PCB protection for 480VAC mains

diameter:

concern.http://www.youtube.com/watch?v=PHs14ZTo96M

lucky)

't_Worry,_Be_Happy

something like this:

?

-Lasse

Wow, that seems perfect. I did a general search and I've looked some = time=20 ago, but either it's fairly new or I just missed it.

And I'm getting an order together from Newark, so I'll add a few of = those=20 babies.

Thanks!

Paul

If they don't pan out you could ask Vishay if they'd endorse their 500V fusible resistors for 600V:

These guys have 1000V fusible resistors, scroll to the end:

--
Regards, Joerg

http://www.analogconsultants.com/

Welwyn also has some rated at 1000V. But Newark doesn't stock them:

Their selector shows them as "level 69V" and the EMC2 as "level 11.7V". = I=20 don't know what that means.

Actually 500V is probably enough. 600V is more of a device class, but = some=20 fuses like the KTK are 600V, and the same size fuses in slow blow, FNQ, = are=20 rated 480V.

480 VAC service is usually 277 to ground, for wye connections, but it = could=20 be delta, and a phase could be as high as 480 VAC to ground. Our new = test=20 sets are not rated for 600V input, but older ones are. And they = typically=20 have a tap switch that has a maximum setting of 600 VAC, even on 208V = input.

Mostly I worry about a fault from line to ground, or something like the =

120=20 VAC control voltage. In that case, depending on phase, a 480V supply = could=20 be 600V to ground in some cases. There is no way to protect against all=20 possibilities. There are also two gate/cathode pairs that come onto the=20 board. Maybe I need to fuse those as well.

But my customer does not believe in anything more than the most basic=20 protection. The earlier test sets had something like ten fuses rated =

30-60A=20 600V, to protect a set of relays, and they have been removed. We also = have a=20 voltage relay which does not allow the test set mains to be energized = unless=20 the power inputs are properly configured for 208/240 or 480 VAC. But we = had=20 some problems with the VRLY that disabled the test set and it had to be=20 jumpered out until it could be repaired, so the technician could use it. = So=20 now my customer sees any protective device as a potential liability, and =

says he would rather have a test set heavily damaged by an incorrect=20 connection, than have it falsely prevent the technician from using it.

Management vs Engineering politics. Dilbert, anyone?

Paul=20

f90d61f488ee.PDF

The voltage rating of the fuse is not all you need to consider with high power circuits, you also need to consider the AIC (Amps Interrupting Current) rating for fuses or circuit breakers used in high power circuits where the available fault current exceeds 10 kA, which is usually the case in industrial power systems.

The current fuse in a Fluke DMM is only rated for 10 kAIC, for instance. I heard the bang (at least as loud as a shotgun, caused hearing damage to the two closest people) and saw the remains of the meter when someone accidentally contacted a 480 volt circuit with about 18 kA available fault current with the meter set for current measurement. The fuse failed to clear, and the probes substituted as fuses, with the probe tips evaporating well back into the plastic probe handle. We opened up the meter and found most of the traces completely evaporated and most leads blown out of IC packages.

The guy with the meter got lucky, the probes were far enough apart that the arcs did not join for a direct line to line arc, which could have caused a potentially fatal arc blast.

The easy way to check the upper limit on available fault current is to divide the current rating of the transformer supplying power by it's impedance factor (both will be on the nameplate). Typical impedance factors are around .05 (or lower) (sometimes listed as a percentage), so the available fault current is limited to about 20 times (or more) the nameplate current. The actual fault current will be lower, usually not much, due to other factors which quickly become rather tedious to calculate and are not normally bothered with unless the simple upper limit is just over a standard AIC rating.

Specifying parts not rated for the available fault current has been known to cause injury and death, I suggest you not do it.

Regards, Glen Walpert

crossing

DF

high

circuits

instance.

have

You raise some excellent points, and I am familiar with the concepts of=20 available fault currents, interrupting rating, and impedance factors = (also=20 sometimes known as regulation, in %). Since you brought this up, it will = be=20 helpful to discuss the details of these test sets and their safety = issues.

