MOV voltage

It has to be rated for maximum continuous line plus small transients. "130VAC" is the most commonly seen value for 120V.

Expect the maximum clamp voltage to be 2-3 times nominal under full surge conditions.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

OK, thanks. I'll order some 130V immediately. They have onboard fuse, but have to bring the $5,000 equipment offline and get the 50 cents fuse in a week.

On Saturday, March 12, 2016 at 11:06:00 AM UTC-5, snipped-for-privacy@gmail.com wrot e:

A varistor must be sized for a worst case surge current. Varistor data she ets provide charts to size it. An undersized protector is why fires happen .

Proper design also means a thermal fuse. To disconnect protector parts as fast as possible during a larger surge (while leaving that surge still conn ected to electronics). Only useful answers include those many numbers and other well understood design precautions. Start by reading varistor data s heets. Recommendations must always be based in those spec numbers.

ote:

heets provide charts to size it. An undersized protector is why fires happ en.

s fast as possible during a larger surge (while leaving that surge still co nnected to electronics). Only useful answers include those many numbers an d other well understood design precautions. Start by reading varistor data sheets. Recommendations must always be based in those spec numbers.

ok, will include a thermal fuse. Should I also include a regular fast acti ng fuse? Thermal fuse might not be accurate for current.

130V has to be an AC voltage rating. MOVs will conduct on the peak voltage of AC.

There is a MCOV (maximum continuous operating voltage) rating, which is the voltage at which a MOV starts to conduct (something like 1ma). A failed (but still functional) MOV is defined as one whose MCOV has decreased by 10%. The peak AC voltage a MOV will see had better be less than that.

Francois Martzloff was a surge protection expert at the US-NIST. He has written: "The fact of the matter is that nowadays, most electronic appliances have an inherent immunity level of at least 600 V to 800 V, so that the clamping voltages of 330 V widely offered by [surge protector] manufacturers are really not necessary. Objective assessment of the situation leads to the conclusion that the 330 V clamping level, promoted by a few manufacturers, was encouraged by the promulgation of UL Std 1449, showing that voltage as the lowest in a series of possible clamping voltages for 120 V circuits. Thus was created the downward auction of 'lower is better' notwithstanding the objections raised by several researchers and well-informed manufacturers. One of the consequences of this downward auction can be premature ageing of [surge protectors] that are called upon to carry surge currents as the result of relatively low transient voltages that would not put equipment in jeopardy." (330V is not the "AC" rating, but is more likely the MCOV.)

Note that the potential of the device power wires can shift, which can cause problems if other external wiring is connected to the device.

In the US, UL1449 has required a thermal disconnect for MOVs in plug-in protectors since 1998. These disconnects are in close (thermal) proximity to the MOV.

Martzloff looked at the amount of energy that could make it to a MOV on a branch circuit. The source was up to the maximum surge with any reasonable probability of occurring on power service wires (10,000 A), and branch circuits 10m and longer. The maximum energy dissipated was 35 Joules. In 13 of 15 cases it was 1 Joule or less. There are a couple reasons the energy is so low, one of which is specific to US wiring products and practice.

The energy rating of a MOV is the single-event energy that will put the MOV at defined end-of-life. Looking at data sheets, if the energy hits are far smaller than that, the cumulative energy rating is far larger.

(MOVs do not protect by absorbing surge energy, but they absorb some energy in the process of protecting.)

Fuses are not fast enough to disconnect during a surge - disconnect happens after. Failing MOVs start to conduct at normal voltages, which is when they are likely to overheat.

The protected equipment can be connected downstream from the thermal disconnect. In case of MOV failure, the protected equipment will not be exposed. (I believe plug-in protectors listed under UL1449 are now required to say if they do NOT disconnect the protected equipment.)

Francois Martzloff has written "In fact, the major cause of [surge protector] failures is a temporary overvoltage, rather than an unusually large surge." An example of TOV is a several kV distribution wire that falls on the wire to your house. The event is a far longer duration than a "surge".

Thanks you for the quote of the nice writeup.

Unfortunately, the specific equipments we have is of marginal design. It is only active for a few seconds, but most of the components are under-rated. Namely, will fail under continuous operations. So, typical surge protectors might not work.

They are working fine in high spec core power area; namely, downtown SF; but failing in area outside.

