25-year old JNIOR and the battery is not dead

[Bruce Cloutier]

The question remains as to what combination of unbelievable circumstances has to occur for a 3V primary lithium coin cell to remain above 3V while being in service for 23 years. Those cells have a very flat discharge curve and remain at 3V for a very long time and until they don't. At which point they fall off fairly quickly. So you have very little warning if you are measuring voltage watching for impending doom. The rated "shelf life" of 10 years doesn't mean that you get 10 years in service. If I had to guess I suspect that we are seeing 6 to 8 years on average in the Series 4. Of course, that battery is replaceable and there is no permanent destruction that occurs while the battery is out.

Obviously 23 years is possible. That is 3 or more times what I would say was the average. Pretty amazing.

But I don't think it is tapping the vacuum energy.

[Harold Hallikainen]

From https://share.google/aimode/gLTA840wj8017kkQv :

In a standard Real-Time Clock (RTC) application operating at a stable room temperature (~20°C to 25°C) with an ultra-low continuous draw (e.g., 0.5 µA to 1.5 µA), a high-quality primary lithium coin cell (like a CR2032) does not deplete due to electrical load alone. Instead, the operational percentage over time is dictated by a race between chemical self-discharge and physical component degradation (primarily electrolyte evaporation and seal wear). [1, 2]
Based on industry reliability models (such as the Analog Devices / Maxim Lifetime Prediction Model), the survival rate of a population of premium cells degrades across 5-year intervals:

0 Years	100%	Baseline / brand-new deployment.
5 Years	98% – 99%	Infrequent infant mortality or manufacturing defects.
10 Years	90% – 95%	Standard shelf-life expiration; worst-performing batches face high self-discharge.
15 Years	40% – 60%	Accelerated electrolyte evaporation through the crimp seal.
20 Years	10% – 20%	Complete dry-out of electrolyte or mechanical seal breakdown.
25 Years	1% – 5%	Statistical anomalies; only pristine cells in cold/dry microclimates survive.
30 Years	~0%	Total chemical and structural exhaustion.

---

Key Drivers Behind the Curve

• The 10-Year Turning Point: Up to year 10, the continuous current draw of an RTC is so minuscule that the cell easily maintains its nominal voltage (>2.8V). The drop-off after Year 10 is almost entirely due to mechanical degradation, not the electrical load. [1, 2, 3]
• Electrolyte Evaporation: The organic solvent inside the cell slowly permeates the structural plastic grommet over time. Once enough electrolyte escapes, internal resistance spikes exponentially, and the voltage collapses instantly under even a nano-amp load. [2, 4]
• The BR vs. CR Variance: If your design utilizes Poly-carbonmonoxide Lithium (BR series) cells instead of Manganese Dioxide (CR series), the curve shifts slightly to the right. BR cells possess a more resilient chemical composition that tolerates longer exposures to high heat, helping stretch the 15-to-20-year survival rates if environmental conditions fluctuate. [5, 6]

Would you like to explore hardware mitigation strategies to extend this window—such as using dual-diode isolated backup circuits or transitioning to a supercapacitor-backed RTC architecture? [5]
[1] https://forum.allaboutcircuits.com
[2] https://www.analog.com
[3] https://vonikoshop.com
[4] https://www.7x24exchange.org
[5] https://www.analog.com
[6] https://www.analog.com


On another issue with coin cells, we found that the contact between the cell and the cell holder where the cells are replaceable deteriorates with low current (in the 10 uA area). This caused a bunch of lost certificates in the USL IMB and caused Dolby to go to soldered in cells. I remember that we looked at applying a surge "sealing current" when we switched to battery power. This high current might be used to charge a capacitor, but the capacitor leakage current was too high to be left connected. Similarly, there was an issue with low current through contacts on tamper switches.

[Bruce Cloutier]

It is a little more complex in my cases in that one or more low power (LL) SRAM are maintained by the cell. If you use the maximum standby current ratings of those components the battery would deplete in less than a year. Those ratings are useless. Meanwhile the 'Typical' rating in the table is all cases is blank.

I did a measurement on another module from the era this morning. I get a draw of about 0.3 microamps. But don't trust that. I do not have an electrometer handy (or at all I don't think). That reading is 1/100th the maximum rating of one SRAM. That makes 23 years a little more feasible but doesn't explain why only one JNIOR-A out a the few I have still has a battery with nominal voltage.

There was a case when I worked at Xerox now 46 years back, when I designed a circuit and outlined its performance for publication. My boss had wanted to use a more conservative figure and doubled mine. When the circuit came out in the final technical report that figure was 8 times! So two additional levels of management beyond that had to have their say. Stupid.