Reply by July 7, 20162016-07-07
On Thu, 7 Jul 2016 13:05:09 +1000, Sylvia Else
<sylvia@not.at.this.address> wrote:

>I'm looking at protecting a power sense circuit from mains power >transients, using a resistor and a zener diode. > >Spice says it works well, but when the zener conducts, essentially the >entire transient voltage appears across the resistor. A high voltage >transient spike is likely to be very narrow, but I can find nothing to >say I can put, say, 5000V, across a 1K resistor, even for a microsecond. >Indeed, it seems entirely possible that a current would flow across the >outside of it, rather defeating the point of having the resistor there. > >Any thoughts? >
Automotive circuits don't have to protect against anything like 5KV (more like 50-100V) but they often use Zener (TVS) diodes with no resistor. Basically the resistor flattens out the transfer curve so the output voltage goes higher than it would without the resistor. Often a pi circuit is also used to filter the transient (though that's not its only purpose).
Reply by Tim Williams July 7, 20162016-07-07
"Jon Elson" <jmelson@wustl.edu> wrote in message 
news:o-OdnSgT3dnKRuPKnZ2dnUU7-QudnZ2d@giganews.com...
> Well, I can tell you what happens to a 10 Ohm, 1 W axial film resistor > when > the line gets a lightning strike. The resistor goes POP and all the paint > and much of the resistor element disappears. Sometimes I have to replace > the gas tube arrestor, too, after these events, but the equipment being > protected has survived several of them. The resistors are in series > before > the arrestor. >
Heh. A very similar, though smaller scale, application I've seen: in high resolution CRTs (e.g., Trinitrons), the wideband video amplifier (50-100V at >100MHz) usually had protection diodes on its output, then series resistors, then GDTs, then carbon comp resistors, before finally connecting to the CRT. The event being: - Internal arcover from 2nd anode (30kV and ~1nF, not your grandfather's ESD!) - Zaps the grid and/or cathode - Carbon comp increases impedance - GDT ignites, shunting pulse (but not after having let through a few hundred volts still) - Second level resistor, and clamp diodes, handle the rest. All together, probably still not all that healthy for the vid amp (which is made with ~1GHz, 100mA BJTs, usually monolithic), but that's all they can afford without killing bandwidth! Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
Reply by Tim Williams July 7, 20162016-07-07
<dagmargoodboat@yahoo.com> wrote in message 
news:66aeb029-2996-4f1f-af0b-7296957b77ed@googlegroups.com...
>> There must be a reason nobody suggests an inductor in seies. How come? > > Inductors can be useful, such as to reject a fast transient by low-pass > filter action (L-C). It has to be able to handle the voltage impressed > across it, which isn't always easy. Turns and cores arc!
That, and flux. If you've never played around with an EFT test machine (or racked your brain over a product while in the test lab), then you'll have to realize the pulses are quite tall indeed. 50ns doesn't sound like much, but it's not just a couple volts of RF, it /just keeps coming/. It's an electromagnetic tsunami. You can literally observe it washing over cables and modules, and reflecting off end nodes. The biggest problem is the high impedance. Normally, such a test is performed at a modest distance above a ground plane, so that the cables have a common-mode transmission line behavior much like microstrip, with an impedance on the order of 150 ohms. Even if you put a huge ferrite toroid in series with that, with many turns, all you can do is take the edge off -- a huge hump of flux still makes its way through. Indeed, if you try shorting the pulse across such a choke, you'll probably saturate it. A 2kV pulse 50ns long contains 100uWb of flux*. That'll saturate a single turn around ferrite (Bmax = 0.3T let's say) of 333 mm^2, which is a sizable pot core, or a 75-100mm toroid, or many turns around the average ferrite clamp. *Not really, but ballpark close. The real figure is probably less than a full rectangular 2kV * 50ns. But it's more than half, because after 50ns, the pulse is only half amplitude, and still drooling on. As for *surges* (~us domain), that's practically DC. Those can be filtered by some iron-cored equipment (which again, largely does the job by saturating from the excess flux), but usually must be dealt with by clamping or absorption -- hence the popularity of MOVs. Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
Reply by July 7, 20162016-07-07
On Thursday, July 7, 2016 at 12:40:28 PM UTC-4, Tom Del Rosso wrote:
> Tim Williams wrote: > > "Tim Wescott" <tim@seemywebsite.com> wrote in message > > news:Xp6dnfzBzolXVODKnZ2dnUU7-UmdnZ2d@giganews.com... > >> Some resistors come with pulse ratings. There's a power/time > >> relationship that's not just as simple as dumping a constant energy > >> pulse of any width into the thing, so you want to specify a resistor > >> that comes with real live charts on the data sheet. > > > > Right. If you see one that gives a single pulse rating, you can > > probably-kinda-sorta-hand-wavey-extrapolate around it. > > > > The slope is usually a goofy function of time, like t^(-1/2) (which > > implies diffusion, a classic thermal effect), or t^(-1/3) (..?!). > > Rarely, it is inversely proportional, even to fairly small time > > scales; this is mostly confined to the "high pulse" (bulk / > > composition) types. > > Naturally: if the datasheet doesn't say explicitly so, you're on your > > own. A guess is just a guess... until you start getting field > > failures back (or not, as the case might be!). > > There must be a reason nobody suggests an inductor in seies. How come?
