Surge supressing for Solid State Relay?

Hi there. So I'm having a problem with a couple SSRs in a test box I built dieing.

Basically, this box (and a power supply) is used to charge a large inductor coil. For that role it works great. Its actually 3 inductor coils (A B and C) that are all attached together at one end. The box is used to direct current as needed (A+B-, A+C-, B+C-). Used to check the polarity of the unit at a number of points. Works fine for that. The solid state relay in question is used to switch the 15V@5A that is used to charge the coils during the test. The relay we are using is rated for 220VDC@25A.
Clearly we should be safe with the rating of that relay (even assuming a voltage spike from when a coil is discharged)

My thought for why the relay has broken is that it is used to check the inductor array AFTER a HiPot test (2000VDC). We know for a fact that there is a bit of capacitance to the unit (if you touch any of the 3 A,B,C terminals together or one of them to the core, it'll give you a little jolt). I'm guessing that when the unit is hooked up to the testing fixture, that charge that the unit holds is giving a little jolt to the relay, and after a few weeks of operating, bye bye relay.

So I'm working a couple angles to try and counter this. Aside from a manual way of shorting all the terminals to a ground to discharge them first, any suggestions on a surge protection method that could be added to the relay?

I don't find your description of the arrangement very informative. Can you provide a schematic? How was the HiPot test carried out?

If you use the SSRs to break the 5A current flow to the coils, you can expect a large inductive spike, quite possibly more than 220VDC depending on the inductance and capacitance of the coils and the core material. When switching an inductive DC load with an SSR or any other device, you should use either a flyback diode (possibly with a resistor in series, if you need the current to decay faster), or a R-C snubber network (but that may not be practical with a 5A 15V load).

What's limiting the current? Are you sure it's limited to 5A?

I don't understand the capacitance either... I agree, we need to see a schematic. 2000V from a capacitor across the SSR will probably kill it.

If there is a capacitor across where the high-pot voltage is applied, you can get "unlimited" current when it's discharged, just as you can get "unlimited" voltage when you break the current path through a coil.

I'd say you need to discharge the capacitor as a routine part of high-pot test, and I'd say discharge needs to be an automated part of the high-pot test to protect both humans and the SSR.

P.S.
You might consider using a regular 'ol mechanical relay. Mechanical relays are generally more rugged and more tolerant of abuse. A solid state relay can last forever if it's not abused but it can also be killed instantly.

A mechanical relay will eventually wear-out, but it's usually cheaper to replace and it's usually in a socket.

You might need a relay driver to power the mechanical relay coil, depending on what's currently driving the SSR.

Controlling inductive loads with A SSR can be very problematic as the resulting voltage and current go out of phase with each other and that can cause major problems with turning on and off of the internal triac/SCRs. Carefully calculated and sized "snubber" circuits are the general solution for using SSRs with inductive loads. Or go with a good quality electromechanical relay instead.

Lefty

retrolefty:
Controlling inductive loads with A SSR can be very problematic as the resulting voltage and current go out of phase with each other and that can cause major problems with turning on and off of the internal triac/SCRs.

Looks like he’s controlling DC not AC, so that doesn’t apply here.

dc42:
I don’t find your description of the arrangement very informative. Can you provide a schematic? How was the HiPot test carried out?

If you use the SSRs to break the 5A current flow to the coils, you can expect a large inductive spike, quite possibly more than 220VDC depending on the inductance and capacitance of the coils and the core material. When switching an inductive DC load with an SSR or any other device, you should use either a flyback diode (possibly with a resistor in series, if you need the current to decay faster), or a R-C snubber network (but that may not be practical with a 5A 15V load).

Sorry about that but I’ve been nowhere near a scanner or camera with which to upload a schematic, and the majority of my work is away from computer as well so I haven’t gotten to make a proper schematic on there either. I’ll throw a crappy drawn one up now, I’ll come up with a better cleaner one later.

Also the HiPot test is carried out by applying 2000VDC to the A,B,C nodes, relative to the core (and some thermistors built into the unit)

DVDdoug:
What’s limiting the current? Are you sure it’s limited to 5A?

I don’t understand the capacitance either… I agree, we need to see a schematic. 2000V from a capacitor across the SSR will probably kill it.

If there is a capacitor across where the high-pot voltage is applied, you can get “unlimited” current when it’s discharged, just as you can get “unlimited” voltage when you break the current path through a coil.

I’d say you need to discharge the capacitor as a routine part of high-pot test, and I’d say discharge needs to be an automated part of the high-pot test to protect both humans and the SSR.

P.S.
You might consider using a regular 'ol mechanical relay. Mechanical relays are generally more rugged and more tolerant of abuse. A solid state relay can last forever if it’s not abused but it can also be killed instantly.

A mechanical relay will eventually wear-out, but it’s usually cheaper to replace and it’s usually in a socket.

You might need a relay driver to power the mechanical relay coil, depending on what’s currently driving the SSR.

The current is primarily limited by the resistance of the inductor coils themselves. But basically its hooked to a power supply and the operator dials the voltage and amperage to 15@5 when hooking up the first unit and then keeps those settings throughout the tests. It is able to jump a bit over 5A from unit to unit, however if one were far enough out of spec to have that drastic of a change, it would have been caught at an earlier test.