First, there is no way of knowing what sort of mains supply it will be=20 connected to, in the field, so we must assume the worst, which would be=20 probably a 480 VAC supply with a 400A or even 1200A capacity, and it is = very=20 possible that the source may be of even higher kVA. We once took our = largest=20 test set to an electrical generating station and connected the input as=20 solidly as possible to their largest service, and we were able to obtain =

more than 100kA on the output into a shorting bar, with an open circuit=20 voltage of about 12 VAC. This was a test set rated about 6000A = continuous,=20 so a rough estimate of the impedance factor (or regulation) was about =

5%, as=20 you said, with the output about 20x nominal. The current draw on the = mains=20 would be about 1/40 that of the secondary, or about 2500 amps.

The early versions of these test sets were protected with only a MCB = rated=20 at 200 to 400 amps, with interrupting rating of about 12,000 amps. Our = newer=20 test sets have a pair of LPS-200 fuses which are rated at 200kAIC, and = that=20 should be sufficient for most normal services which are protected by = current=20 limiting fuses. When an extreme overload occurs, such fuses trip in less =

than 1/2 cycle, even before the first peak, so they limit the actual=20 "let-through" current to something that a MCB can handle, and in fact = will=20 probably stop the current before the breaker even has time to operate. I =

have seen a movie comparing the performance of fuses and MCBs with = various=20 available fault currents, and even at the rated IC, the breaker was = badly=20 damaged, but it did its job of clearing the fault. At much higher fault=20 currents, however, the breaker exploded in a quite spectacular manner, = and=20 an arc was maintained until the source was turned off. Under the same=20 circumstances, fuses with ACIR of 100k or 200k only exhibited slight=20 movement and perhaps a puff of magic smoke.

My colleague who has me design these test sets was once performing=20 switchgear maintenance with an older engineer, during a shutdown of a=20 commercial facility after hours. The engineer was using an adjustable=20 wrench, which did not have a properly insulated handle, to loosen some = bolts=20 on live buswork, and the wrench slipped, falling across the main bus = from=20 the distribution transformer, and it created a bolted short and a huge=20 fireball. My friend had just enough time to turn aside, and was still = badly=20 burned, but the engineer received injuries that were eventually fatal. = They=20 were both wearing safety glasses, but no other protective gear. I found = one=20 video clip that really demonstrates the need for arc blast suits, and = shows=20 what can be survived:

The company I originally worked for did breaker testing, and one of my = first=20 jobs was at a large government facility in Newark, OH. We removed, = tested,=20 serviced, and replaced hundreds of metal clad breakers, with no = problems.=20 But a few years later one of our technicians was seriously injured when = a=20 stab finger cluster came loose during racking and fell across an = energized=20 bus. So I have great respect for what can happen when high power = electricity=20 is involved. And it's technically still "low voltage" since it's les = than=20

600V.

Back to the protective measures of the test sets we make, the LPS-200 = fuses=20

df=20 limit the maximum available fault current to 20kA where available fault=20 current is 100kA. So for all practical purposes, the fusing or other=20 protection beyond that can have an ACIR of 20kA or even 10k, which is in = the=20 range of most circuit breakers and fuses used for instrumentation. The=20 control circuitry is then fused with something like FNQ-5 fuses which = limit=20 fault currents to less than 100 amps, and a true interrupting rating of=20

10kA.

But the real problem is the connection of the gate wires to the SCRs, = which=20 are capable of several thousand amperes during normal use. The SCR = trigger=20 boards we have used in the past have split bobbin transformers for the = gate=20 circuitry, and opto-isolators rated at 5kV breakdown. We have not had = any=20 problems with these boards. The phase reference has always come from the =

120=20 VAC control circuitry, and that is safely limited and protected. Much of = the=20 safety of these boards is built into their physical design, with large=20 clearances between the AC mains and the low voltage control circuitry. = But=20 this new design will be in a smaller package, and will optionally derive = the=20 zero crossing detection from the voltage across the SCRs, so that a 12 = VDC=20 supply can power the board circuitry.

