Thread: OT: How to choose proper varistor to protect bridge rectifier

We have a simple 120 AC in 120 DC out motor control manufactured for us. It uses a KBPC5010 bridge rectifier and the DC motor it drives draws around 9 amps full load. In a small number of cases (1/2 %) the bridge rectifier fails. I am thinking it might be due to voltage spike. Wondering if someone could help with proper selection of a transient/Surge Abbsorber. Another manufacture of a similar device is using a HEL-20D471K.

As you are aware diodes are very touchy about their reverse voltage, your one is rated at 1000V seems enough!!. The trouble is that brushed motors and switching them off can generate high voltage spikes, which it appears in your case exceed 1000V . The device you mention (HEL-20D471K) is meant to clamp the voltage to about 775 V max, which though close in theory should help a lot. While looking for the data sheet for the 471, I came across a 431, which has a clamp voltage of 330V, which is better.
The problem with these devices is that they are no faster then the diodes, i.e. if a spike came along its a guess which one blows first. I prefer to put capacitors from the live to neutral to bypass these spikes, though these capacitors must be of a huge working voltage, say 2KV, else again its a guess which one blows first, the capacitor need only be .001 MF, very cheap, pennies each. Don't ignore spikes that might come down the mains lead either.
Years ago we tested a mains spike detector in our office, big office block BUT no heavy machinery at all, just computers etc. the gizmo registered loads of spikes, the frightening thing was that it gave one of > 7KV every couple of minutes!!
Frank

what is your drive model and mfgr? I will assume at this point it is a simple SCR drive since you state your 120vac input gives 120vdc output: must be simple bridge rectifier and no capacitor input to smooth it to 160vdc peak; hence, cheap SCR drive on it since most servo drives will not run well on simple bridge rectified DC well.

that said, and scr will not backfeed high voltage spikes. see reason for me asking what your drive is?

now if it IS a FET or IGBT drive, then it does not produce ANY voltage spikes when the semiconductor turns OFF: virtually all industrially mfgred drives today have backward polarity diodes across the power switching devices to totally prevent a voltage spike. BUT... they WILL produce a voltage spike upto 2-3x the input bus voltage at TURN ON due to ringing in the tuned Xl/Xc output load. so 3x 120= 360v so this is not likely your cause of bridge failure. I would look closely at its CURRENT rating vs your load rating - is it undersized? does it get hot?

lastly I would consider spikes coming in from the ac input side. are you in india? Mozambique? a location where input voltage is very filthy? rich in spikes?

more info needed to help you.....

Thanks for your help so far. To further explain our application. The drive is not purchased. We have a local shop build it for us. Basicly it is just a 30 amp bridge rectifier with two relays to allow the motor to be reversed. It has no capacitors or other protection of any kind. The application is for motorizing boat lift winches. We have sold these for the past three years. I would guess we have around 1000 120V units in service. Because they are on lakes it is possible that they are exposed to more than a normal share of lighting. I should also mention that they all have ground fault protection. To date we have gotten a dozen or so back with blown rectifiers. We have tried changing the mfg of the bridge rectifier but they have had a couple failures as well. Our competitor who builds a unit very simiar to ours started using a HEL-20D471K surge protector accross the AC side. So am I correct to do things right we should have protection on the AC side for surges and also on the DC side to.

MOV and similar devices are EXTREMELY variable in voltage.... they are very diffiult to properly size if your application is at all fussy.

You can probably do VERY well with a small capacitor across the *motor*.... better yet, one of the resistor/capacitor combination "snubber" parts which are MADE for this purpose.

You are most likely killing the rectifier yourself with the arc and voltage spike as the relay opens. The capacitor/resistor combination damps the spike better than an MOV device will.

Typically they are 0.1 uF and 10 to 20 ohms, in series. The cap absorbs the current of the spike, and the resistor dissipates the small amount of energy, to prevent ringing.