They're the lowest loss option, and it would be more of bead than an inductor= predominantly resistive dissipation of the high frequency components of the transient. Epcos has detailed design notes centered on their product line.
Reply by July 7, 20162016-07-07
On Wednesday, July 6, 2016 at 11:05:17 PM UTC-4, Sylvia Else wrote:
> I'm looking at protecting a power sense circuit from mains power > transients, using a resistor and a zener diode. > > Spice says it works well, but when the zener conducts, essentially the > entire transient voltage appears across the resistor. A high voltage > transient spike is likely to be very narrow, but I can find nothing to > say I can put, say, 5000V, across a 1K resistor, even for a microsecond. > Indeed, it seems entirely possible that a current would flow across the > outside of it, rather defeating the point of having the resistor there. > > Any thoughts?
You really should use a resistor specifically designed to handle surges. Ohmite has several lines depending on the type of surge, one line even for lightning surges. Most of them are designed around IEC 64000 transient profiles, so you need to look at that and determine which one best fits your circuit's risk exposure.
> > Sylvia.
Reply by Jon Elson July 7, 20162016-07-07
John Larkin wrote:


> There are some telecom-type resistors that come in low values and are > specified to withstand lightning surges, usually 5KV peak. > >
Well, I can tell you what happens to a 10 Ohm, 1 W axial film resistor when the line gets a lightning strike. The resistor goes POP and all the paint and much of the resistor element disappears. Sometimes I have to replace the gas tube arrestor, too, after these events, but the equipment being protected has survived several of them. The resistors are in series before the arrestor. Jon
Reply by July 7, 20162016-07-07
On Thursday, July 7, 2016 at 12:40:28 PM UTC-4, Tom Del Rosso wrote:
> Tim Williams wrote: > > "Tim Wescott" <tim@seemywebsite.com> wrote in message > > news:Xp6dnfzBzolXVODKnZ2dnUU7-UmdnZ2d@giganews.com... > >> Some resistors come with pulse ratings. There's a power/time > >> relationship that's not just as simple as dumping a constant energy > >> pulse of any width into the thing, so you want to specify a resistor > >> that comes with real live charts on the data sheet. > > > > Right. If you see one that gives a single pulse rating, you can > > probably-kinda-sorta-hand-wavey-extrapolate around it. > > > > The slope is usually a goofy function of time, like t^(-1/2) (which > > implies diffusion, a classic thermal effect), or t^(-1/3) (..?!). > > Rarely, it is inversely proportional, even to fairly small time > > scales; this is mostly confined to the "high pulse" (bulk / > > composition) types. > > Naturally: if the datasheet doesn't say explicitly so, you're on your > > own. A guess is just a guess... until you start getting field > > failures back (or not, as the case might be!). > > There must be a reason nobody suggests an inductor in seies. How come?
Inductors can be useful, such as to reject a fast transient by low-pass filter action (L-C). It has to be able to handle the voltage impressed across it, which isn't always easy. Turns and cores arc! Cheers, James Arthur
Reply by Tom Del Rosso July 7, 20162016-07-07
Tim Williams wrote:
> "Tim Wescott" <tim@seemywebsite.com> wrote in message > news:Xp6dnfzBzolXVODKnZ2dnUU7-UmdnZ2d@giganews.com... >> Some resistors come with pulse ratings. There's a power/time >> relationship that's not just as simple as dumping a constant energy >> pulse of any width into the thing, so you want to specify a resistor >> that comes with real live charts on the data sheet. > > Right. If you see one that gives a single pulse rating, you can > probably-kinda-sorta-hand-wavey-extrapolate around it. > > The slope is usually a goofy function of time, like t^(-1/2) (which > implies diffusion, a classic thermal effect), or t^(-1/3) (..?!). > Rarely, it is inversely proportional, even to fairly small time > scales; this is mostly confined to the "high pulse" (bulk / > composition) types. > Naturally: if the datasheet doesn't say explicitly so, you're on your > own. A guess is just a guess... until you start getting field > failures back (or not, as the case might be!).
There must be a reason nobody suggests an inductor in seies. How come?