There are also other safeguards built into the newer test sets. After = the=20 primary fuses, we have a pair of large contactors which are energized = from=20 the AC control voltage, through a series of interlocks. These interlocks = can=20 be thermal sensors, safety switches, or circuitry which monitors output=20 current for overload conditions. For this we have designed a = Programmable=20 Overload Device (POD), which calculated the safe operating overloads and =

shuts down as needed. We have also considered a "Ground Overload Device" =

(GOD), to detect ground faults.

But, as my colleague asserts, having too many safeguards can also be a=20 liability, if the circuitry itself fails and renders the test set = unusable.=20 And also there is the premise that a technician may even get careless if = he=20 thinks there are all sorts of protective safety devices that will = prevent=20 damage or personal injury if the test set is misused.

So, there needs to be a tradeoff and limitation on what can happen to = cause=20 a dangerous condition, and what can be done to avoid it in the first = place,=20 or deal with it safely and effectively if it does happen. Permanent=20 electrical installations are covered by many standard practices and the = NEC=20 code, while portable test equipment is covered somewhat by NEMA = standards,=20 but because of its nature is not as straightforward. In the case at = hand, I=20 am trying to look at as many ways as possible to determine what could = happen=20 that could affect safety, and what can be done, reasonably, to prevent=20 problems or deal with any that could happen. I think the surface mounted =

500V fuses are a reasonable precaution, and I may also include them in = the=20 gate connections. But in reality, the board should just be designed with =

adequate clearances and perhaps some insulating barriers to preclude any =

problems.

The most likely source of failure is the optoisolators, since they are = very=20 small and there could be some deterioration or manufacturing defect that =

could cause a breakdown. The only way to absolutely be sure is to use=20 separate transmitters and receivers separated by a substantial length of =

light pipe. The other source of breakdown would be the transformers, and =

properly designed split bobbins are very reliable. Other than that, the = only=20 failure mechanism would be external sources of conductive material, such = as=20 loose strands of wire or contamination from carbon or metallic deposits, = and=20 those can be prevented by proper mounting and perhaps a protective = cover.

Sorry for the long essay, but I wanted to present these points for my = own=20 analysis as well as invite other comments. Thanks.

Paul

*Sigh*. Another of those charming outfits where they think that the fuses are there to protect _the equipment_. A couple of minutes trawling on YouTube will get you some useful videos taken by security cameras at substations and other places where they had an arc event when somebody was changing a breaker, or something like that.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058

hobbs at electrooptical dot net
http://electrooptical.net

f90d61f488ee.PDF

b245-16e25e01cbd4.pdf

Thanks for the long essay, I think anything which increases awareness of the hazards of high power electrical distribution is a good thing. Good video link too. Clearly you have given this matter the serious consideration it deserves. The only other thought I have is that it may be desirable to design the enclosure so that if an internal arc blast were to occur it would vent out the back, away from the operator, either with adequate open vent area or with the back panel being designed to blow open without detaching. I expect you have already considered this too.

I was fortunate in that no one was ever hurt or killed on any of the jobs I worked on involving high power electrical systems (mostly military facility design and startup testing, ending in 1998). But people I worked with lost 4 coworkers on other jobs; 1 low voltage electrocution,

2 from medium voltage burns and 1 arc blast death from the shock wave. All due to human error rather than equipment failure. The arc blast death occurred when an electrician opened an air-cooled breaker scheduled for replacement, went to get the drawing rack, came back with it and pulled the wrong breaker, which was under load. The scariest indecent occurred at the New London CT submarine base, where an electrician went to inspect a new enclosed pad mount substation installation, where the installers neither grounded the enclosure, bolted it down. A truck then backed into the enclosure pushing it into contact with a medium voltage terminal. The electrician grabbed the lock to open it, and suffered burns requiring amputation of an arm and a leg, plus such severe neurological damage that he was still unable to talk a year after the incident. Prior to that, I never once considered the possibility that the enclosure of distribution equipment might be hot, and touched numerous enclosures with my bare hand without concern.

