Pages

Thursday 25 March 2021

Relay Life - Tests going forward from 5 Million Cycles


After a cumulative 5 million cycles in total the increase in contact resistance is conclusive. Shown below is the final 3.1 million cycles.


 OK, so I was surprised to lean just how robust these inexpensive Chinese relays are. But I also learned what failure looks like and by using various statistical measures I have concluded that going forward I will:

  1. measure the voltage across the contacts and save only the sum of 4 readings. This gives me just as good a measure of increasing resistance as 8 readings.
  2. Opening and closing times have no strong correlation with contact resistance so I will not be recording them any more.
  3. Measurement will start at the same time reading 6 is currently taken. The average of readings 1-5 are statistically different to the last three readings, but that difference is decreasing i.e. reading 1 is more different than reading 5 when both are compared to reading 8. Without the RC network this effect is still present. So perhaps this is a heating effect of passing current through the contacts?

Which all means the external flash memory chip I used can now records over 2 million cycles. Which is fortunate because my interest is switching to these 2Amp signal relays for use as a 50 watt 6m changeover relay.

 

I will test one of these shortly at a constant 2 Amps using a constant current generator. These should fail relatively quickly at 2Amps. I might then redo the test with the RC network with an expectation they will fail really quickly. But all I need is 10,000 cycles to meet my goals. 

73's

Wednesday 17 March 2021

Relay life - Degradation after 4.7 million cycles

 Just to recap:

  • 1.8 million cycles switching a 2 Amp constant current
  • Added an RC network giving some 10Amps on contact closure falling to 2Amps constant current
  • Ran for a further  1.3 million cycles
  • Got impatient, reduced R to give >10 Amps on contact closure falling to 2Amps constant current
  • Ran a further 1.6 million cycles (Run 6 and onwards below)

 Total cycles now 4.7 million. For the 2.9 million observations with the RC network we see:

 



The analysis of variance by group shows run 11 had an increase in average contact resistance of around 8% and much greater variability. 

 

The 15% increase from run6 to run11 means the relay contacts are dissipating 32% more heat when closed.

I'm going to hit a total of 5 million cycles with the next run. 

73's