Solid State Relay Endurance

Thanks for the heads up on that Jack.

I will definitely see if I can simulate a crimp failure. I am thinking that a short tail of N80 crimped between terminal and wire tail will be a good start.

We have seen crimp failures in the field, but they are very rare.

These controllers often run 24/7 for weeks at a spell, usually they see a 30°C ambient, perhaps a little more, but some field units seemed to be getting a bit warmer than we could ever get them in our testing. Not failing, but enough to warrant a deeper dig. Hence the cook-up.

I am astonished, but pleased that the SSR has taken this much punishment without failing. My prediction, made after the first week at 50°C and having just cranked it up to 70°C, was that it would expire within 24 hours. Very confidence-inspiring.

I would add that short term tests do not replicate aging and thermal cycling effects.

The heat-sink compound do ages. Some brands more than others. It can harden when subjected to high temperature or be pumped out when fresh if frequent thermal cycles are experienced.

Do you use the same thermal compound that production use? Do they all know how to apply it? Maybe the "new" assembler has not been trained properly.

Yes, a very good point on the crimping of the wires to connectors and pins. Loose connections make for unanticipated hot spots.

I would also examine the strain relief and mechanical dressing of these overheating hookup(?) wires. Remember, with hookup wires the metal conductor is the only internal mechanical support. Consistent vibration of hookup wires during shipping and operation can induce metal fatigue breaks in the conductor. This is more obvious with the complete failure of solid or single strand conductors but in multi-strand wires this failure mode will only appear as extra self heating of the wire.

It could be a bad batch of relays...

I would expect at high heat an SCR would fail in turning OFF and likely closer to maximum currents. I doubt your 1750W heater uses 70 VAC so if you were using 120 VAC you were only just over half the maximum RMS current.

At this hour, my solid state textbooks will remain on the shelf.

You state that some internal components,such as heat shrink and PVC wiring insulation,you did not mention any failures internal to the SSR.

If it only the insulation that concerns you,then upgrade to a higher temp insulation,such as fiberglass,etc.

However, if it is really an internal active component,and the application is frequently switched or modulated you need to duplicate the same conditions to give valid results.

Are the field failures a result of the input or output failure?

Triac's and SCR's require power to be applied before the gate is activated.

A weak or out of spec internal capacitor could cause failure of the Triac by allowing the trigger to gate prematurely.

As previously mentioned,I would test the SSR under a switching load.

Yeah, uncertainty is a +/- thing. I doubt if you conduct the test with not just one but 10, probably some will fail or may be not.

You must have taken a greater design factor that's why.

The main purpose is to make it close to fail as you designed it. Otherwise, it might cause too much for the purpose which is bad for bean counters and your company's business as a whole.

Perhaps after taking some of the good advice suggested in the comments here (switching loads, bad crimps, meaningful sample size, etc), if you become very confident about the durability of your controller in high heat, consider making a video (for marketing or just bragging, either way sent a copy to us) in which you use the backplate of the controller to fry an egg...each day for a week.

Your test results are no surprise.

Automotive components with similar heat ratings are typically "design verified" by testing at up to 80 Deg C ambient and also down to -40. Storage temp ranges even higher/lower.

Failure usually happens as the result of multiple simultaneous deviations. For instance, your "test" might actually be cooling the device by sinking heat away. Testing ovens sometimes have air flow that can assist the DUT by blowing heat off the item.

The manufacturer's max temp has to consider all applications. Encapsulant coating of the board and then installed inside a closed cabinet could have device surface temps far greater than you imagine.

A separate test might be to insulate the DUT and monitor its temp as you increase the loading. This will give you some indication of the capability of the relater insulation and nearby items you have observed failing.

Typically as experienced in the field, SSR failures that are heat related starts to be more common is when the Thermal compound between the heatsink and the device started to deteriorate and dries up...

Most of the time the poor conduction of heat is the root cause..

We carried out a series of tests last year to achieve EMC compliance for our SSRs and the test rig got so hot that it melted the label off the SSR and heated the test room so much that the doors got jammed with the expansion. SSRs are pretty tough components, much better than electromechanical relays which tend to burn out contacts and sometimes explode - not a good thing when they are installed in a petrol pump.

The only drawback with standard SSRs is they produce a lot of conducted noise - well above the guidelines in the EMC Directive and you have to add line filters which tend to be expensive and problematic. This is why we now make EMC compliant SSRs which produce hardly any noise in use.

I could not read/understand the graphs. But...

Kinda looks like temperature *may* not be the item associated with early failures.

You may want to look at other factors... Humidity? Airborne pollutants? Particulates?

I've been using/contributing to CR4 for almost 13 years, and I continue to be amazed and highly disappointed that CR4 does not provide some means of attaching higher-resolution images.

