how to automate relay testing

[debugtimog]

Hi All,
I would like to ask help to help me design a relay tester, that will buzz or light up if the contacts are,say more than 2 ohms. I dont want to use microprocessor, just a simple circuit, maybe a comparator? I want to switch it several times to power up the relay, and I would just know that the relays is faulty each time I press the power button.

Thanks in advance for your help.

[budm]

Re: how to automate relay testing

First, understanding the test method.
http://www.te.com/commerce/DocumentD...=CS&DocLang=EN

[stj]

Re: how to automate relay testing

you could do it with a microcontroller, a transistor, and a load resistor.

[eccerr0r]

Re: how to automate relay testing

Like any testing situation you'll need a well regulated supply. With a regulated supply, a simple resistor can be used with the supply versus the contacts to determine whether they are better than 2 ohms. With a regulated 3 volt supply, if you try to use the relay to short a 1 ohm resistor, you can just use a comparator to make sure that the voltage swings more than 1 volt to ensure that the contact is 2 ohms or less. Of course you should scale this as worst (best?) case is that the contacts need to sustain 3 amperes of current. Doing this to a tiny reed relay may fry the contacts, but a heavy duty relay may be bad if the contacts have 2 ohms of resistance in it.

But just that -- like transistors, there's tons of different kinds of relays out there. Are you targeting a specific type, brand, etc. of relay? What about the normally closed contact which may have slightly higher resistance?

As they all have different contact strength, different coils, pin outs, throws, etc. it makes automation tough.

[Curious.George]

Re: how to automate relay testing

I would like to ask help to help me design a relay tester, that will buzz or light up if the contacts are,say more than 2 ohms.

That's not a sufficiently well-specified test criteria.

Relay contacts aren't "ideal switches". Their performance will vary depending on the actual operating conditions. E.g., passing a couple of volts through some contacts is usually much less reliable than switching "many" volts. Likewise with current. There are physical phenomena at play (contact shape, oxide buildup on contact surfaces, etc.) that make low voltage/current operation problematic.

This is why you see gold-plated contacts on some offerings.

Or, mercury-wetted.

You need to figure out how the relay(s) will eventually be used in order to be able to design a suitable test environment/criteria for them. Else your test results may be wholly inapplicable.

[goontron]

Re: how to automate relay testing

^ George is right. Without any particular set of standards, or, expected operating environment to model a test around, well, it's not a test.

[Curious.George]

Re: how to automate relay testing

... well, it's not a test.

Oh, it's a TEST, alright... just not one whose results MEAN anything!

By way of example, I had a summer job (college) working for a company that manufactured "hand tools" (think: hammers, screwdrivers, chisels, etc.). One of my duties was "testing" those tools.

Q: How do you "test" a hammer? Or, a screwdriver? Or, a tape rule?

Of course, any manufacturer needs SOME way of testing their product to get an idea as to whether they "quality" is as it should be. At the very least, to raise a flag when "test results" seem to show a decrease in quality/performance.

One of the ways that hammers were tested was to fasten the handle in a jig on a machine that would "wind up" and "bang" the hammer into a fixed plate with a predetermined amount of force. Lather, rinse, repeat The "data" that would obtain from the test was the number of such "blows" that the hammer could tolerate before the handle snapped.

[There were other tests for other aspects of the hammer].

So, the test tells you that (now broken!) hammer was a "372". But, what does that mean to a carpenter or other customer?? <shrug> It's only value was as a RELATIVE measure: if the (sampled!) hammers last week were yielding values in the 420's, then obviously something has gone wrong with the manufacturing process!

But, just how wrong? Does that loss of ~50 blows represent a 12% reduction in quality?

And, because manufacturers don't publish the results of this (bogus?) test, there is no way to know how well your product stands up to the competitors -- short of buying a few dozen of THEIR hammers and testing them on YOUR test rig!

If your hammer yields a better value, should you feel good about this? How do you know that your test is MEANINGFUL to the end user? If the force you are simulating is only attainable by someone with "Hulk"-like strength, then no one will likely make a purchase decision based on that aspect of your product!

[And, given the willingness of folks to buy HF crap, maybe all they are really interested in is purchase price and quality doesn't factor into the calculus!]

[petehall347]

Re: how to automate relay testing

so you want to say something like 5 amps for a relay is guaranteed to 15000 times so real test fails at 45000 times .like rope and chains are more or less tested at ?

[llonen]

Re: how to automate relay testing

I use an oscilloscope and a simple octopus circuit for testing relays, allows me to look at contact quality at varying rates and loads. I will post some details when I get back to the workshop.

[Curious.George]

Re: how to automate relay testing

I use an oscilloscope and a simple octopus circuit for testing relays, allows me to look at contact quality at varying rates and loads. I will post some details when I get back to the workshop.

Any characteristics you are "looking at" on a 'scope are likely not to be compatible with the OP's goal to "design a relay tester, that will buzz or light up if the contacts are,say more than 2 ohms" -- esp with just a "simple circuit".