Very informative comments

The comments from Martzloff, is that from a book?

Cheers

Klaus

ting fuse? Thermal fuse might not be accurate for current.

Fast acting fuse is too slow. Destructive surges are done in microseconds. Fastest fuses take milliseconds. That same defective design is why so ma ny (maybe millions) of APC protectors had to be removed immediately - a ser ious human safety threat. Concern is to disconnect MOVs that fail catastro phically - that become too hot and on the verge of creating a fire before t hat fast acting fuse can respond.

Return to relevant numbers. All properly designed 120 volt electronics (be fore PCs existed) would withstand 600 volts without damage. Today's electro nics are even more robust. A potentially destructive surge that might over whelm existing internal protection is maybe once every seven years. Even l ess often in venues such as San Francisco.

Apparently you are having repeat failures. Known should be which type of s urge exists from critically important facts such as each damaged internal p art. There is no one type surge. Which type of current is causing damage? To have that answer means you know what part is damaged, what was the inc oming path, and what was the outgoing current path. Only then can effectiv e protection be recommended.

Adjacent MOVs address one type of surge. Which of many is causing damage? An effective solution is not possible if that anomaly is not first defined. Best evidence is always found in a dead body - an apparently repeatedly d amaged item. What was a destructive incoming and outgoing electrical curre nt path?

Even if the mains protection device will pass peaks of thousands of Amperes, you should know the current path in the electric system. The peak goes in many cases indirectly or directly into a grounding electrode, with has some grounding resistance.

The current peak through the grounding resistance will cause a ground potential rise at the equipment. If the device has separate distant ground connections e.g. through telephone, ADSL or CATV network connection, there can be huge voltage differences between the local and remote grounds.

Thanks to the mains protecting devices, the mains supply may survive, but components close to the external signal connections may be fried. Most people then blame that overvoltages came through the external connection, while the real culprit is the chassis potential rise due to the house grounding electrode grounding resistance (and impedance).

acting fuse? Thermal fuse might not be accurate for current.

s. Fastest fuses take milliseconds. That same defective design is why so many (maybe millions) of APC protectors had to be removed immediately - a s erious human safety threat. Concern is to disconnect MOVs that fail catast rophically - that become too hot and on the verge of creating a fire before that fast acting fuse can respond.

The fast acting fuse protects the equipment from over current. Thermal fus e and MOVs protect it from over voltage.

The failed equipments are in San Mateo, 30 miles south of SF. One of them was replaired for $1000. The techician simply replaced a pair of boards. It failed again. Preliminary analysis of the board showed on-board fuse de ployed. The on-board fuse is behind the recifier and caps (2000uF), namely :

120AC -> Bridge recitifer -> 340DC -> fuse

The $1 fuse would need to be replaced by at least $100 labor cost, even if we stock additional boards. So, we are thinking:

1. Increase 1A on-board fuse to 1.5A (can't find anything in-between)
2. Add 1A and thermal fuses in series to line input
3. Add 120V, 130V and 150V MOVs in parallel
4. Place fuses and MOVs in external plug-in module
5. Replace 120V MOV monthly, 130V MOV every 3 months and 150V MOV every 6 m onths

My worry is that the initial cap charging would be more than 1A, if we plac e it outside the box. Should I short out the fuse with a relay during powe r up?

In the case of sub-par equipment I see only one option: A serious UPS. One that doesn't just pipe line power through and switch over but a UPS that always runs, always supplies the connected equipment with well regulated 120V made right there by its inverter regardless of what comes in.

MOVs are way too inexact. If you dimension them too tight you can get spectacular failures with lots of smoke. They are like a bank account, they can take this many Joules in hits and then ... phssst ... PHUFF ...

*BANG* ... The thermal protection mentioned may be a good intention but I have seen my share of devices where MOVs have burnt out themselves and some stuff around them. A visit from OSHA or the fire marshal after an event is going to cost more than the $100 in labor.

Big fat TVS may be better than MOV but they also have a large spread between rated stand-off voltage and the voltage where they actually come on hard.

--
Regards, Joerg 

http://www.analogconsultants.com/

Many of Martzloff's original articles can be obtained by searching the NIST website.

RL

Yes, it is one option under consideration.