I can't read any of the text on your graphs, even after enlarging and using a loupe. Can you please provide a text version of the legend for your graph, so we can have some idea of the meaning thereof?

Legend is:

Red Trace (top) = SSR temp, measured between the mounting screws with a thermocouple stuck to the back panel(1.5mm s/steel).

The next two traces Orange and Yellow are in the controller under the controller PCB and in the wiring loom.

The green trace is the oven temp.

The lowest trace Blue is test room ambient.

So the SSR spent a week with the measured temp at a bit over 115°C, followed by a second with it mostly over 120°C. There is a 2-day spell just after the start of week 2 where the PID control was used on a 1-second switching cycle to give this function some exercise.

The load was a 230V load so line current was 9.7A.

Yes, the SSR is overframed but they are cheap enough that using them seems like good insurance against unreliability. Especially if they can take this level of abuse - far more than would be possible in the field where most of our experience with this controller has been acquired.

You state in your OP:"(heatshrink, PVC wire insulation) were degrading due to heat."

If this is the only problem and not an internal component failure then as others have stated,the likely cause is a high resistance connection(crimp or other connection inside of the heat shrink.An undersized conductor can also result in excessive heat,especially at junctions.

I recommend properly sized THHN conductors with a 90C rating.

A #14AWG copper minimum is recommended.Aluminum conductors are seldom used anymore in small gauges because they are problematic and require special splicing connectors when connecting to copper.

I may be preaching to the choir,but below is a link that explains the variables that can cause conductors to overheat.

I have used SSR's for over 30 years with no problems,even with high inductance loads,such as motors,etc.

https://iaeimagazine.org/magazine/2000/03/16/wire-temperature-ratings-and-terminations/

Ask the user to provide a sample of the wire with the crimp or junction attached and make certain that the proper crimp tool was used.I have seen many instances where the crimp was done with lineman pliers instead of a crimp tool and gave problems with overheating.

This is an example of an electromechanical relay fitted in supermarket petrol pumps - 30A relay controlling a 3/4hp motor! I have replaced hundreds of these and they all show serious damage to the contacts! Downtime costs, (loss of sales) £10k/day.

SSRs do break down, (usually one of the back-to-back thyristors goes short circuit), but I have only ever seen one that 'blew up' so the reliability is far superior.

By comparison this is a solid state relay undergoing EMC testing - yes 'components' ARE required to comply with EMC Directive 2014/30/EU, (at least in this neck of the woods) and definitely in sensitive environments.

The small single phase SSR puck in the lower centre of the picture will switch all 10 of those 2.5kw hot plates without any problem, (apart from wrecking the test room with the heat).

Always do a carriage return before inserting an image, so the image starts at the left. Once the image appears, click on it to select it, then drag a right corner of the image to the right side of the window to enlarge it to usable size.

Also, on that first image, use your image editing software to lighten the shadows. That's easy to do in Photos on a Mac, I presume there is similar software on PCs.

Thanks for that - not the easiest way to submit pictures on this site and I tried every which way to do it, but thought they might be of general interest.

Tom

Hello, you may have provided another clue to the concern.

In use, the controllers seem to be intended for pump (motors) and such. Testing with a resistive load, especially when switching can provide significantly different results.

Switching incandescent headlamps (DC 21V) provides inrush currents typically 7x the SS condition. Heater elements seldom have the same characteristics as the load experienced in the field.

EMC from switching of heaters could be significantly different from switching motors.

One of our Customers commissioned an independent test house to carry out comparison EMC tests using several solid state relays loaded with resistive and inductive, (6 electric motors) loads. The only relay which successfully passed the conducted emissions test was the one we manufacture, but it was interesting to note that the inductive load tests produced less noise than the resistive so that's the test we now use for certification. All waveforms are available if anyone is interested?

Inrush current is less of a problem with thyristor SSRs which can take 1,200A surges, but you have to be careful when selecting line fuses and circuit breakers. Surges such as this can occur in localised generator situations such as on ships.

With inductive loads we have found that the worst case is when switching off at peak waveform. The back emf causes arcing across contacts in electromagnetic relays and contactors, but SSRs seem to take the surges with no apparent damage.

Do you guys not use inter-phase contacts dealing from low to high voltage/power lines, instead use SSR? I don't know if that's a good practice.

I am gladly will be hearing any objection.

This comment appears to be totally unrelated to anything being discussed here, what exactly are you talking about?

I didn't see anyone ask about the brand name of the SSR.

FOTEK has so many China vendors that make counterfeit copies, as well other brand names.

I had a failure of a 25amp SSR powering a radiant heater of 10amp. The TRIAC was not bonded to the back shell of the SSR. This is inside a potted assembly.

My search back then I found a guy who had an eBay auction to advertise this fact of counterfeit SSRs he experienced.

Forgive me if this issue was brought up as a possible root cause.