# varistor calculations needed

• posted

Hi to newsgroup

I have a problem that needs some assistance. I have a 220V 50 HZ industrial relay at 20 amps. The arming inductor is rated at 6 watts . I need to use a varistor as surge arrester but i don't know to calculate the energy that will be produced and the varistor ratings . As i understand it must be rated for

275 VAC rms.But how is the current capability related and the energy (joules) that it is rated .Is there a rule of thumb for such calculations ?. Any help would be apprecciated .
• posted

Joules are Watt seconds. Watts are current x voltage and i^2 R. You know R from the varistor on properties. You need to work out the peak current and it's decay time constant so that the total power dissipated by the varistor is within limits (typically 6 joules for pulses

• posted

VA is the critical figure, from that the peak energy storage of the inductor can be computed (multiply by half an AC cycle and sqrt(2))

can you take a current reading with a clamp meter?

• posted

I presume the surge arrester is for the coil, which (when switched off) will possibly cause some switch damage. The 20A rating of the contacts is irrelevant unless the load circuit also needs some surge protection.

The energy stored in an inductor is 0.5 * L * I **2, and the varistor will absorb that much energy (less, actually) each time the relay is cycled. The "6W" rating of the coil suggests AC current in the 30 mA range, so peak current would be 50 mA. Normal AC drive reverses the current

50 times a second, so it can't take more than 10 ms to quench: If your surge protector clamps at 400V, that makes a worst-case

E =3D 400 * .05 * .010 =3D 0.2 joule

Detailed measurement of the coil inductance and equivalent resistance (eddy current losses in the iron are important) can get you a real number. The "6W" might only be related to heat losses, not necessarily inclusive of the reactive current.

• posted

In my country mains is rated at 220V rms That is 220*1.414=311 volts peak. That means regarding votage overshoots 275 vrms varistor is safe to use ???. I've measured current thats how i calculated power so i imagine a 6 joule varistor is ok . My other question is i see on digikey 275 v rated varistors characterized as

175 vrms so a varistor is rated by clamping voltage ??? Another question i have is that why is not a zener used for DC inductor relays and a general usage diode is used . For example if i have a 200 mw 12 vdc coil can i use a 500 mw 12Volt zener as snubber ???

I presume the surge arrester is for the coil, which (when switched off) will possibly cause some switch damage. The 20A rating of the contacts is irrelevant unless the load circuit also needs some surge protection.

The energy stored in an inductor is 0.5 * L * I **2, and the varistor will absorb that much energy (less, actually) each time the relay is cycled. The "6W" rating of the coil suggests AC current in the 30 mA range, so peak current would be 50 mA. Normal AC drive reverses the current

50 times a second, so it can't take more than 10 ms to quench: If your surge protector clamps at 400V, that makes a worst-case

E = 400 * .05 * .010 = 0.2 joule

Detailed measurement of the coil inductance and equivalent resistance (eddy current losses in the iron are important) can get you a real number. The "6W" might only be related to heat losses, not necessarily inclusive of the reactive current.

• posted

as

ener

Relay contacts were not relevant to your original question? This assumes you are only discussing discharge of energy stored in the coil. If DC, a routine solution is a reverse polarity diode from the

1N400x series. When DC current disconnects, coil energy is shorted by a diode of polarity so that the energizing DC current is blocked by the diode. So that current from a discharging coil is shorted by the same diode.

If AC, a simple avalanche diode (ie from Littelfuse or Tyco) rated for above peak AC voltage (185 for 120 VAC operation or 370 for 240 VAC) is sufficient.

MOVs typically are not for repeated transient operation. MOVs degrade. Voltage further varies widely depending on currents. For example, an MOV rated at 175 volts for 120 VAC operation may not conduct significant current until voltage rises near to 300 volts OR may conduct at up to 1000 volts if current is even higher. That 175 volt number is a ballpark selection number; is not sufficient for design purposes. Actual voltages are obtained from MOV manufacturer data sheets where voltages vary widely with current.

For a 20 amp relay coil, the ubiquitous 1N400x diode or bidirectcional avalanche diode is more than sufficient AND does not degrade.

For switch contacts, a simple resistor capacitor snubber is more than sufficient. But for a relay switching 20 amps, even that is not typically used because that relay coil is so small.