That's why we want to put it outside the box, perhaps with metal enclosure.

From what you wrote so far that appears to me your only option. Also the most sensible one because you can easily buy such gear at a cost that is most likely much lower than the labor and materials expenses you have sunk into it so far.

It would likely have to be much more intelligent than MOV or even TVS. For example, what I do on DC rails a lot is add ballasts that are normally off but come on when a voltage gets exceeded. Controlled by precison references such as TL431 so you can have a very small windows. But I am not aware of any such module off-the-shelf and all your custom design efforts would easily dwarf the expense for a line-conditioning UPS. Even a line conditioner without a backup battery might work if you do not need to tide over power interruptions although with PCs in the mix that would be a smart thing to have.

--
Regards, Joerg 

http://www.analogconsultants.com/

In response to klaus kragelund:

Martzloff sources were all papers written by Martzloff. He has created an anthology consisting mostly of what he has written:

SPD is surge protective device - a surge protector SRE is a surge reference equalizer - a surge protector with added ports for cable, telephone ... to wire through Low-voltage is less than 600V.

from

Remembering better - UL1449 tests surge protectors by hitting them with a series of test surges. The protector must survive these tests and remain functional. The largest voltage during a surge at a plug-in protector is the "maximum let-through voltage". This voltage is rounded up to one of the standard let-through voltages in UL1449. 330V is the lowest standard voltage.

As Tim wrote, the let-through voltage is much higher than the nominal MOV clamp voltage.

from

Also with slightly different results:

10,000A maximum surge current - his comes from another Martzloff paper - I couldn't find it in a rapid search. It is based on a 100,000A lightning strike to an adjacent utility pole in typical urban overhead distribution. This is an extremely close strike, and only 5% of strikes are stronger. It the worst case with any reasonable probability of occurring. About 10,000A is on each power service wire to a house (H-H-N).

This comes from an IEEE guide to surges and surge protection (primarily for homes, but principles are the same):

This surge guide is aimed at technical people (anyone here) and IMHO it is well worth reading.

from

Also is good on surges generated by utility power line switching, which are generally the most damaging surges after lightning.

Adding some details to what you have written: Suppose you have a (US) power system earthed by only a ground rod, and the rod has a near miraculous resistance to earth of 10 ohms (and ignore the impedance of the wire from the rod to the electric service). Then suppose you have a strong surge on power wires that results in 1000A current to earth. The power system ground will be 10,000V above 'absolute' earth potential. (As a rule of thumb, 70% of the voltage drop away from the rod is in the first 3 feet.) In the US, the service neutral is bonded to the system ground. If panel busbars are about

6,000V from the enclosure there will be arc-over. The established arc will be hundreds of volts. (This dumps most of the surge energy to earth.) Power wiring may be less of a problem than signal wiring. Much of the signal wiring coming into the house(cable TV, satelite TV, telephone,....) can be many thousands of volts from the elevated building ground. That voltage can show up at anything connected to ground and power and signal.

Much of surge damage may be caused by high voltage between signal wires and power/ground.

In the US, it is standard practice for incoming phone wires to go through an entry protector that is tied to the power earthing system. The connecting wire must be short for protection. (I believe in the UK it is common for telephone entry protectors to not be tied to the earthing system.)

Similarly (US) coax entering the building (antenna, cable, dish) must go through a ground block and the shield is connected to the earthing system. Again this must be a short wire for good protection. (An example of a wire that is too long is in the IEEE surge guide, link elsewhere, starting page 30.) This actually leaves the coax center conductor unprotected except for arc-over at connectors, which may be 4kV.

Wires may rise thousands of volts from 'absolute' earth potential. Much of surge protection is that all wires rise together.

Martzloff has written "the impedance of the grounding system to `true earth' is far less important than the integrity of the bonding of the various parts of the grounding system."

UPS is too bulky and we don't need battery backup. Probably just a 12V plug feeding an AC inverter.

Put some inductance or (wiring)resistance between the MOVs to protect the lowest voltage MOV.

Hey, that's VERY ODD. Fuses for AC interruption are cheap; fuses that can usefully interrupt a 340VDC circuit are expensive.

Or consider something more decisive than an MOV, namely a Sidactor (SCR crowbar) for overvoltage, and have a resettable circuit breaker. Maybe a polyfuse?