I seems that many still have a cavalier attitude towards electrical safety, as evidenced by an article in the November 2011 Control Engineering magazine titled "Electrical Controls Dirty Little Secret: We Don't Follow NFPA Rules", which basically mocks NFPA 70e safety rules. Pity the magazine was foolish enough to publish it.

Regards, Glen

I

n,

al

y

ere

PC

t n

The

by

?v=3DPHs14ZTo96M

orry,_Be_Happy

a bit of 'fiddling' but a series cap rated per those pesky EMC specs for direct AC mains connections?

y y r t

I still remember watching the guy working on a 'small' 200A 480V service access having something go off sounded like a shotgun blast and either he jumped, or was knocked, a good ten feet backwards into a sitting position on the floor, more startled than injured.

Well, if anything goes wrong (incorrect wiring, faulty switches or components, etc.) there may not be enough of the unit left to even do a port-mortem! A serious fault on 240 V can be messy, but even a minor fault at 480 V can lead to massive arcing and enormous flash damage. I stay away from 480 as much as possible.

Jon

From a liability point of view they should be rated for the maximum applied peak voltage. If something bad happens that'll be the first thing the expert witness will be looking at, the datasheet. 480V is nothing to sneeze at, it can cause major facial injury, loss of eyesight and so on.

Aha! That would be the fire extinguisher on the wall, I assume :-)

Show them this :-)

--
Regards, Joerg

http://www.analogconsultants.com/

Yes, I've seen that. I recommend that we supply or require the same=20 protective gear that is used in this video:

My colleague has witnessed first hand what a 480 VAC arc flash can do. = He=20 was badly burned and his associate had his face in the cubicle where he=20 dropped his uninsulated wrench, and received severe injuries that proved =

fatal. There have been many unsafe conditions where I work for him, as a =

consultant. We have a source of 480 VAC at 400A, and a sub-panel fused = at=20

30A for testing units that we are unsure of. And even a portable unit = that=20 boosts 120 VAC to as much as 600 VAC but only capable of about 2-3 amps=20 before kicking out the breaker. And also some 208 VAC sources that are=20 nominally 20A.

But I had hooked up a test set to one of these sources, which had three=20 wires coming out of the box to connect to the test set. IIRC it had two = red=20 wires and a black wire, so I assumed the reds were two phases and the = black=20 wire was ground. But actually one of the red wires was ground, and=20 identified only by having a lug rather than a clip. So the test set only = had=20

120 VAC supplied to it, and the chassis was at 120 VAC to ground.=20 Fortunately nobody was injured, and I informed the management about this =

problem, and eventually they used a properly identified green wire for=20 ground.

Sometimes the customer does not connect the ground. I found this once on = a=20 job in Alabama where I was trying to calibrate a 50,000 amp test set and = my=20 test equipment was acting strangely. And I also was getting tingles from = the=20 cabinet. So I pulled it out from the wall and found that it was not=20 grounded, and apparently never had been, as they had to run another = wire. We=20 tried to design a ground integrity sensor, but it did not work on all = line=20 configurations. Later test sets use an isolated green binding post for a =

separate connection to an external ground, and a low voltage current is=20 applied that must flow through this connection back to the main = equipment=20 ground on the test set. But usually the customer just jumpers the two=20 together and uses one ground wire.

We have also attempted to design a GFCI circuit, but it is just about=20 impossible to detect 20 mA when the source can draw surges of over 2000 = amps=20 in normal use!

