Reply by John Fields April 25, 20122012-04-25
On Sat, 21 Apr 2012 16:08:58 -0500, John Fields
<jfields@austininstruments.com> wrote:

>On Fri, 20 Apr 2012 13:36:49 -0700 (PDT), panfilero ><panfilero@gmail.com> wrote: > >>I need a 30V zener to drop the a high voltage of 100V down to 60V so I >>can use it with a switch mode power supply chip. So while this >>circuit is in use the zener will be exposed to current pulses from the >>switching, these current pulses are in the neighborhood of 1A. >. >. >. > >--- >What's the quiescent current through the Zener, how long do the pulses >last, and how often do they occur?
-- JF
Reply by Michael A. Terrell April 24, 20122012-04-24
"krw@att.bizzzzzzzzzzzz" wrote:
> > I proved > that Sprage devices would never fail (without avalanching first), so that was > the fix.
Sprague -- You can't have a sense of humor, if you have no sense.
Reply by Tim Williams April 23, 20122012-04-23
"mike" <spamme9@gmail.com> wrote in message 
news:jn0i5l$98l$1@dont-email.me...
> Go back a couple of weeks and review the thread on "slightly reverse > biased tantalums". Perfect example of allegedly rational engineers > arguing that it's OK to do something irrational.
Ah, but you see.. I would never insult the word "rational" by using it to describe most of the engineers here :) Tim -- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
Reply by krw...@att.bizzzzzzzzzzzz April 22, 20122012-04-22
On Sun, 22 Apr 2012 02:14:20 -0700, mike <spamme9@gmail.com> wrote:

>On 4/22/2012 12:50 AM, Tim Williams wrote: >> "mike" <spamme9@gmail.com> wrote in message >> news:jmvlsl$ood$1@dont-email.me... >>> In my opinion, there is NO pulse width that allows 40W of peak power >>> in a 1/2W device to be considered an acceptable design...none. >> >> Well, why not? >> >> Back in the toob days, horizontal sweep outputs were actually rated for >> this service. Typical datasheet reads 30W plate dissipation (abs. max., >> continuous service), 260W short circuit during startup only. This rating >> today would be labeled as something like "single pulse non-repetitive", >> with a maximum pulse width of a leisurely 60 seconds or so. >> >> Admittedly that's a peak-to-average ratio of not quite 10, rather than >> 80. Tubes are essentially limited by dumb joule heating until voltages >> cause vacuum breakdown or ion bombardment destroys the cathode; the peak >> could be much higher for shorter periods, but the cathode will saturate >> before you reach the actual pulsed thermal limit of the electrodes >> themselves. >> >> There are plenty of other examples in semiconductors, anyway. >> >> If your statement were unconditionally true, real switching circuits >> would essentially never work. A transistor (optimistically) rated for >> ~500W, will easily see, under hard inductive switching, power pulses >> over 20kW peak. That's a peak-to-average ratio of 40 right there. If you >> count against the actual continuous power rating, it could be 100W >> (ratio of 200) for typical heatsinking, all the way down to just 1-2W >> without any heatsink at all (for a ratio of a whopping 1000!). >> >> Clearly, there is SOME pulse width that allows a peak-to-average ratio >> over 80 to be considered an acceptable design, for some component types. >> >> Real engineers don't go by hearsay, they actually run experiments and >> think for themselves. >> >> Tim >> >So, you're saying that it's OK to use a device designed and specified >for such use??? >Couldn't agree more. > >Apparently, you've never got the panicked call from production, >"hey mikie, your product is exploding. The parts have the right part >number, but they're a different color. Incoming inspection says >they meet the spec. Lead time on the old parts is 20 weeks. >We've got enough of the old parts in rework to last 'till Wednesday. >FIX IT!!!??" > >Or the call from purchasing, "hey mikie, your idiot engineer wants >to add a 40W pulse power spec on a jellybean 500mW part. >The vendor just laughed at me. Whatdayawannado???" > >Or you've never had to explain safe operating area to an engineer >who wanted to use a transistor at max voltage and >current simultaneously.
SOA isn't about (not) using a device at max voltage and max current. The max power spec says you can't do that (unless it's an IRC part ;-). SOA is about transitioning between the two safely. ;-) One project I was sucked into turned out to be a SOA problem that the more senior engineer in the group designed in. Because of some completely avoidable design decisions, he put freewheeling diodes across contactors instead of across the transistors driving them, some 20' away. In one series of mainframes the wiring inbetween was just "black wire" (basically a wad of individual 24GA wires) which was quite inductive. Oops. The other two models used twinlead or trilead between the two points, which was quite capacitive. No problems with them. One brand of drivers (Motorola) would go through secondary breakdown and fry, causing the contactors to fail closed and the magic smoke to roll out of some ballast resistors. Not so good. I proved that Sprage devices would never fail (without avalanching first), so that was the fix. Not the best, but it worked.