The OP needs to define what his criteria are for pass/fail before a solution can be proposed, let alone debated for its relative merits!

[sam_sam_sam]

Re: how to automate relay testing

The OP needs to define what his criteria are for pass/fail before a solution can be proposed, let alone debated for its relative merits!

If you were doing this project what would your criteria be can you please provide an example

Thanks

[Curious.George]

Re: how to automate relay testing

If you were doing this project what would your criteria be can you please provide an example

How are the relays intended to be USED? Unless you're the manufacturer of the relays -- trying to establish an initial manufacturing specification and, later, test against it -- you undoubtedly have some particular use in mind. You want to test that they will perform "reliably" (however you define that) in your particular application.

If you're switching high voltages and there is a tendency for arcing as the contacts open (under load), you would be more concerned with the contacts pitting as a result of that than if you were switching low currents at low circuit voltages. Etc.

As you're only taking a peek at the relay(s) at one point in their life (which the OP hasn't identified: are these "pulls" that need to be recertified or suspected failures or new items or...), you have to rely on your own historical records (past performance) to extrapolate how they will fare at some other point in their life.

[llonen]

Re: how to automate relay testing

Mainly I look for degraded contacts which shows up really well on a scope tells me more than measuring the contact resistance itself as this can vary significantly from throw to throw. Essentially using a variable rate relay pulsar poor contacts will show up as excessive bounce and or noise on the trace, I appreciate this method is rather different to a pass no pass solution but its one I developed for testing relays on commercial bar aid washing machines that have about a dozen on each board.

[llonen]

Re: how to automate relay testing

I have come into the workshop 10 Baraid controller boards with a large number of relays on each for testing, so I looked into the way I was testing them again and came up with a 'simpler' method. I am not sure how this got overlooked last time but rather than a crude pulser I got around to using a function generator, square wave output, a variable power supply (100ma) set for the test signal and the scope.

Inject the square wave into the relay coil, set desired frequency and watch the signal developed across the switching contacts from the power supply on the scope. Should be in theory be a square wave, but it will show up contact bounce and poor contacts, use the function generator to vary the test frequency.

When I get around to it I might put some pics up.

[redwire]

Re: how to automate relay testing

Copy pasta from the TE paper:

"Used relays... cannot be reliably tested for contact resistance after installation or use due to the pitting that results from normal switching and the arc residue that accumulates on the surface of the contacts."

"Do not monitor contacts using an ohm-meter or digital logic levels"

"...Used relays (that have been installed or have switched any load current) must be tested for functionality at much higher voltages and currents - typically about 12V, 100 mA (or 500mA)."


So it's 12VDC at 100mA, 500mA preferred for testing contacts.
I wonder if a car light bulb is good enough?
Most cheap bench power supplies in constant-current mode have terrible overshoot and spikes and some can wreck the relay.

[llonen]

Re: how to automate relay testing

Well I could test at a much higher current as the relay contacts are rated at 5amp 30 volts, the object though is not a destructive or damaging test rather to assess contact quality and bounce which clearly shows up even at a 100ma. The function generator I use will generate a signal up-to 20 volts peak to peak which so I can dial in 12volts. I will be working on a batch of these boards so will get some scope pics up of good and failing / faulty relays.

Just use a 120 ohm 1 or 2 watt resistor, that will give you a 100ma draw current at 12 volts.

[llonen]

Re: how to automate relay testing

So here we are then, having worked my way through 10 of these hobart bar aid boards here are some of the resultant relay failures.

So first the test layout, then a good control sample then a selection of bad relays.

[llonen]

Re: how to automate relay testing

And of course not to forget some of the dross, I had 15 final pulls from 10 boards, some tested completely welded closed and some were completely open. the rest tested as intermittent but showed as clear fails when scoped.

[redwire]

Re: how to automate relay testing

Are you saying the debounce scope waveforms tell you pass/fail? (for a relay that still switches, not stuck).
I wasn't sure what scope traces are OK or not for your criterion.

You can also AC couple a loudspeaker and listen to the switching. I had some small SMT signal relays drive me nuts with flaky switching, and a loudspeaker and tap tap tap on the relay housing instantly showed the defective parts. Like a noisy potentiometer scratchy sounding.

Inductive loads with no snubber or MOV eat up relay contacts, like on appliances and heaters.

I found HongFa relays cheap but they work OK.

[llonen]

Re: how to automate relay testing

With the waveform I am feeding into the coil actuator, that being a square wave the resultant switching waveform consisting of a dc 12 supply limited to 100ma or 500ma should look within the limits of the switching arrangements to be a square wave as it is simply switching the relay between two voltage / current states. I find this method good for identifying failing relays where the failure mode is attributed to degraded contact and contact resistance since a simple measure of resistance rarely identify's this unless the seriously degraded.

I used to use an octopus circuit and pulsar but prefer the level of control that a function gen gives me, I can more easily vary test frequency and coil voltage etc. In the case of the hobart controller boards I use the customer fault report along with this testing methodology to verify relays on each board.

I am sure there are many other ways of testing relays but this method seems to work reliably for the work I carry out.