• posted

:Thanks whit3rd and the rest of the guys for your answers : : :In my country mains is rated at 220V rms That is 220*1.414=311 volts peak. :That means regarding votage overshoots 275 vrms varistor is safe to use ???. : I've measured current thats how i calculated power so i imagine a 6 joule :varistor is ok . :My other question is i see on digikey 275 v rated varistors characterized as :175 vrms :so a varistor is rated by clamping voltage ??? :Another question i have is that why is not a zener used for DC inductor :relays and a general usage :diode is used . :For example if i have a 200 mw 12 vdc coil can i use a 500 mw 12Volt zener :as snubber ??? : :

You still have not detailed what you are trying to achieve by limiting the inductive voltage produced by the relay coil. Is it to prevent EMC into adjacent wiring or circuits? Is it to protect sensitive electronic equipment or other devices? Is it to minimise arcing at a series contact or switch? Tell us what you are trying to do and we may be able to help. All you have told us is that the relay coil is for 220Vac and is rated at 6W. We don't really care what you are trying to do with the contact - that is another issue entirely.

• posted

Lets make it clear what i want

First i have one 12volt dc relay controlled by a microcontroller with contacts at 220Volt arming a large relay 220V 6W I also have another dc relay from the microcntroller arming an ac motor at

25W. I used R-C snubber at the contacts of the small relay with a 275V AC X2 capacitors, 100 R resistor Problem was the capacitor was failing by open circuit after some thousands of operations .Measured the spike and it was over 1KV .The motor was producing nasty spikes during the turn off. I decided to use a varistor in shunt with RC but i need minimum space so i need the calculations for minimum requirements . Test i made improved both symptoms so to me it is an accepted solution . Varistor clamps and capacitor smothens dv/dt Symptoms from the innapropriate snubber were microntroller reset due to watchdog. and limited relay life due to arcing. As i understand i must not use a varistor since it's not for repetitive usage and use avalance diodes?????

My second question is related to the 12 dc relay .Instead of a 1n4001 why not use a zener in the 12V dc relay coil ?.I have ,as i understand, switch off protection since the zener is forward biased and also if there is a spike (not related to the relay switch ) in my supply clamp over the 12VDC .Is my assumption wrong ??? In that case a 200 mw relay inductor can be protected by a small 500mW zener in shunt???

• posted

:Lets make it clear what i want : : :First i have one 12volt dc relay controlled by a microcontroller :with contacts at 220Volt arming a large relay 220V 6W :I also have another dc relay from the microcntroller arming an ac motor at :25W. :I used R-C snubber at the contacts of the small relay with a 275V AC X2 :capacitors, 100 R resistor :Problem was the capacitor was failing by open circuit after some thousands :of operations .Measured the spike and it was over 1KV .The motor was :producing nasty spikes during the turn off. I decided to use a varistor in :shunt with RC but i need minimum space so i need :the calculations for minimum requirements . :Test i made improved both symptoms so to me it is an accepted solution . :Varistor clamps :and capacitor smothens dv/dt :Symptoms from the innapropriate snubber were microntroller reset due to :watchdog. :and limited relay life due to arcing. :As i understand i must not use a varistor since it's not for repetitive :usage and use avalance diodes????? : :My second question is related to the 12 dc relay .Instead of a 1n4001 why :not use a zener in the 12V dc relay coil ?.I have ,as i understand, switch :off protection since the zener is forward biased and also if there is a :spike (not related to the relay switch ) in my supply clamp over the 12VDC :.Is my assumption wrong ??? :In that case a 200 mw relay inductor can be protected by a small 500mW zener :in shunt??? :

Start by reading as much literature as you can.

and

I see you have also included a CR snubber across the series contact but the X2 rated cap is failing due to high peak voltage across it. The source of the high voltage, I assume, is the 220Vac relay so you need to suppress this directly at the relay coil. I would suggest a Littlefuse V250LA40C varistor will do the trick.

Provided you select a suitable varistor for the relay there is little chance of failure over an extended period of operation.

Using an X2 cap for the CR snubber is not a good idea. I would suggest a self healing polypropylene type. The WIMA FKP1 looks to be an ideal type.

For your 12Vdc relay coil you can use a 24V zener in series with a 1A silicon diode - NOT a 1N4001 btw. I would use a 1N4002 or preferably a 4004 - there is no point in trying to specify as low a voltage as you can get away with.

