag in a car. After it has opened you cannot continue to drive and it'll be an expensive repair bill but it may have saved a lot of grief.

--
Regards, Joerg

http://www.analogconsultants.com/

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

Right. The difference is the cost, not the price. The cost of the electronics is certainly not in the tens of thousands of dollars. The stuff I did where the cost was in the millions had all sorts of belts on. With supplies rated in the kA, things get smoky fast and smoke was rather more frowned upon. These weren't jellybean regulators, though.

Fuses are there to keep the heat, if not a little smoke, inside. Fans dissipate the rest rather quickly. ;-)

A false trip on the crowbar has the same effect for us. Show doesn't go on, customer mad as hell. He doesn't much care if it was $1 or $10000s of electronics smoking. It's not working.

If it went off when there was no accident, you still have at least second degree burns, broken arms and thumbs, and perhaps no ear drums. ...and haven't saved a damned thing. See, I can burn strawmen too. ;-)

and

With a bunch of hotshot FPGA you can easily get to $1k. Many of my boards drive very expensive gear.

See, that's why fuses are a good thing :-)

Mine are working. I've yet to see a false trip of one of my crowbar designs and it's been >>20 years now.

But they rarely go off without an accident. You need to exceed life-threatening G-forces to make the sensor signal an accident.

:-)

--
Regards, Joerg

http://www.analogconsultants.com/

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

Yep, Joerg, Sometimes _even_ _you_ can toss out strawmen and be aggravating ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    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     |

and

Sure, the Virtex-IIs and 4's I last used were about $4K each. Even here a service call at the wrong time (incoming shells) makes that chump change. That's not the point though. I don't spend a lot of money to protect bring-up (which was the original point). We'll put in 0-ohm resistors for debug but that's about it.

BTW, the FPGA I'm most likely to use next[*] is more like $2.

[*] Firmware is so hosed up there isn't much point in making new hardware to play with.

I never said I didn't use fuses, only that I don't have any illusions of them saving electronics. As I said before, their purpose is to keep any fire inside the covers.

Circuits that never fail? Why don't you just use one of them on the power supply, instead? ;-)

That's the theory. Reality is much different, of course.

We did a similar, but bidirectional, diode clamp arrangement on a PowerQUICC SBC. Everything worked fine in the lab and in production test, but the customer reported some cases of the SBCs not starting up. In our case the customer supplied a 5V rail, and we used TI TPS54610 switchers to provide 3.3V and 2.0V. We found that the customer 5V startup was much faster than our lab and production power supplies. One TPS54610 started quickly, pulling up the output of the other via one of these diodes, which confused it and prevented it from starting. We had to adjust the soft-start capacitor values to ensure that the switchers started in the correct order, so that they didn't get confused by their outputs being pulled up via the clamp diodes during startup.

================================

Greg Neff VP Engineering

*Microsym* Computers Inc. snipped-for-privacy@guesswhichwordgoeshere.com

OK, we fixed the power supply and replaced both BGA chips.

ftp://jjlarkin.lmi.net/First_Light.jpg

On the aluminum plate is the Kontron SBC, running Linux. It connects to our controller board through the short silver PCI Express jumper cable.

So far, it's powered up, things look reasonable, and we have Linux talking to the PCIe chip. Next step is to get the FPGA configured and doing stuff.

The NewHaven LCD looks real nice. The Kontron is talking to it RS-232, with a little AVR processor on the user interface board handling the LCD and buttons and LEDs.

John

Do all the testing you see fit. This case can be used to argue for "hard start" testing.

board=20

would=20

=20

one.

continue=20

a=20

drums.

Just the same, krw was being a twit.

Interesting power distribution used there.