What are the most reliable control relays? SSR's?

[jtruncale]

Hello all,

I work in an oil refinery and we have many small 120VAC control relays such as Potter-Brumfield icecubes all over the plant controlling everything from lights to small MOV's. Most applications do not present a problem when a $50 relay fails, but occasionally we end up shutting down something that costs the plant thousands, if not millions of dollars.

Naturally, most of these failures occur when a relay has been switching too often, say.. once a minute for 6 months. Of course, certain design changes can usually be made to remedy the issue of the constant switching, but I would like to hear from some different folks on what relays they have used in the past (brands, types, models, whatever) that have proven to be the most reliable in their own fields.

I have looked into solid-state relays and they seem like a good solution, but I am not familiar enough with them to confidently recommend them for use in our plant. I understand they can get pretty hot and may have other issue unbeknownst to me.

Any and all suggestions are welcome! Thanks!

 

[Cow]

Try posting your question on this site, I bet they'd have a good idea of what works and what doesn't:

http://www.plctalk.net/qanda/forumdisplay.php?f=2

 

[Jraef]

You cannot make a universal decision, you need to evaluate each incidence of failure on it's own merits.

SSRs are a good choice for high switching duty, but are not suitable for all types of loads and most importantly, they do NOT provide true isolation of circuits (they "leak").

"Ice Cube" realys are really the cheapest thing you can use, but have the advantage of being able to be replaced without tools and very very quickly. Sometimes that's a good thing.

Machine Tool (Industrial) Relays are better in terms of life span, but are more difficult to replace and are physically much larger and more expensive.

 

[jtruncale]

Thanks

Thanks

@ Cow: Thanks for the link. I will give them a try.

@ Jraef:
Thanks for the info on the SSR's! And I completely agree with regard to making universal decisions. In fact, that is a habbit we are trying to break in our plant; the tendancy in the past has been to slap the cheapest minimum solution in and run until failure.

You mentioned the SSR's do not provide true isolation... just to clarify: I understand that the load switching mechanism is a transistor that naturally has some leakage even in the off state, but are you saying that current can leak from the control circuit to the the load circuit through the SSR?

 

[Smart $]

... but are you saying that current can leak from the control circuit to the the load circuit through the SSR?

I believe that is what he is saying... but that is not true of all SSR's. Look into SSR's with optical coupling...

On the other hand, "leaks" can happen from-to any two terminals. It's a matter of whether leaking is normal or faulty operation :roll: and whether the affected system is designed for and-or can tolerate it.

 

[Jraef]

...
You mentioned the SSR's do not provide true isolation... just to clarify: I understand that the load switching mechanism is a transistor that naturally has some leakage even in the off state, but are you saying that current can leak from the control circuit to the the load circuit through the SSR?

Load switch is more often an SCR or triac, it's only a transistor if it's for DC loads (because an SCR doesn't turn off on DC). There is leakage current through SCRs; it's low, but measurable. There are designs where the control and switched circuit are the same, some are isolated optically. The problem is, there are almost as many different kinds and formats of "SSR" as there are electro-mechanical relays, hence my comment on no universal solutions.

 

[Besoeker]

I have looked into solid-state relays and they seem like a good solution, but I am not familiar enough with them to confidently recommend them for use in our plant. I understand they can get pretty hot and may have other issue unbeknownst to me.

Any and all suggestions are welcome! Thanks!

One thing you might light to think about is how many "contacts" you get with a simple SSR.
The11-pin ice cube will give you four changeover contacts. Replicating that with SSR devices might be a bit of a challenge.