• posted

Thanks Ross for the reply . The thing is in the end the cost and size of the components is more that the relay it's self . I've checked the FKP 1 capacitors and they seem monstrous for the design i had in mind . I know it's playing on the safe side but putting a product on the market without thorough examination leads only to more research . Can i ask why not only a Zener alone why should somebody use also a diode .If you forward bias a zener it will conduct as an ordinary diode .So why do we need a series diode also ? My supply is regulated at 12 volts so a 12 volts zener should be ok as i think

Ross Herbert wrote:

• posted

On Sun, 08 Mar 2009 19:34:16 +0200, michael nikolaou wrote:

:Thanks Ross for the reply . :The thing is in the end the cost and size of the components is more that :the relay it's self . I've checked the FKP 1 capacitors and they seem :monstrous for the design i had in mind . I know it's playing on the safe :side but putting a product on the market without thorough examination :leads only to more research . :Can i ask why not only a Zener alone why should somebody use also a :diode .If you forward bias a zener it will conduct as an ordinary diode :.So why do we need a series diode also ? My supply is regulated at 12 :volts so a 12 volts zener should be ok as i think : :

I suppose the FKP1 could be too expensive and difficult to source so you could use a cheaper polypropylene self healing cap since the MOV across the relay will limit the voltage trying to maintain any arcing at the contacts.

With regard to your 12Vdc relay; As you know a common method of relay coil suppression is a common silicon diode such as 1N4004. Used by itself, the release delay period of a relay is often too prolonged, which can lead to problems of extended arcing on any inductive circuits controlled by that relay's contacts. Ideally, a relay should release as fast as possible within its mechanical constraints, but due to the destructive transients produced by the coil, some suppression is required. A simple resistor, a combined capacitor + series resistor or a combined zener + silicon rectifier are options, each having its own advantages and disadvantages. If cost is a problem a simple resistor shunting the coil is often useful. A better alternative is the C-R combination. LM Ericsson x-bar exchanges of the 60's era used a 0.25uF + 600 ohm C-R unit across the coil of almost every general purpose relay. This produced the best compromise between relay release delay and contact protection for any contacts in the operate paths of the relays. As semiconductor relay control became common the simple resistor or C-R unit was not appropriate and subsequently the use of either MOV's or transient suppression diodes (Transorb or other proprietary name, eg. 1.5KE series) became the norm. Due to the relatively high cost of Transorbs it was often much cheaper to use a standard zener + diode to produce similar results.

This Tyco note gives a little information on the zener + diode approach.

A zener across the relay coil by itself must be connected with the cathode towards the negative side of the coil and the anode to the positive side if it is to limit the back emf surge voltage produced by the coil during release. However, it is then obvious that, if used alone, the zener is now biased in the forward conducting state and will be permanently shunting the coil so that the relay will never operate. The silicon diode must be included and connected so that it is normally biased off and thus will not allow the zener to conduct.

In your case a 12V zener is just too close to the supply voltage. Since the purpose of it is to reduce the peak transient induced by the relay to below that of the operating transistor voltage maximum, is all that is required. Since most relay operating transistors would be in the region of at least Vce/Vbe = 40V, you can use a zener of say 20 - 25V without causing any problems for your electronics.

: : : : : : :Ross Herbert wrote: :> On Sun, 8 Mar 2009 10:15:31 +0200, "michael nikolaou" :> wrote: :> :> :Lets make it clear what i want :> : :> : :> :First i have one 12volt dc relay controlled by a microcontroller :> :with contacts at 220Volt arming a large relay 220V 6W :> :I also have another dc relay from the microcntroller arming an ac motor at :> :25W. :> :I used R-C snubber at the contacts of the small relay with a 275V AC X2 :> :capacitors, 100 R resistor :> :Problem was the capacitor was failing by open circuit after some thousands :> :of operations .Measured the spike and it was over 1KV .The motor was :> :producing nasty spikes during the turn off. I decided to use a varistor in :> :shunt with RC but i need minimum space so i need :> :the calculations for minimum requirements . :> :Test i made improved both symptoms so to me it is an accepted solution . :> :Varistor clamps :> :and capacitor smothens dv/dt :> :Symptoms from the innapropriate snubber were microntroller reset due to :> :watchdog. :> :and limited relay life due to arcing. :> :As i understand i must not use a varistor since it's not for repetitive :> :usage and use avalance diodes????? :> : :> :My second question is related to the 12 dc relay .Instead of a 1n4001 why :> :not use a zener in the 12V dc relay coil ?.I have ,as i understand, switch :> :off protection since the zener is forward biased and also if there is a :> :spike (not related to the relay switch ) in my supply clamp over the 12VDC :> :.Is my assumption wrong ??? :> :In that case a 200 mw relay inductor can be protected by a small 500mW zener :> :in shunt??? :> : :> :> Start by reading as much literature as you can. :> :>