>I have a lot of experience with what engineers do. >Product designers expect the product to sail through manufacturing >and be trouble free in the customer's hands. Sometimes, that >means overruling what "engineers" "think".
Don't understand the above paragraph.
>Go back a couple of weeks and review the thread on "slightly reverse >biased tantalums". Perfect example of allegedly rational engineers >arguing that it's OK to do something irrational.
>Abusing a part WAY, WAY beyond its specification when there >are conservative alternatives is bad business.
Now comes the argument about what constitutes "WAY beyond specifications". ;-)
Reply by Jeroen April 22, 20122012-04-22
On 2012-04-22 18:41, Fred Abse wrote:
> On Sun, 22 Apr 2012 02:14:20 -0700, mike wrote: > >> Or you've never had to explain safe operating area to an engineer >> who wanted to use a transistor at max voltage and >> current simultaneously. > > Apparently a usage of the word "engineer" with which I am unfamiliar. > > Where I come from, engineers are experienced graduates who design and > develop equipment that works, and is suitable for purpose. They know about > SOA, and where to find the relevant data.
I knew one, long ago. It was just around the time power MOSFETs became common, even though they were still expensive. He was testing some and he had a whole heap of devices that "didn't conform to specs". You'd think that after a pop or two he'd start to wonder if he'd misunderstood something, but not this one. Oh well, I thought it was funny. No idea what's become of him. Popping GaN transistors somewhere, probably. Jeroen Belleman
Reply by Spehro Pefhany April 22, 20122012-04-22
On Sun, 22 Apr 2012 09:41:15 -0700, the renowned Fred Abse
<excretatauris@invalid.invalid> wrote:

>On Sun, 22 Apr 2012 02:14:20 -0700, mike wrote: > >> Or you've never had to explain safe operating area to an engineer >> who wanted to use a transistor at max voltage and >> current simultaneously. > >Apparently a usage of the word "engineer" with which I am unfamiliar. > >Where I come from, engineers are experienced graduates who design and >develop equipment that works, and is suitable for purpose. They know about >SOA, and where to find the relevant data.
They probably didn't learn about SOA in University.
>"Product designers" are people who decide what sort of pretty box it's to >go in.
Best regards, Spehro Pefhany -- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
Reply by Fred Abse April 22, 20122012-04-22
On Sun, 22 Apr 2012 02:14:20 -0700, mike wrote:

> Or you've never had to explain safe operating area to an engineer > who wanted to use a transistor at max voltage and > current simultaneously.
Apparently a usage of the word "engineer" with which I am unfamiliar. Where I come from, engineers are experienced graduates who design and develop equipment that works, and is suitable for purpose. They know about SOA, and where to find the relevant data. "Product designers" are people who decide what sort of pretty box it's to go in. -- "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." (Richard Feynman)
Reply by mike April 22, 20122012-04-22
On 4/22/2012 12:50 AM, Tim Williams wrote:
> "mike" <spamme9@gmail.com> wrote in message > news:jmvlsl$ood$1@dont-email.me... >> In my opinion, there is NO pulse width that allows 40W of peak power >> in a 1/2W device to be considered an acceptable design...none. > > Well, why not? > > Back in the toob days, horizontal sweep outputs were actually rated for > this service. Typical datasheet reads 30W plate dissipation (abs. max., > continuous service), 260W short circuit during startup only. This rating > today would be labeled as something like "single pulse non-repetitive", > with a maximum pulse width of a leisurely 60 seconds or so. > > Admittedly that's a peak-to-average ratio of not quite 10, rather than > 80. Tubes are essentially limited by dumb joule heating until voltages > cause vacuum breakdown or ion bombardment destroys the cathode; the peak > could be much higher for shorter periods, but the cathode will saturate > before you reach the actual pulsed thermal limit of the electrodes > themselves. > > There are plenty of other examples in semiconductors, anyway. > > If your statement were unconditionally true, real switching circuits > would essentially never work. A transistor (optimistically) rated for > ~500W, will easily see, under hard inductive switching, power pulses > over 20kW peak. That's a peak-to-average ratio of 40 right there. If you > count against the actual continuous power rating, it could be 100W > (ratio of 200) for typical heatsinking, all the way down to just 1-2W > without any heatsink at all (for a ratio of a whopping 1000!). > > Clearly, there is SOME pulse width that allows a peak-to-average ratio > over 80 to be considered an acceptable design, for some component types. > > Real engineers don't go by hearsay, they actually run experiments and > think for themselves. > > Tim >
So, you're saying that it's OK to use a device designed and specified for such use??? Couldn't agree more. Apparently, you've never got the panicked call from production, "hey mikie, your product is exploding. The parts have the right part number, but they're a different color. Incoming inspection says they meet the spec. Lead time on the old parts is 20 weeks. We've got enough of the old parts in rework to last 'till Wednesday. FIX IT!!!??" Or the call from purchasing, "hey mikie, your idiot engineer wants to add a 40W pulse power spec on a jellybean 500mW part. The vendor just laughed at me. Whatdayawannado???" Or you've never had to explain safe operating area to an engineer who wanted to use a transistor at max voltage and current simultaneously. I have a lot of experience with what engineers do. Product designers expect the product to sail through manufacturing and be trouble free in the customer's hands. Sometimes, that means overruling what "engineers" "think". Go back a couple of weeks and review the thread on "slightly reverse biased tantalums". Perfect example of allegedly rational engineers arguing that it's OK to do something irrational. Abusing a part WAY, WAY beyond its specification when there are conservative alternatives is bad business.