If the motor inductance is known, a very good guess as to the exact optimum part can be made..... but I'd not bother.

As a teen I worked a couple of summers for a (very) small Midwestern motor control maker. At that time they typically sold into the farm industry for things like silo elevators and ventilators. Quite a few returns due to farm environment abuse, including shoddy end-user wiring, lightning strikes, high-pressure manure sprayers and (more than once) cattle urinating on the equipment. It was not unusual to get a control back with several square inches of 4oz or 6oz copper traces just vaporized.

They did eventually figure out that it was cheaper in the long run to build in robust protection than to handle the returns on minimal-cost products. They apparently got out of the farm equipment segment, too. 40 years later, they are still in business and rather larger now.

as jst said that you are likely killing the rectifier yourself with the non suppressed contacts, i would like to add:

99.9% prbability you are killing the rectifier yourself with those contact openings - when they open they can easily generate well over 1000v!

Snubber as jst said will probably fix it. wouldnt a reverse biased diode across the contacts be a good/better solution even than rc snubber? or 300v MOV across the motor armature itself?

So to be double safe I could put a surge absorber accross the AC side and a "snubber" accross the DC? JST- I would also appreciate further discussion on merits of diode accross relay contacts.

oragrad, if you are mfgring this at 1000pc quantities, you really should be doing more homework that asking here on a forum... do you own an oscilliscope? If you are building 1000pc as oem of a device that is beginnig to fail, you need to hook your scope up and SEE FOR YOURSELF what the spike is you are generating, then add snubber, and or diodes and or MOVs TO FIX IT FOR YOUR CUSTOMERS! This is not a 1 time maybe I can make it live a little longer home shop question - you owe it to your customers to research and fix this better than you have so far from your previous posts.

as an oem of this product, dont you agree?

Diode will damp the spike also......

HOWEVER.......

A slight re-design will fix a lot of this...

1) interlock so the reverse MUST have the main relay open before being switched....

2) leave ONLY the reverse relay between the rectifier and motor, and move the power relay to BEFORE the rectifier.... in teh AC side (it may be there already).

With the rectifier connected that way, it forms the diode across the motor. No extra parts...... the motor is never disconnected from the rectifier without power being off to it.

I would still put in the snubber unit, right across the motor.

http://www.alliedelec.com/images/pro...S/70186531.pdf

http://www.alliedelec.com/images/pro...S/70030263.pdf

Originally Posted by ORAGRAG

I should also mention that they all have ground fault protection. To date we have gotten a dozen or so back with blown rectifiers. .

If 1000 volt spikes are destroying that bridge, then you are replacing the ground fault, coffee pot, telephone equipment, etc just as often.

If 1000 volts is destroying the bridge, then the same current is also inside destroying other parts.

View how a bridge works. When one side is trying to stop a reverse voltage, the other side is connecting current so that a lower voltage exists.

More likely failing is an exceeded current rating OR the bridge is not properly connected to a required thermal conductor.

From your post are zero reasons to suspect a high voltage spike. And plenty of reasons to suspect other problems.

Lightning would never be a problem if the facility has what all such facilities must have. A properly earthed 'whole house' protection solution. Where the word 'earth' is most important and means more than just meeting code. No problem especially if the bridge is over 600 volts.

Originally Posted by westom

If 1000 volt spikes are destroying that bridge, then you are replacing the ground fault, coffee pot, telephone equipment, etc just as often.

If 1000 volts is destroying the bridge, then the same current is also inside destroying other parts.

There ARE no other parts, essentially, the control box seems to be two relays, control buttons, and the bridge......

Overcurrents on a 50A bridge rectifier are not too likely..... rectifiers are quite tough, and one would suppose the unit is fused against long term overload.

The "Ifsm" (overload) is 500A single cycle for this part (KBPC5010), ON TOP OF rated load.

http://www.comchiptech.com/cms/UserF...ies%20RevA.pdf

Of course it isn't impossible, but should require some other errors in fusing/circuit protection.