:> and :>
:> :> I see you have also included a CR snubber across the series contact but the X2 :> rated cap is failing due to high peak voltage across it. The source of the high :> voltage, I assume, is the 220Vac relay so you need to suppress this directly at :> the relay coil. I would suggest a Littlefuse V250LA40C varistor will do the :> trick. :> :> Provided you select a suitable varistor for the relay there is little chance of :> failure over an extended period of operation. :> :>
:> :> Using an X2 cap for the CR snubber is not a good idea. I would suggest a self :> healing polypropylene type. The WIMA FKP1 looks to be an ideal type. :>
:> :> For your 12Vdc relay coil you can use a 24V zener in series with a 1A silicon :> diode - NOT a 1N4001 btw. I would use a 1N4002 or preferably a 4004 - there is :> no point in trying to specify as low a voltage as you can get away with.

• posted

Stop trying to block those transients. The solution is to eliminate those transients or shunt (conduct, short) them out. Ross has repeated what I had summarized previously. The 1N400x diode (x is any digit from 1 to 4) is located so that a charged relay coil "does not" flow current through the 1N400x diode. And so that a discharging relay coil "does" flow current through the diode. That diode must be located as close to the coil as possible so that current through the discharging relay coil does not flow through wires used by anything else.

Remember, wire is also an electronic component - not a perfect conductor.

Also not mentioned is a critical design fact. No complete isolation exists between a relay wiper and relay coil. Parasitic components mean any voltage across opening relay wiper contacts will also flow current through relay coil (and microprocessor).

A routine design puts a buffer transistor between the relay and microprocessor. To power the relay, current flows from microprocessor to digital ground. Therefore current flows from relay coil to same ground. Buffer transistor necessary to protect microprocessor from currents flowing through relay coil.

Any current that might flow from coil to ground does not also find a path to ground through the microprocessor. If a uprocessor is driving the coil directly, then current from the relay's wiper is flowing through the processor - causing the watchdog timer reset.

No motor should be generating 1Kv. A better solution is to use a zero crossing thyristor. Cutoff occurs only when voltage is crossing zero. Power is applied (delayed inside the thyristor) only when voltage crosses zero. Therefore an open circuit does not occur when motor current is maximum - as your relay is doing. Don't suppress the transient. Instead eliminate the reason for that transient.

Again, varistor has a life expectancy. They are not intended for frequent transients. Protection from frequent transients means something better - semiconductors. Ross has recommended zener diodes. However a superior solution is a type of zener diode designed just for this purpose - avalanche suppression diodes. See the 1.5KE series. So strongly recommended that it should be your very next web site visit. In your case (for AC motor transients), get a bidirectional avalanche or transient suppression diode as posted previously.

General Semiconductor adds a story about this solution. The original L1011 jumbo jet was on a maiden flight carrying Spiro Agnew, Mayor Lindsay, etc. However somebody forgot to install some avalanche bidirectional suppression diodes. That champagne flight ended early due to no working toilets. Yes, this technology and semiconductor solution is that old and that long understood. Stop wasting time with varistors. Varistors are not intended for your problem.

• posted

Ross Regarding the capacitors can i ask why a X2 capacitor is a bad choise and one should use polypropylene type ?. The thing is as i see i must use a varistor so no overshoot should be observed to destroy the capacitor. I want to use a 5mm varistor though so i don't know if it is sufficient. The thing that worried me though was a suggestion that i must not use a varistor since they seem to self destruct . Do you have an opinion on that ?. Now regarding the zener arrangement . As i know the spike is according to -L(di/dt) so the voltage produced is negative. I was thinking for the reverse connection to the zener as it is suggested so it is forward biased during the negative spike and clip during turn on . I read the article you suggested .It makes sense but i don't know if applied to a 180 mw relay there is actually such a critical behaviour that should make us consider a speed up during turn -off. Anyway as is see i must test it . Ross your suggestions are invaluable . Thank's once again

• posted

I got the idea westom

I checked the 1.5PKE and seems ok regarding size , cost . The motor i'm driving is operated every minute for 5 seconds so i get

24*60=1440 transients per day. I will actually speed thing up for testing and see what will happen after 5 years of operation. I imagine it will be enough .So your suggestions will have iummediate actions. Now regarding the 12VDC and isolation pls check my post to Ross . I know it's ancient story but the fact is that the new relays are extremely low power so somebody could check the minimum solution . Can somebody use a TVS unidirectional zener in shunt with the 12V relay inductor to protect spikes originating from the wiper as you mentioned since the zener-diode should create a similar behaviour ? The 1.5PKE seems fine though and you have my thanks for that.