 

[CONTROL FREQ]

This is just my 2 cents... I've found (many of you may wonder why) for some reason the 'average maintenance man' has a hard time successfully and confidently trouble shooting SSR's, SCR's and the like. A former employer of mine lost so much money on SSR's being the "root cause" and the off-shift guys' inability to figure it out, that he started replacing them ---ALL OF THEM--- as an annual PM:slaphead:... If you already have ice cubes, I would suggest you stick with them, BUT check out grainger, and buy the omrons for $10 each, instead of paying $50... then, if you want to do what I personally prefer, try out that "pro-active" maintenance approach. Change 100, or 1000 ice cubes as an annual PM. Wether it's $1000, or $10,000----IT AIN'T TENS OF THOUSANDS OR MILLIONS:lol:. They seem to be easy for everyone to troubleshoot, and their cheap, the down side is electronics always faill on a bath tub curve (also the upside IMO), if they last a week, your probably going to be 100% reliable for the other 51 (less any outside influence). You may even find you can go 2 or 3 years... just depends on how critical they are to your ops.

 

[gar]

110914-2008 EDT

jtruncale:

Your application determines what you want to use. This means you need an understanding of the design and characteristics of different relay types, and your loads.

First, I will describe a brutal application where I applied a P&B KUP 11D15 24 VDC coil relay. This application has been in a substantial number of machines in automotive production over the last 40 years. The relays need to be replaced periodically. A good time frame is 3 months, but some have worked for a year. The cycle rate on some may be 3 to 4 times per 24 seconds, or about 6 to 8 per minute.

This relay has contacts rated at 10 A AC and uses silver-cadmium contacts. These contacts are very poor at low voltages, but at 120 V provide better durability than plain silver. In my application I paralleled the DPDT contacts, and the contacts were used as a SPST relay.

This relay switches a highly inductive load fed from DC at 110 V. The steady state current is slightly over 1 A. There is an RC snubber across the load. Our goal was the shortest drop out time of the electromagnetic clutch. Thus, a shunt diode across the load was not feasible.

The failure mode is unidirectional transfer of contact material from one contact to the other. One advantage of not doing the following to lengthen the life was that this gave the plant electricians something to do.

To lengthen the life a second relay could have been added that during the off state of this switching relay the second relay would reverse the direction of current that would flow thru the switching relay.

What happens when switching DC is that on average (meaning always) some contact material transfers from contact A to contact B. This produces a conical cavity in contact A and a conical mound on contact B.

Change to an AC current thru the contact and on average for a randomly operated relay or switch there will be metal flow in both directions.

In the old days when breaker points were switching the DC to an ignition coil there was this one way transfer of contact material.

Filed May 17, 1943 is patent http://www.freepatentsonline.com/2402543.pdf by Henry Ford and Emil Zoerlein for a distributor design that automatically reversed current direction each cycle. I do not believe it was ever put into production. In the early 50s Ford Motor developed a breaker point set with a hole in the center of the contact that grew the mound. This somewhat increased breaker life.

An electromechanical relay that switches moderate voltage AC at a relatively low current will probably fail of mechanical fatigue. At high currents it will probably be contact failure. Contact size and cycle life will determine the relay sized. Lots of AB machine tool relays, back in the days of relay logic, probably had cycle life in the 10,000,000 range. One cycle every 12 seconds is 2,628,000 cycles per year for 24 hours per day. I suspect a lot of machine control panels had some relays that were never changed (repaired or replaced) in 10 years, but not all panels ran 24 hours 365 days per year. However, some relays would operate more that once per 12 second machine cycle.

Now to SSRs. For your type of application almost all will have optical or other high resistance isolation between the control signal and the output switch. The output switch is a transistor or equivalent device for DC circuits. Any appropriate solid state device can be the output power switch for AC loads. The amount of leakage current thru this output switch will vary, but is many times significant, and results from the switch device itself, or shunt circuitry around the switch.

Major failures of solid state switches will result from excessive current, or overvoltage. If operated within reasonable limits the cycle life can be very great compared to an electromechanical relay. Also SSRs may be operated rapidly. Some normally do not turn off when commanded, but wait until the next current zero crossing of the load current.

You need to learn more about the various devices and evaluate the requirements of the load.

.

