Recent thread about SSR vs mechanical relay got me thinking. I have been replacing mechanical contactors for 40 years and find they die at about 250,000 cycles from pitting of the contacts. Apparently a solid state relay has an estimated lifetime of infinity because it has no contacts.

On the other hand, I've replaced 3 or 4 mercury bulb thermostats in 40 years but I've replaced 3 or 4 digital thermostats in ONE year and they aren't the most commonly used thermostats...yet.

I think this bears discussion in this chat forum. Why should I believe an SSR has an estimated lifetime far in excess of a mechanical relay when SS thermostats have such a miserable track record?

Anybody with real world experience?

 

If a solid-state circuit is used comfortably within its operating limits it will operate for an indefinite time. But if the device is used close to or exceeding it's limits than it can fail in a short time period. Also there are other components (such as electrolytic capacitors) that often have a higher failure rate than the semiconductors in the circuit.

I've had SS thermostats that have operated for years without a problem. You have apparently had experience with some poorly designed ones.

An SSR should have a very long lifetime as long as it is used within its voltage and current ratings, and is connected to a proper heat sink to keeps its operating temperature below its rating at the highest ambient temperature it will see. (SS relays do dissipate some power due to the current through them and their forward voltage drop. Mechanical relays have a much lower drop.)

 

I've had experience with all sorts of SS thermostats, poorly designed and not. HVAC is my day job. Point is, it's not a personal problem, it's a survey of real world experience. I'd love to hear from somebody that perhaps works at a factory where hundreds of SSR's are used to run motors on a production line basis. That kind of experience would be significant, even if it only confirmed that proper sizing and heat sinking results in very good longevity.

 

The action of a relay contacts & a refrigeration mechanical thermostat are quite different. In a relay the contacts open & shut in a straight line. Mecanical thermostats the contacts have a slight wiping action as they open & close. My current fridge has an electronic control board in it for temp & defrost, its switching is done by small relays, its now about 5ys old, dont know how mutch longer it will last. I have seen old fridges with mechanical thermostats go for 20Yrs or more.

 

It's a myth promulgated in the early days of solid state that SS devices have an indefinite lifetime.

It is even stranger, considering that solid state diffusion was (is) used to manufacture such devices, that anyone believed this.

Of course we know better today, don't we?

They do have a long lifetime but modern devices are under attack from another direction.
As currents diminish, random charge fluctuations and incoming radiation has an increasingly deleterious effect.

SS devices do run cooler than thermionic devices, but not necessarily than mechanical devices, and this does lead to a long life provided they are not run near their limits.

These limits generate another vulnerability as compared to thermionic or mechanical devices. SS devices are vulnerable to overvoltage and many SS services fail due to fluctuating mains or other supplies.

 

It's a myth promulgated in the early days of solid state that SS devices have an indefinite lifetime.

It is even stranger, considering that solid state diffusion was (is) used to manufacture such devices, that anyone believed this.
.......................

They do have an indefinite lifetime (although certainly not infinite), meaning there is no defined wear-out mechanism for the devices that gives a predictable lifetime estimate.

The diffusion used to make solid-state devices occurs at a very high temperature. The diffusion rate at the device's normal operating temperature is so low that it is unlikely to affect the device's operation for any practical time period that the device is likely to be used.

 

I personally think that RoHS is a huge train wreck in the making. I'm seeing lots of transient shorts in new power electronics that I think are from 'Whiskers'.

http://www.hlinstruments.com/RoHS_articles/Modern%20Marvels%20-%20Engineering%20Disasters%2019%20Tin%20Whiskers.wmv
http://nepp.nasa.gov/whisker/index.html

http://en.wikipedia.org/wiki/Whisker_(metallurgy)

 

studiot reminds me of the early transistorized televisions. The myth of indestructable transistors was so strong that the training and repair manuals NEVER came to the conclusion that a transistor should be replaced. In the first year of working on those TV's, none was ever repaired in the field because following the instructions always resulted in an infinite loop that never landed on a bad transistor. We weren't even issued transistors as truck stock!

Two years later I was running a TV repair shop and 70% of the repairs were transistors or IC's. Magnavox had a particularly hot video amplifier IC in a TO-5 package but they were nice enough to mount it on a breakout board so it could be replaced without soldering. A lot can change in 2 years.

 

meaning there is no defined wear-out mechanism

Even at absolute zero only perfect crystals posses no vibratory movement and no tendency to undergo internal diffusion.
And don't forget #12's comment as to how hot some semiconductor components run.

 

Even at absolute zero only perfect crystals posses no vibratory movement and no tendency to undergo internal diffusion.
And don't forget #12's comment as to how hot some semiconductor components run.

The impurity diffusion process in a semiconductor manufacturing takes place at a temperature of 1000°C or more and, even at that temperature, it takes many minutes to hours to generate the desired diffusion depth.

Since the diffusion process is exponentially proportional to temperature it takes a very long time for any significant diffusion to occur at temperatures near room ambient, even at the maximum Si operating junction temperature of 125°C. This reference states that at room temperature an impurity atom will move to the next position at the rate of about 1 jump every 1e50 years (for comparison the estimated age of the universe is "only" about 13e9 years). So obviously, even at an elevated operating temperature the diffusion impurities in a semiconductor are essentially locked in place for any practical time period and thus are not part of any wear-out mechanism.

 

1e50 years sounds about right to me.

 

This reference............

Yes indeed your reference is a bit of physics about the theory of perfect lattices.

Real semiconductor chips are not perfect lattices, although manufacturers strive for the best and they are gettting better.

Further the reference does not take into account the working environment of the chip - radiation, electric magnetic fields, etc.

How fast does a dislocation move?
How fast does a dendrite grow?
Very much faster than Ficks law.

Here is an article from CISCO about semiconductor aging. Google has many more. The problem has become sufficiently acute to spawn standardised testing.

http://www.cisco.com/web/about/ac50/ac207/crc_new/university/RFP/rfp11068.html

........NBTI (Negative Bias Temperature Insatiability) & PBTI (Positive BTI): NBTI aging degrades the performance of P-channel devices and has become a significant reliability and yield concerns since 90 nm silicon technology.................

 

a survey of real world experience. I'd love to hear from somebody that perhaps works at a factory where hundreds of SSR's are used to run motors on a production line basis.

There are hundreds or thousands of SSRs in my plant. I've replaced 3 in > 2 years. One was failed shorted, another was undersized and improperly heat sinked and failed twice, accounting for 2 of the 3. They are used almost exclusively for heaters. Actually, I can't remember ever seeing one switching a motor. Motor controls have been exclusively VFDs or mechanical contactors in my experience. I assume because of the inductive load, & high start-up current draw, a dumb SSR is a bad choice. There are smart solid state starters for motor control, which are decked out in features; overvoltage protection, undervoltage protection, phase loss, and probably a dozen other things I can't think of. A while back we had a motor fire after a contactor welded 2 of it's contacts shut and single-phased a motor. I was tasked to find a solution that we could implement plant-wide that would prevent this happening again. That's when I learned about these solid state starters; after reading the details on them, even they use mechainical contacts and are not immune to the same failure mode. My assignment is still pending....