Stop trying to block those transients. The solution is to eliminate those transients or shunt (conduct, short) them out. Ross has repeated what I had summarized previously. The 1N400x diode (x is any digit from 1 to 4) is located so that a charged relay coil "does not" flow current through the 1N400x diode. And so that a discharging relay coil "does" flow current through the diode. That diode must be located as close to the coil as possible so that current through the discharging relay coil does not flow through wires used by anything else.

Remember, wire is also an electronic component - not a perfect conductor.

Also not mentioned is a critical design fact. No complete isolation exists between a relay wiper and relay coil. Parasitic components mean any voltage across opening relay wiper contacts will also flow current through relay coil (and microprocessor).

A routine design puts a buffer transistor between the relay and microprocessor. To power the relay, current flows from microprocessor to digital ground. Therefore current flows from relay coil to same ground. Buffer transistor necessary to protect microprocessor from currents flowing through relay coil.

Any current that might flow from coil to ground does not also find a path to ground through the microprocessor. If a uprocessor is driving the coil directly, then current from the relay's wiper is flowing through the processor - causing the watchdog timer reset.

No motor should be generating 1Kv. A better solution is to use a zero crossing thyristor. Cutoff occurs only when voltage is crossing zero. Power is applied (delayed inside the thyristor) only when voltage crosses zero. Therefore an open circuit does not occur when motor current is maximum - as your relay is doing. Don't suppress the transient. Instead eliminate the reason for that transient.

Again, varistor has a life expectancy. They are not intended for frequent transients. Protection from frequent transients means something better - semiconductors. Ross has recommended zener diodes. However a superior solution is a type of zener diode designed just for this purpose - avalanche suppression diodes. See the 1.5KE series. So strongly recommended that it should be your very next web site visit. In your case (for AC motor transients), get a bidirectional avalanche or transient suppression diode as posted previously.

General Semiconductor adds a story about this solution. The original L1011 jumbo jet was on a maiden flight carrying Spiro Agnew, Mayor Lindsay, etc. However somebody forgot to install some avalanche bidirectional suppression diodes. That champagne flight ended early due to no working toilets. Yes, this technology and semiconductor solution is that old and that long understood. Stop wasting time with varistors. Varistors are not intended for your problem.

• posted

: :I got the idea westom : :I checked the 1.5PKE and seems ok regarding size , cost . :The motor i'm driving is operated every minute for 5 seconds so i get :24*60=1440 transients per day. :I will actually speed thing up for testing and see what will happen after 5 :years of operation. :I imagine it will be enough .So your suggestions will have iummediate :actions. :Now regarding the 12VDC and isolation pls check my post to Ross . I know :it's ancient story but the fact is :that the new relays are extremely low power so somebody could check the :minimum solution . :Can somebody use a TVS unidirectional zener in shunt with the 12V relay :inductor to protect spikes originating from the wiper as you mentioned :since the zener-diode should create a similar behaviour ? :The 1.5PKE seems fine though and you have my thanks for that. : :

Just to clarify, the 1.5KE series are not varistors but are of the Transorb (and other names) variety. If not appropriately rated for the task they can fail just as readily as varistors due to overstress. Whether you use a varistor or a TVS requires the same attention to selecting one which will continue to absorb the transients without failure over many years. If varistor manufacturers produced items which were prone to the failure rates suggested by some posters they would have law suits coming out of their backsides. Choose the right device for the job at hand and it will not fail. These devices are used in just about every telecommunications device connected to telephone lines and they rarely suffer the failures suggested by some in this thread because they are rated for the task.

With regard to your small low power relay we preferably need a make and model number. I would venture to guess that your 12V relay is of the printed circuit board type measuring approx L=20mmx B=10mmx H=12mm with 2 C/O contacts with a coil of around 360 ohms and 400mW power rating. If this is about right then use a TVS (eg. 1.5KE24CA) or a 1N4748A (22V) zener + 1N4004 diode if you want the relay to release quickly.

I also don't understand the reference to the term "wiper" used here. Such a term does not apply to any relay I have ever come across. Relays have an electromagnet, an armature and yoke assembly and a set of isolated contacts and that's about it.