 

[Besoeker]

This is just my 2 cents... I've found (many of you may wonder why) for some reason the 'average maintenance man' has a hard time successfully and confidently trouble shooting SSR's, SCR's and the like

SSRs I can understand. SCRs almost always fail as a short circuit.

 

[GeorgeB]

First, I will describe a brutal application where I applied a P&B KUP 11D15 24 VDC coil relay. This application has been in a substantial number of machines in automotive production over the last 40 years. The relays need to be replaced periodically. A good time frame is 3 months, but some have worked for a year. The cycle rate on some may be 3 to 4 times per 24 seconds, or about 6 to 8 per minute.

This relay has contacts rated at 10 A AC and uses silver-cadmium contacts. These contacts are very poor at low voltages, but at 120 V provide better durability than plain silver. In my application I paralleled the DPDT contacts, and the contacts were used as a SPST relay.

This relay switches a highly inductive load fed from DC at 110 V. The steady state current is slightly over 1 A.

I've not researched that particular relay, but I would have hooked the contacts in series for DC rather than parallel. Greater path, quicker arc extinguish, less material transfer. We always put contacts in series when available with DC.

 

[petersonra]

A lot depends on the application.

I use a fair number of Siemens 3RT contactors as relays. They are DIN rail mountable and just about indestructible, and the smaller sizes cost less than an 11 pin relay and socket. I have used well in excess of 10k of them. Have yet to have a customer report a failure. The smallest sized one takes about about the same amount of panel space as a socket mounted relay does.

I don't use them for low current applications any more as they don't work real well without some wetting current.

These are for applications that are 28VDC, 270VDC, or 115/230VAC (300-900Hz), for switching capacitors, inductors, or resistive loads, along with mixed loads. Some of the profiles used are as fast as 10X a minute, and sometimes they are used for weeks on end.

Been using them for >10 years now. Indoor in controlled temperatures, also outdoors with ambient temperatures. Zero problems.

 

[jdsmith]

Hello all,

I work in an oil refinery and we have many small 120VAC control relays such as Potter-Brumfield icecubes all over the plant controlling everything from lights to small MOV's. Most applications do not present a problem when a $50 relay fails, but occasionally we end up shutting down something that costs the plant thousands, if not millions of dollars. ...

I work in a refinery as well so I have a good idea what your applications tend to be. We have defined two classes of relays in our specs - general service and critical service.

General service means:
-Failure of the relay would not cause a process unit of even a small part of a process unit to trip offline.
-Failure of the relay would not cause failure of a pollution control device
-Failure of a relay would not prevent MOVs from operating - we don't have a lot of MOVs, but the big high risk ones that come to mind are in the ultraformer unit.
-A relay failure is permitted to trip off a motor that has an installed spare.

Critical service:
-Failure of the relay would cause shutdown of a portion of a process unit
-Faiure of the relay would cause an environmental release
-Failure of the relay would cause a non-spared motor to shutdown, leading to a unit shutdown

Aside from the criticality level, another consideration is the rated number of operations. For a general duty relay that will switch often we would use an industrial relay with contacts rated for 10 million operations to save maintenance costs

For years we used Potter Brumfield KUP series ice cube relays for general duty applications. We have now switched to a Phoenix Contact ice cube relay with 2 SPDT 10A contacts that is physically quite a bit smaller to save panel space. For critical applications we use heavy duty industrial relays, primarily Allen Bradley 700P series and also GE CR120B series. Allen Bradley has the most flexible line of industrial relays by offering reed type contacts, standard 10A contacts, and 20A contacts. All of the contact cartridges in industrial relays are interchangeable and field convertible between NO and NC configurations.

As jraef indicated, industrial relays are much larger, are not very quick to change because all of the wires must be unhanded and remanded, and they are pricey. However, with the problems you indicated and especially the cost of downtime in the refining industry, you will probably be best served by using industrial relays in some of your trouble spots.