Reply by John Larkin July 7, 20162016-07-07
On Thu, 07 Jul 2016 11:26:31 -0400, Ecnerwal
<MyNameForward@ReplaceWithMyVices.Com.invalid> wrote:

>In article <du5v77F102vU1@mid.individual.net>, > Sylvia Else <sylvia@not.at.this.address> wrote: > >> I'm looking at protecting a power sense circuit from mains power >> transients, using a resistor and a zener diode. >> >> Spice says it works well, but when the zener conducts, essentially the >> entire transient voltage appears across the resistor. A high voltage >> transient spike is likely to be very narrow, but I can find nothing to >> say I can put, say, 5000V, across a 1K resistor, even for a microsecond. >> Indeed, it seems entirely possible that a current would flow across the >> outside of it, rather defeating the point of having the resistor there. >> >> Any thoughts? >> >> Sylvia. > >Kinda depends - if you are Jeorg-pricing a solution it's rather >different than if not. > >My default whole-house (single-phase 120/240 VAC) mains-input regimen at >this point is Delta LA 302R (SOV) and CA 302R (surge capacitor - also >facility EMI filter). It's not magic, but it sure does seem to reduce >issues - of course, that could be confirmation bias, since it's all been >negative results so far. And it's around $100 with shopping, more >without shopping. Squish it where it comes in, protect everything. You >probably have something less broad and far cheaper in mind ;-) > >High voltage lab methods were sometimes to spend for the high voltage / >high power resistors, often to string a bunch of normal axial-lead >resistors in series and stuff them in a plastic tube (tube does not help >power dissipation, does help not making inadvertent contact with the >high voltage junctions.) Can do the same on a PCB if you are careful >that the layout is not self-defeating. > >Vishay (just first place that came up looking for datasheets) offers a >line of chip resistors they market as "pulse-proof" but you're still >looking at a peak voltage of 500V on a 2512, or 200V on 1206. They also >have medium voltage chips at 1415V in 2512, or 550V in 1206, and >high-voltage but with very high resistance as a minimum (so you'd be >hard pressed to get 1K from any reasonable number of those.) > >Of course, you could also put a high-voltage capacitor in parallel with >your resistor (or simply across the line) to take the edge off the >spike. Mouser claims to have a 5KV AC/DC rated 220pF for just over a >buck in onesies, 75 cents at 100. Still too rich for Jeorg, of course.
Right, the 1K value adds difficulties. Most HV resistors come in correspondingly high ohmic values. 5KV across 1K is 25 kilowatts. A 1 us pulse dumps 25 millijoules, which should be OK into a few 1206 or 2512 size resistors in series. I guess the issue is demonstrated survival vs regulatory approval. There are some telecom-type resistors that come in low values and are specified to withstand lightning surges, usually 5KV peak. -- John Larkin Highland Technology, Inc lunatic fringe electronics
Reply by Ecnerwal July 7, 20162016-07-07
In article <du5v77F102vU1@mid.individual.net>,
 Sylvia Else <sylvia@not.at.this.address> wrote:

> I'm looking at protecting a power sense circuit from mains power > transients, using a resistor and a zener diode. > > Spice says it works well, but when the zener conducts, essentially the > entire transient voltage appears across the resistor. A high voltage > transient spike is likely to be very narrow, but I can find nothing to > say I can put, say, 5000V, across a 1K resistor, even for a microsecond. > Indeed, it seems entirely possible that a current would flow across the > outside of it, rather defeating the point of having the resistor there. > > Any thoughts? > > Sylvia.
Kinda depends - if you are Jeorg-pricing a solution it's rather different than if not. My default whole-house (single-phase 120/240 VAC) mains-input regimen at this point is Delta LA 302R (SOV) and CA 302R (surge capacitor - also facility EMI filter). It's not magic, but it sure does seem to reduce issues - of course, that could be confirmation bias, since it's all been negative results so far. And it's around $100 with shopping, more without shopping. Squish it where it comes in, protect everything. You probably have something less broad and far cheaper in mind ;-) High voltage lab methods were sometimes to spend for the high voltage / high power resistors, often to string a bunch of normal axial-lead resistors in series and stuff them in a plastic tube (tube does not help power dissipation, does help not making inadvertent contact with the high voltage junctions.) Can do the same on a PCB if you are careful that the layout is not self-defeating. Vishay (just first place that came up looking for datasheets) offers a line of chip resistors they market as "pulse-proof" but you're still looking at a peak voltage of 500V on a 2512, or 200V on 1206. They also have medium voltage chips at 1415V in 2512, or 550V in 1206, and high-voltage but with very high resistance as a minimum (so you'd be hard pressed to get 1K from any reasonable number of those.) Of course, you could also put a high-voltage capacitor in parallel with your resistor (or simply across the line) to take the edge off the spike. Mouser claims to have a 5KV AC/DC rated 220pF for just over a buck in onesies, 75 cents at 100. Still too rich for Jeorg, of course. -- Cats, coffee, chocolate...vices to live by Please don't feed the trolls. Killfile and ignore them so they will go away.