• posted

5

=A0know

Testing without first doing calculations is a bad design practice. 5 year testing is only to confirm those calculations; not to see if a circuit works.

Varistors are not intended for repeated spikes. Manufacturer datasheets demonstrate the concept. With use, they degrade. Therefore varistor type devices once used for protection are now routinely replaced with semiconductor type devices. Avalanche diodes (ie Transorb, Transil, etc) do not degrade. Either it works the first day or it does not work at all.

Varistors must never fail as implied by "lawsuits", etc. A failed (degraded) varistor must have no visual indications of that failure. Datasheets make that obvious. Varistors are intended for limited events. With each use, varistors degrade - as datasheets define with numbers. Any varistor that fails by creating a visual indication leaves the circuit designer liable for a lawsuit. A completely degraded varistor has no indication of its failed - a normal event defined by the size and number of 'spikes'. And varistors are typically used where rare, not normal, and excessive currents might exist.

Current pulses induced by the wiper (armature or whatever) typically would be completely ignored by a transorb across the coil. Appreciate the two different types of current. A transorb across the coil may see no current (or voltage) as that current passes out the on one or both wires.

This is another reason for buffering the coil with a transistor. Transient current conducts through transistor to ground AND the large CB or drain to gate voltage in series with a resistor buffers the microprocessor.

Most modems do this slightly different. Rather than grounding the coil with an NPN transistor, the coil is powered from the V+ side with a PNP transistor. Even better. Therefore both types of transients (discharging coil and transient from relay contacts) are kept from the uprocessor and always have a direct connection to ground.

12 volt zener on the coil does nothing useful. That diode must be nonconductive to powering current from the microprocessor. And it must be 0.7 volts conductive to discharge the coil. Using a 12 volt zener does nothing useful and adds costs. Better is to enhance buffering on the base or gate connection to the uprocessor.

Meanwhile, some MOSFETs are perfect because the coil's shunting diode is already part of the MOSFET.

Unmentioned is the reason for spikes from the motor. Best is always to eliminate the reason for spikes rather than suppress them - as discussed previously. And not just for circuit stability. Doing so also increases life expectancies elsewhere, reduces EMI, and other reasons. Best is to not create that harmful energy in the first place.

• posted

A diode across the coil is a common fix. As the tyco reference from Ross indicates it is not necessarily a good fix. The OP read the tyco note.

One of the MOV datasheets I have used is:

It has curves for up to 1,000,000 operations. One million is not frequent?

That is almost 2 years of operation for the OP. MOVs could be further derated for more operations. The next parallel curve would be

10,000,000. May or may not be the best component for the application but is a possibility.
• posted

:On Mar 9, 10:45 am, "michael nikolaou" : wrote: :> The motor i'm driving is operated every minute for 5 seconds so i get :> 24*60=1440 transients per day. :> I will actually speed thing up for testing and see what will happen after 5 :> years of operation. :> I imagine it will be enough .So your suggestions will have iummediate :> actions. :> Now regarding the 12VDC  and isolation  pls check my post to Ross . I  know :> it's ancient story but the fact is :> that the new relays are extremely low power so somebody could check the :> minimum solution . :> Can somebody use a TVS unidirectional zener in shunt with the 12V relay :> inductor to protect spikes  originating from the wiper as you mentioned :> since the zener-diode should create a similar behaviour ? :> The 1.5PKE seems fine though and you have my thanks for that. : : Testing without first doing calculations is a bad design practice. 5 :year testing is only to confirm those calculations; not to see if a :circuit works. : : Varistors are not intended for repeated spikes. Manufacturer :datasheets demonstrate the concept. With use, they degrade. :Therefore varistor type devices once used for protection are now :routinely replaced with semiconductor type devices. Avalanche diodes :(ie Transorb, Transil, etc) do not degrade. Either it works the first :day or it does not work at all.

What do you mean by "repeated"? Is it once every second or once per minute or whatever? Provided that the varistor is selected to have the right voltage and current surge rating for the task it will not fail as you suggest.

: : Varistors must never fail as implied by "lawsuits", etc. A failed :(degraded) varistor must have no visual indications of that failure. :Datasheets make that obvious. Varistors are intended for limited :events. With each use, varistors degrade - as datasheets define with :numbers. Any varistor that fails by creating a visual indication :leaves the circuit designer liable for a lawsuit. A completely :degraded varistor has no indication of its failed - a normal event :defined by the size and number of 'spikes'. And varistors are :typically used where rare, not normal, and excessive currents might :exist.