Originally Posted by JST

Overcurrents on a 50A bridge rectifier are not too likely.....

Everything in that post remains unchallenged. Furthermore, mike_kilroy defined what should be obvious to everyone. Even an informed rookie technician does not change a design on speculation. First a problem is identified. Ie use a scope. Then a solution is implemented. Anyone fixing something only on wild speculation (ie varistors to stop failing diodes) wants to surrender his job to a Chinaman.

Never fix anything on wild speculation. Identify a reason for failure before fixing that failure. If 1000 volts were destroying the bridge, then that facility as a long list of other items simultaneously destroyed.

EDN magazine features “Tales from the Cube”. Stories of how people solved the strangest problems by simply doing what is said even in CSI. “Follow the evidence”.

Originally Posted by westom

Everything in that post remains unchallenged. Furthermore, mike_kilroy defined what should be obvious to everyone. Even an informed rookie technician does not change a design on speculation. First a problem is identified. Ie use a scope. Then a solution is implemented. Anyone fixing something only on wild speculation (ie varistors to stop failing diodes) wants to surrender his job to a Chinaman.

Never fix anything on wild speculation. Identify a reason for failure before fixing that failure. If 1000 volts were destroying the bridge, then that facility as a long list of other items simultaneously destroyed.

EDN magazine features “Tales from the Cube”. Stories of how people solved the strangest problems by simply doing what is said even in CSI. “Follow the evidence”.

Technically, that is correct.....

In practice, experience shows that the design as described is GOING to have voltage spike problems.... If properly protected, it will NOT have overcurrent issues, and in any case a lot of blown fuses with no other faults would also be a factor. That has not been described.... Perhaps because it was not considered relevant, of course....but......

And, it does not sound as if the designers are going to look with a scope, nor that they even own one. If they DID, they would already have looked, seen the problem, and fixed it.

Putting a snubber on would be a routine necessity of sensible design..... and it SHOULD have been on there already. It should NOT be necessary to "prove" that a destructive spike will occur, unless the bean counters are asses.

Once the elementary protection is in place as it should have been already..... THEN we can talk about investigations that they likely will not do.

As for all the other office equipment that should also be destroyed....... this is DOCK EQUIPMENT AT A LAKE.... likely hundreds of feet from any such "office", and the spike problem is a LOCAL issue arising from the design.

Don't let the perfect world interfere with reality...... And I agree about the investigation..... but I ALREADY KNOW that there WILL be spikes if ther reverse relay is triggered while the motor is running. That CAN also lead to overcurrents, but where are the blown fuses with no cause? (I expect the fuse to blow after the bridge fails, that's not in question)

Originally Posted by westom

If 1000 volt spikes are destroying that bridge, then you are replacing the ground fault, coffee pot, telephone equipment, etc just as often....

If 1000 volts is destroying the bridge, then the same current is also inside destroying other parts.

View how a bridge works. When one side is trying to stop a reverse voltage, the other side is connecting current so that a lower voltage exists.

From your post are zero reasons to suspect a high voltage spike. And plenty of reasons to suspect other problems.

You suggest the job may go to china..... I THINK IT IS IN CHINA ALREADY. See OP.....I think this is why they have the problem and I suspect JST is correct that they don't own a scope.

But, there is good reason to know there is a hi voltage spike. they are interupting the current flowing in a motor w/o any snubber system in place. This WILL generate a hi voltage spike. Our guys designed a system with small 24vdc 8 amp solenoids: after frying fw rectifiers like these and a few relays, I got involved and scoped what they had: 600v spike on turn off. his motor at 120vdc will produce more voltage than my little 24v solenoid.

this is not a hi current spike - a hi voltage spike with low joules and it is on the back side of the diodes. both these things will keep it from burning out the coffee pot across the street, just take out the rectifier and make the relay contacts life very short.