Paul=20

Most of our earlier units were basically a 14 gauge steel box, with a = bottom=20 frame on casters, and expanded metal to keep mice out of the interior. = So=20 any major arc flash would expel hot metal and flames out the bottom.=20 Hopefully the operator is not wearing sandals! But one of our new units = is=20 built with aluminum T-slot channels, and the side panels are merely 1/4" = ABS=20 plastic. At least the top is a steel plate. You can see a video of these =

breaker test sets here:

and others along the same line:

(4000A DC nominal, tested up to 20 kA) (a USB based analyzer I designed for=20 calibration)

And yes, I should be wearing safety glasses and other protective gear!

[snip]

Perhaps it was an example of what NOT to do.

Paul=20

crossing

the

control

high

about

that

to

usually

of

be

is

about

amps.

a

and

job

and

removed,

no

"low

fault

in

SCR

into

detection

the

have

safe

a

misused.

a

any

defect

of

from

=20

=20

=20

=20

jobs=20

=20

scheduled=20

=20

=20

Scary indeed. For the past few years my job has expanded to include arc flash hazard issues and i have my own copy of IEEE 1584 and some other related stuff.

?-)

t

ed

ck

had

a

my

the

We

e

mps

Paul,

*IF* you're serious about GFCI circuitry to detect that 20mA out of 2000A+, we should talk - I think I see a way to do it.

welcome to come back directly Robert .DOT. A .DOT. Macy .AT. gmail .DOT. com

I have to ask what- the f*ck is it with electricians dropping uninsulated wrenches all the time? Nobody was hurt, but a giant power outage at a datacenter I used was caused by guess it -a electrician dropping an uninsulated wrench into some 480V switch gear. It's like clockwork that this happens.

tip for electricians or anybody working on live ciruits with dangerous amounts of energy present- use insulated tools.

I'm not sure of the sensitivity, and I'm guessing it's more than 20mA but in mining they have weird GFCI breakers which operate by injecting an extra signal into the power and then looking for this signal in the ground. Looking for plain current imbalances just doesn't work for this stuff. I have no idea how they detect an electrified puddle with cables in it over plain leakage from the 800Hz (something like that) that they inject into the power.

On Sat, 17 Dec 2011 17:58:25 -0500, P E Schoen wrote:

Interesting videos. Some safety gear is no doubt appropriate, perhaps safety glasses, face shield and hearing protectors. My feeling is that it is a good idea to read the current safety standards and conform to their requirements completely, mostly because those standards are very much experience based and proved effective but also so that if there is an injury it won't be due to negligence on my part. None of the current standards were out when I last worked with high power distribution equipment, so I can only guess what is now required for the fairly low energies in your testers. I like to go a bit beyond in some cases and duct tape rubber mats over nearby conductive material if it is for some reason necessary to do hot work - in addition to using properly insulated tools.

The 4000 amp frame breaker in your video is much smaller than one with the same rating I recall well, which was a 200 kAICR ITE-Siemens switchboard breaker designed for remote operation. Final customer acceptance test underway, and the breaker will not operate except manually - remove a cover from the breaker, insert a large crank, turn about 20 times to compress the opening spring and then close the breaker. We racked out the breaker and found that the 120 VDC control power rectifier was open (the supply consisted of taps from the 480 input terminals to a big cartridge fuse (no wimpy glass fuses in switchboards) to a ~500 VA transformer to a single house numbered stud diode, no unreliable capacitor of course, and not required for a trip). I called the factory, gave them the serial number, and was told that it was still under warranty so they could not sell me any parts or provide any information, I should pack up the 1600 pound breaker in it's original crate and send it back to the factory for repair, they would turn it around in 120 days max. I explained that I had 20 people on site waiting to continue a plant acceptance test and that wouldn't do, so they said if we bought a new one they could probably get it to us in only 60 days. I told them we would consider our options, then went to the nearest Radio Shack and bought a diode suitably rated to protect the expensive fuse (just like the original), resuming testing in about 2 hours. Memorable only because it was the least useful call to a vendor I ever made.

If you can't be an example, be a warning :-).

Glen