What you failed to state is that varistors degrade when they repeatedly have to absorb surges "beyond" their specified ratings of voltage and current. Provided that the transients absorb do not produce excessive surge currents they will continue to limit transients for many years. It all depends on the correct device selection. Transorbs which are not appropriately selected for the task will fail just as easily as varistors. It is just that when they do fail it is usually on the first occurrence and the destructive effect is more violent. I have seen a transorb blown to pieces just as I have seen a varistor when subjected to ecessive voltage and current. I have also seen varistors continue operating in 240Vac operated environments for 30 years without failing.

: : Current pulses induced by the wiper (armature or whatever) typically :would be completely ignored by a transorb across the coil. Appreciate :the two different types of current. A transorb across the coil may :see no current (or voltage) as that current passes out the on one or :both wires.

This statement makes absolutely no sense at all. What are the "two different types of current" you refer to? The fact that you use the term "wiper" in relation to electromagnetic relays indicates you have little knowledge or experience with them. A transorb or varistor across a relay coil will "always" see the voltage produced by the back emf as the relay is de-energised - that is the whole purpose of using such devices. The degree of current produced by the back emf is dependent upon the magnitude of the voltage transient produced by the back emf and the coil resistance through which this current must pass.

: : This is another reason for buffering the coil with a transistor. :Transient current conducts through transistor to ground AND the large :CB or drain to gate voltage in series with a resistor buffers the :microprocessor.

In the case of the OP's 12V dc relay we know that this is controlled by a solid state device. As such, it must include some appropriately rated relay driver transistor whether this is discrete or in an IC package. Your talk of producing ground currents as a result of surges produced by relay coils is meaningless. Provided that a separate ground scheme is maintained purely for the relay circuits there will be no deleterious effect on the sensitive digital control circuits. It is all about appropriate printed circuit design and layout technique. Transorbs are more appropriate to low voltage dc operation where solid state control is used.

: : Most modems do this slightly different. Rather than grounding the :coil with an NPN transistor, the coil is powered from the V+ side with :a PNP transistor. Even better. Therefore both types of transients :(discharging coil and transient from relay contacts) are kept from the :uprocessor and always have a direct connection to ground.

??? what the heck are you on about with this statement? It has absolutely nothing to do with the OP's question.

: : 12 volt zener on the coil does nothing useful. That diode must be :nonconductive to powering current from the microprocessor. And it :must be 0.7 volts conductive to discharge the coil. Using a 12 volt :zener does nothing useful and adds costs. Better is to enhance :buffering on the base or gate connection to the uprocessor.

What utter crap. For starters, if the relay supply is 12V regulated, it makes no sense to use a 12V zener across the relay coil. You have also fallen into the trap of thinking that the zener must not be conductive to the normally applied votage polarity. This is totally incorrect. The purpose of the zener is to limit the back emf produced by the relay coil when it is de-energised. If you know anything at all it is that the back emf produced by the coil is ALWAYS of OPPOSITE polarity to that of the applied operating voltage. Consequently, in order to limit the back emf the zener MUST be conducting to the NORMAL operating polarity and only when subjected to the reverse polarity back emf does the zener "break-over" into normal zener mode. If connected as you suggest the result is no different to using a standard diode across the coil.

If the relay manufacturers (eg.Tyco, Siemens and others) suggest that a standard silicon diode MUST be used in conjunction with a zener diode for this application, then I would suggest that you are not as smart as you make out.

: : Meanwhile, some MOSFETs are perfect because the coil's shunting :diode is already part of the MOSFET.

Yes, but this is not relevant to the OP's question. Trying to demonstrate your knowledge of MOSFETS which include an integral diode is immaterial.

: : Unmentioned is the reason for spikes from the motor. Best is always :to eliminate the reason for spikes rather than suppress them - as :discussed previously. And not just for circuit stability. Doing so :also increases life expectancies elsewhere, reduces EMI, and other :reasons. Best is to not create that harmful energy in the first :place.

Oh great. And just how do you suggest that you eliminate spikes produced by a motor? Transients produced by motor windings are just part of the deal - you can't just "design them out" of the equation. As long as there are inductive windings in a motor there will be transient spikes - end of story. So you are stuck with having to suppress or moderate these spikes in exactly the same way as suppressing the back emf from relays. BTW, varistors are commonly used to perform this function in electric motors as well. Even motors rated at hundreds of horsepower have their winding transients suppressed by the common varistor. Admittedly these are much bigger than the disc types we commonly encounter. It would be logical to assume that in this particular application the varistors would be subject to repeated and constant transients, but do we hear of frequent failure of electric motors due to the failure of varistors which have to endure these transients? I would suggest not. It is all about selecting the right varistor for the job and it will rarely, if ever, fail.

