Selecting an MOV for a power supply

[Guineafowl]

I have a 30V 5A variable linear power supply, which has failed twice from transients killing the LM358 quad op-amp IC that controls the output.

These transients are from driving things like a car ignition coil, for making an electric fence energiser. Simply replacing the LM358 gets it going again, but can I add protection to prevent this happening again?

My idea is to place an MOV across the output. However, I’m not sure of the rating - the electronics supplier quotes Max DC voltage, Clamping voltage and Varistor voltage for each unit. Given my power supply’s max output is 32V, how do I select the correct MOV?

 

[berkeman]

Can you post a schematic? And you're saying these are transients from your load on the power supply output?

 

[Guineafowl]

Can you post a schematic? And you're saying these are transients from your load on the power supply output?

Yes, it’s a commercial power supply, but a cheap one. Not much of a schematic, but one time the load was a car ignition coil, and I was showing a young lad how a spark is produced at the secondary when the primary was disconnected. Then we added a capacitor across the primary to induce oscillations and make a better spark. Then the power supply blew.

Now I think of it, I remember @Tom G suggesting a suitably rated zener opposed with a regular diode across the output, but this would suit repeated transients. I think an MOV would be a simpler solution for infrequent transients.

To clarify: The PSU output does not exceed 32 V, so I need an MOV to clip voltages just above this, say 36V. What is the meaning of clamping, rated and varistor voltages in MOV datasheets? How do they apply to selecting an MOV for this application?

 

[berkeman]

failed twice from transients killing the LM358 quad op-amp IC that controls the output.

Yes, it’s a commercial power supply, but a cheap one. Not much of a schematic,

Well, the LM358 is not likely directly connected to the outputs, right? There must be at least a current-gain BJT driving the output that the opamp is driving. Whatever.

Yeah, just connect a Zener or MOV across the output to limit the transients seen by the output. Check the Zener and MOV curves to select a value that will not conduct during normal operation, but will conduct for transients above that.

 

[Tom.G]

A minor problem with using a MOV is they degrade a little bit every time they clamp the voltage. Of course you can always go to a higher power one to increase the lifetime.

Other things to consider:
MOVs have a less precise operating voltage than a Zener does
High power Zeners cost upwards of $20, MOVs often under $1

Another approach is to use a circuit with a Zener driving an SCR (or maybe a MOSFET) to clamp (crowbar) the output. That is probably the most effective at a reasonable price; I'm actually a little surprised the power supply doesn't already have one bulit-in. Using an SCR is most appropriate when the power supply has either a built-in current limit or an easily replaceable fuse.

Google for 'crowbar circuit' or 'crowbar circuit design'

Cheers,
Tom

 

[Averagesupernova]

Did quick google. I would look into these:
-
https://www.google.com/search?q=p6k...droid-verizon&sourceid=chrome-mobile&ie=UTF-8

 

[Guineafowl]

Thanks all for your replies. I had a look at this:

Suggested mods: Because the power supply has a current limit, there would be no need (I guess) to try to blow a fuse, so the anode of the SCR could just be connected to the positive rail directly above it. The cap is to prevent tripping for brief transients, which I want, so could be omitted.

However, both this circuit and the TVS seem best suited to clipping ‘forward’ transients FROM the power supply. The transients that killed the IC were presumably retrograde and reverse voltage (coil flyback). Would the crowbar or TVS set-ups clip these?

I was thinking MOV for its bidirectional action, as well as simplicity and cheapness. Your thoughts would be welcomed, however.

 

[Averagesupernova]

I would be a little careful the way you hook an SCR to the output. Go with what another company has done. Astron has been building reliable power supplies for years. There are many places on the web that their schematics are posted. You need a certain amount of capacitance in order to prevent nuisance firing of the SCR. Also, be careful that the SCR is protected from overcurrent on the gate. The SCR is not infinitely fast, so have some current limit on the gate.

 

[Tom.G]

reverse voltage (coil flyback). Would the crowbar or TVS set-ups clip these?

The TVS (MOV) would, so would a simple diode (rated at a few amps) connected reverse polarity across the output. You are right about a crowbar, it would not protect against reverse polarity spikes.

 

[berkeman]

Thanks all for your replies. I had a look at this:

View attachment 227269

Suggested mods: Because the power supply has a current limit, there would be no need (I guess) to try to blow a fuse, so the anode of the SCR could just be connected to the positive rail directly above it. The cap is to prevent tripping for brief transients, which I want, so could be omitted.

However, both this circuit and the TVS seem best suited to clipping ‘forward’ transients FROM the power supply. The transients that killed the IC were presumably retrograde and reverse voltage (coil flyback). Would the crowbar or TVS set-ups clip these?

I was thinking MOV for its bidirectional action, as well as simplicity and cheapness. Your thoughts would be welcomed, however.

So you want to blow your input fuse any time there is a transient on your PS output? I understand that you are frustrated at having to replace the opamp IC, but going for a protection strategy that blows the fuse instead or just absorbing the transient in a protection circuit doesn't seem wise to me...

 

[Guineafowl]

So you want to blow your input fuse any time there is a transient on your PS output? I understand that you are frustrated at having to replace the opamp IC, but going for a protection strategy that blows the fuse instead or just absorbing the transient in a protection circuit doesn't seem wise to me...

The supply has a current limit, so a crowbar would just activate that.

 

[Tom.G]

Putting all the above comments together, it seems that the damage is caused by a significant reverse polarity voltage entering the supply, getting thru the voltage sense circuitry, and destroying the op-amp. An active crowbar, as I first suggested, will not necessarily prevent this.

The reverse voltage must be either blocked or clamped before it gets into any active circuitry. Blocking could consist of a series diode at the output terminals, but this partially defeats the voltage regulation due to the diode voltage drop.

Clamping would be something across the output then effectively shorts any reverse voltage. This clamping could be a reverse-biased diode or a surge suppressor (TVS or MOV). A MOV is bi-directional and also will not conduct until the threshold voltage is reached in either polarity, fine for over-voltage, but will not clamp a reverse voltage until the threshold voltage is reached.

Therefore, I believe a reverse polarity diode across the output is the best solution for your present problem. A crowbar is fine for overvoltage clamping, but that is a separate problem that you don't seem to have (so far).

Cheers,
Tom

 

[berkeman]

The supply has a current limit, so a crowbar would just activate that.

Not the way your schematic is drawn. The crowbar is connected to the Input to the power supply, just after the fuse.

View attachment 227269

 

[Guineafowl]

Not the way your schematic is drawn. The crowbar is connected to the Input to the power supply, just after the fuse.

View attachment 227269

Because the power supply has a current limit, there would be no need (I guess) to try to blow a fuse, so the anode of the SCR could just be connected to the positive rail directly above it.