Reply by Tim Williams April 22, 20122012-04-22
"mike" <spamme9@gmail.com> wrote in message 
news:jmvlsl$ood$1@dont-email.me...
> In my opinion, there is NO pulse width that allows 40W of peak power > in a 1/2W device to be considered an acceptable design...none.
Well, why not? Back in the toob days, horizontal sweep outputs were actually rated for this service. Typical datasheet reads 30W plate dissipation (abs. max., continuous service), 260W short circuit during startup only. This rating today would be labeled as something like "single pulse non-repetitive", with a maximum pulse width of a leisurely 60 seconds or so. Admittedly that's a peak-to-average ratio of not quite 10, rather than 80. Tubes are essentially limited by dumb joule heating until voltages cause vacuum breakdown or ion bombardment destroys the cathode; the peak could be much higher for shorter periods, but the cathode will saturate before you reach the actual pulsed thermal limit of the electrodes themselves. There are plenty of other examples in semiconductors, anyway. If your statement were unconditionally true, real switching circuits would essentially never work. A transistor (optimistically) rated for ~500W, will easily see, under hard inductive switching, power pulses over 20kW peak. That's a peak-to-average ratio of 40 right there. If you count against the actual continuous power rating, it could be 100W (ratio of 200) for typical heatsinking, all the way down to just 1-2W without any heatsink at all (for a ratio of a whopping 1000!). Clearly, there is SOME pulse width that allows a peak-to-average ratio over 80 to be considered an acceptable design, for some component types. Real engineers don't go by hearsay, they actually run experiments and think for themselves. Tim -- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
Reply by mike April 21, 20122012-04-21
On 4/21/2012 2:08 PM, John Fields wrote:
> On Fri, 20 Apr 2012 13:36:49 -0700 (PDT), panfilero > <panfilero@gmail.com> wrote: > >> I need a 30V zener to drop the a high voltage of 100V down to 60V so I >> can use it with a switch mode power supply chip. So while this >> circuit is in use the zener will be exposed to current pulses from the >> switching, these current pulses are in the neighborhood of 1A. > . > . > . > > --- > What's the quiescent current through the Zener, how long do the pulses > last, and how often do they occur? >
That's the problem with most questions asked here...lack of context. A person is certainly free to ask a specific question and get a specific answer. BUT You can often judge the experience (I was gonna say competence. That's the right word, but would ignite the ethernet and we don't want that.) of the person asking the question by the form and words used. When the big red flags go up, it can be helpful to the person to elaborate...whether they like it or not. The side effect is that we turn on each other, but that's just the way of the web. In my opinion, there is NO pulse width that allows 40W of peak power in a 1/2W device to be considered an acceptable design...none. And it's difficult to filter the current pulses. You start adding caps and discover that something else blows up when you crowbar the 100V supply. Are we comforted by the fact that most designers wouldn't consider that a problem? Without context, we're left with only common sense to guide us. Ignoring the 10V discrepancy in the question, someone stated that the zener could sustain 17mA if you didn't derate it. For many single chip switchers, the duty factor is closely related to the ratio of input and output voltages. Narrow 1A spikes seem out of place. I would prefer a design that didn't have 'em. Yes, it's possible for a system integrator to be in a situation where they have to mate two bad designs. That's why you need your BEST engineers doing the integration. One of the jobs of a project manager is to stamp out tunnel vision. Everybody needs to be aware of the big picture...not an expert, but aware. Communication is key. I used to tell engineers at the first project meeting that I was gonna personally test the design before the second prototype phase. The list of abuses would contain Crowbar the power supplies. Any connector that could be put on the wrong socket or backwards or off-by-one on the right socket would be tried and the power turned on. Bunch of other single-point failures that could happen in manufacturing. You could have heard the whimpering and wailing in the next county. Probably increased the cost a few pennies too. But the concept propagated to other parts of the design so engineers considered more than just what happened when everything was working right. Yes, you can easily come up with single-point failures that can't be protected against and will reduce the project to a smoldering lump. The key point is that MANY, MOST can be mitigated by paying attention. Enough rant... Back to the current situation. More context would be helpful. If we were voting based on what's been divulged, I'd bet that this is a disaster waiting to happen. I refuse to be comforted by the fact that most designs are faulty in ways that could have been easily avoided by paying attention. Excuses are cheap. Field failures aren't.