I do want to have a quality product. That is why I am looking into how to correct this problem. My company is primarily a distributor of marine products and this is just a little side line buisness that we have going. While I have a strong backround in mechanical design, myself or no one in our comapny is strong in electronics. We do import the raw motors and gear boxes fron China but we have the all of the electricals assembled by a local company here. However they are not design people and do not understand what is causing the problem. The low failure rate was and is a bit perplexing to me because my thought is that if we had a "design" issue the failure rate would be higher. So going into this thread I was just looking for a little help to get pointed in the right direction and to see if it was somthing simple or if we needed to get professional help. Also if we needed to get professional help I wanted to be better informed as to what the possible problem is. I even thought that possibly some"qualified member" on the forum might be interested in providing design help for hire.
So with that said any recomendations on where to go with this for professional help?

Originally Posted by ORAGRAG

I do want to have a quality product. That is why I am looking into how to correct this problem. ...So with that said any recomendations on where to go with this for professional help?

First understand the fundamental bottom line point. You saw what failed. So tried to fix a failing part rather than first learn the problem. Most any electrical engineer can find the problem IFF he has proper equipment. Ie oscilloscope.

However significantly cut costs by first identifying patterns. Or find some means of making the failure repeatable. Intermittents are many times more expensive to find.

Fixing is trivial. Finding the reason for such failures is harder. And must be performed BEFORE fixing anything. That was the point of previous posts. What is creating more than 1000 volts or more than 50 amps? Or what is not conducting heat? Any decent engineer can fix it (ie snubber) only after the relevant questions are answered. Always discover the problem long before even considering a solution.

Westom- I totaly agree with you on all your points. Most importantly your point of the problem needs to be identified before it can be fixed. This is the main reason I started the thread.

To date we have been unable to recreate the problem at our facitlity in any way shape or form. Also in talking to customers who have had problems we have not found anything that stands out. The usual story is that they go to use their lift and it fails to work. We bring the unit back here and the rectifier is shorted out. In a couple cases one of the relays also had a stuck contact. With my limited knowledge I was trying to get some idea if the problem is on the DC, AC or possibly both sides.

Originally Posted by ORAGRAG

The usual story is that they go to use their lift and it fails to work. We bring the unit back here and the rectifier is shorted out. In a couple cases one of the relays also had a stuck contact.

In trying to second guess what this circuit is. And now with that post - WOW. Stuck relays? I now have a completely different picture. Someone lowering the boat. Then immediately switching those relays to raise the boat enough to stop its drop faster. DC motor's momentum means it is a generator literally drawing many times more current from the utility's nuclear power generators. Fortunatly the diodes blew to protect human life from a motor that does not stop.

What are ratings of those relay contacts?

What is shorting out energy generated by (stored in the windings of) a still spinning DC motor? Those fused(?) relay contacts were critically important information. Exactly what was the operator doing when relay contacts fused? I have a very bad feeling about a motor that continues running because relay contacts fused closed. This now sounds like an 'emergency must be solved now' human safety threat.

Originally Posted by ORAGRAG

....To date we have been unable to recreate the problem at our facitlity in any way shape or form.

huh? You either have 1000+v spike or you are plug revesing the motor with no current limit or both. BOTH ARE EASILY REPRODUCIBLE AT YOUR SHOP.

WHERE ARE YOU LOCATED?

DID YOU BUY A SCOPE YET?

So to reproduce do this:

1) hook up a boat to the winch. begin raising it then stop it - watch the relays for big sparks - any? that is your voltage spike generator.

2) 1) hook up a boat to the winch. begin lowering it then stop it - watch the relays for big sparks - any? that is your voltage spike generator.

3) hook up a boat to the winch. begin raising it then reverse it before stopping it - watch the relays for big sparks - any? that is your over current source.

4) hook up a boat to the winch. begin lowering it then reverse it before stopping it - watch the relays for big sparks - any? that is your over current source.