• posted

Guys

I will be more specific about my points . The relay i will use is

5V 120mw . Contact rating is 5Amps at 220V . Its clever as i understand to use NPN,PNP to drive the relay but i imagine a clever diode arrangement can drive this relay from an MCU pin.As i understand what Ross is trying to say is the differential modes of noise (turn off relay transient ) is dealt with diode but common mode (armature leakage ) must be dealt differently by isolating low voltage circuit from hi side. Now the thing is, after checking tyco notes, do all these things apply to a 120 mw relay about turn off time expansion etc., or i'm after ghosts ?.My feeling is that a simple diode 1n4148 should be sufficient. So that will be checked.

The relay i'm driving is drawing 30ma at 220V .But creates fast rising spikes without any protection. Actually it is a common din-rail relay found almost everywhere.I imagine a 600 W avalanche diode should cope with that and the thing about these components is small size and smd case compared to varistors.

The motor i'm driving is rated at 25W @ 220Volts . I will measure the inductance as it is described in TYCO notes . From that point how do you calculate if 600 W is enough or 1500 W is the minimum transorber or not ?? The motor operates every minute one time and the relay 50 times per day. Any help would be appreciated .

Ross Herbert wrote:

producing

no

limit

operating

zener

standard

frequent

• posted

:Guys : :I will be more specific about my points . The relay i will use is :5V 120mw . Contact rating is 5Amps at 220V . Its clever as i understand :to use NPN,PNP to drive the relay but i imagine a clever diode :arrangement can drive this relay from an MCU pin.As i understand what :Ross is trying to say is the differential modes of noise (turn off relay :transient ) is dealt with diode but common mode (armature leakage ) must :be dealt differently by isolating low voltage circuit from hi side. Now :the thing is, after checking tyco notes, do all these things apply to a :120 mw relay about turn off time expansion etc., or i'm after ghosts :?.My feeling is that a simple diode 1n4148 should be sufficient. So that :will be checked. : :The relay i'm driving is drawing 30ma at 220V .But creates fast rising :spikes without any protection. Actually it is a common din-rail relay :found almost everywhere.I imagine a 600 W avalanche diode should cope :with that and the thing about these components is small size and smd :case compared to varistors. : :The motor i'm driving is rated at 25W @ 220Volts . :I will measure the inductance as it is described in TYCO notes . : From that point how do you calculate if 600 W is enough or 1500 W is :the minimum transorber or not ?? :The motor operates every minute one time and the relay 50 times per day. :Any help would be appreciated . : : :

Quite frankly, I would get away from using relays at all. If you are intending to use 5V logic to control a small relay and then use the contact on this small relay to control the 220Vac relay, it seems a bit archaic.

For starters, why not use a snubberless triac opto-coupler such as Fairchild FOD410

as in Fig.1 to directly control the motor. You don't have any problems with inductive spikes or contact arcing and you get a miunimum of 5kV isolation between the 220Vac side and the

5V logic side. I don't think your 5V, 120mW relay has 5kV isolation rating between the contacts and the coil does it?

If you are still inclined to use your small relay then be aware that if you put this relay on your mcu board there will be 220Vac across its contacts. I don't consider it wise to allow potentially dangerous voltage to be connected to a board where there is a probability of careless fingers coming into contact with it. It is far more preferable to have the logic output of your mcu board buffered to an output connector and cabled away to a separate board on which any relays with 220Vac on them are mounted in an enclosure. Going down this path you still need to suppress the back emf from the small relay and use a C-R spark quench across the contact which will be driving the 220Vac relay. Also, as discussed you will also need to suppress the transient spikes from the 220Vac relay coil to minimise any arcing at the contact of the small relay.

If you want to see which is the best method why not try them out and see what happens. Start with the simple diode across the small relay and a 250V (or 275V) varistor across the 220Vac relay coil. If a 20mm varistor is too big then use a

14mm type, but I wouldn't go any smaller than this. If you notice any appreciable arcing at the contact of the small relay in low light then you should use either a Transorb or zener + diode combination as suggested previously to speed up the realy release and see if this reduces the arcing.

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.