Me I don’t understand the capacitance either. Checking it with a voltmeter on about all the points I can’t find a C higher then 200fF which really shouldn’t amount to anything.
On the automation side of discharging the unit, I’ve thought about having our testing bed pull all terminals to ground after the full test but (in the developer’s infinite wisdom) they never added a feature like that.

As far as the mechanical relay, the boss says no. (for now, probably after another relay breaks it might be a doable compromise)

retrolefty:
Controlling inductive loads with A SSR can be very problematic as the resulting voltage and current go out of phase with each other and that can cause major problems with turning on and off of the internal triac/SCRs. Carefully calculated and sized “snubber” circuits are the general solution for using SSRs with inductive loads. Or go with a good quality electromechanical relay instead.

Lefty

Yeah DC only for tests, the unit is operated in DC as well. As far as a snubber circuit, I looked them up, but that whole carefully calculated part gets kinda in the way (ME intern, not EE, which leads me to wonder why I’m the one fixing this also)

I’ve been thinking to opt for a flyback diode, however I wonder where exactly to put in. I could attach it across the terminals of the PSU, or the relay (or perhaps I could throw them everywhere, flybacks for everyone)
Note in the schematic, the selector switch is rated at 125VAC@25A, though it is of much better make then the relay is. The emergency stop switch is also rated for 125VAC@25A, but is kinda cheap. Neither of these have failed (yet)

Scan1.PDF (75.1 KB)

Nikarus:
Also the HiPot test is carried out by applying 2000VDC to the A,B,C nodes, relative to the core (and some thermistors built into the unit)

Are the coils isolated from the SSR when the HiPot test is carried out? If so, how? If you are relying on the selector switch contacts all being open when the HiPot test is carried out, then you need to use a selector switch rated for 2000V.

Nikarus:
I've been thinking to opt for a flyback diode, however I wonder where exactly to put in. I could attach it across the terminals of the PSU, or the relay (or perhaps I could throw them everywhere, flybacks for everyone)

The flyback diode should be connected between the SSR output/selector switch junction and the battery -ve terminal. That will protect the SSR from negative voltage transients. You might want to connect a 100V TVR diode there as well, to protect the SSR from positive voltage surges during the HiPot test.

dc42:

Nikarus:
Also the HiPot test is carried out by applying 2000VDC to the A,B,C nodes, relative to the core (and some thermistors built into the unit)

Are the coils isolated from the SSR when the HiPot test is carried out? If so, how? If you are relying on the selector switch contacts all being open when the HiPot test is carried out, then you need to use a selector switch rated for 2000V.

Yes the HiPot test is completely separate, often happening a couple days before the polarity test that this box is for

Nikarus:
I’ve been thinking to opt for a flyback diode, however I wonder where exactly to put in. I could attach it across the terminals of the PSU, or the relay (or perhaps I could throw them everywhere, flybacks for everyone)

The flyback diode should be connected between the SSR output/selector switch junction and the battery -ve terminal. That will protect the SSR from negative voltage transients. You might want to connect a 100V TVR diode there as well, to protect the SSR from positive voltage surges during the HiPot test.

-ve terminal? Negative? iight then. About how big of a diode should I use though?

Use a flyback diode rated at 5A or more. Connect anode to battery negative, cathode to SSR output.

dc42:

retrolefty:
Controlling inductive loads with A SSR can be very problematic as the resulting voltage and current go out of phase with each other and that can cause major problems with turning on and off of the internal triac/SCRs.

Looks like he's controlling DC not AC, so that doesn't apply here.

Well then how will be able to get the SSRs to turn off it indeed he is using DC as the voltage source, SSRs require the AC zero crossings to switch off current flow.

Lefty

DC SSRs are mosfet-based, not triac-based. So no zero crossings are needed to turn them off. He said in the OP that the SSR was rated 220VDC.

Any reason why the SSR would have a small transformer inside of it then?
After dismantlement, theres a MOSFET, little transformer, and a board with an LED that I'm assuming does the optical isolation to protect the microcontroller thats triggering the relay.

The transformer is probably what provides the isolation.

dc42:
DC SSRs are mosfet-based, not triac-based. So no zero crossings are needed to turn them off. He said in the OP that the SSR was rated 220VDC.

It would be helpful if the OP just posted a link to his actual SSR so we don't have to guess or assume what he is using. DC SSRs are rather specific to manufactures and are not as generic in nature as AC SSRs are. I know you know this stuff but it's really an assumption, in my opinion, on what SSR device he is actually using and a link would make me more comfortable participating any further on this thread.

Lefty

http://www.ebay.com/itm/Solid-State-Relay-SSR-DC-DC-25A-3-32VDC-5-220VDC-25A-/170961637625?pt=LH_DefaultDomain_0&hash=item27ce1b90f9

It matches the numbers on the relay we have

Sry for bumping my own post, but that project got sidetracked at work for a while.

Just wondering, would this be a good diode to use as a flyback? If not why? Cause I've read a bit and I still can't understand how to properly spec out a diode.

Thanks for the help everyone.