Anyone have any SOC vs voltage data for button cells? Li Manganese oxide

[viper]

So I have a piece of equipment that has volatile memory and if it is lost, the machine is a total.  The OEM plays games where you must send the equipment in to have batteries replaced and reload machine.  Now machine is not supported so......we do what we must.  Which means, we hot swapped a soldered pair of button cells in the machine years ago.  Actually 8 long years ago!  I sort of forgot!  I think the OEM specified sending machine in every 5yrs or so.  However, I recall selecting bigger button cells to kick the can down the road as much as possible. 

Somewhere I have the original voltages but as of today, the current CR2450s are at 3.15V and 3.05V.  What is frustrating is Li has such a flat discharge curve, I am trying to assess where I stand.  Because this process requires 3 people doing various things like holding PCBs with cables while keeping everything plugged in and soldering batts, I don't really want to have to do this more than needed.  It is risky as one wrong move and the machine is toast. 

So...to the point now, does anyone have any data on these cells and their voltage relative to SOC?  I could not even know what the load is on the batts, but it can't be much to be over 3V at 8yrs. 

[Kleinstein]

With the Li cells the voltage and low or no load does not change much with charge state. Even when essentially empty they can still give >3 V but the internal resistance goes up a lot. I have looks at some 10 empty CR2032 cells (mix of manufacturers). They still have > 3 V , but the internal resistance gets really high and I have not found a strong correlation between the residual voltage and internal resistance. So cells than can deliver less current can still have the higher no load voltage. So I am afraid the voltage alone is not sufficient. It may take a current pulse (towards some 2.9 or 3.3 V) to tell.

[amyk]

I would figure out how to save the calibration data or whatever else important is stored in the battery-backed RAM than constantly worry about when and how to replace the batteries.

[viper]

I would figure out how to save the calibration data or whatever else important is stored in the battery-backed RAM than constantly worry about when and how to replace the batteries.

I really wish that were the case but it's not just the calibrations, it's everything.  Software, settings, etc.  Many have tried several techniques to retrieve the data or make a mirror, but I don't think there is any success in the wild. 

What we did last round was keep machine plugged in, just like we are supposed to do with a CNC machine, and swap batteries then, because the system is being supported by that power source.  However, these batteries are soldered in, so it gets tricky. 

The reality is I will probably only do this one more time, then the machine will likely just get retired.  Technically it's already obsolete but a new one is $50k so.....  I think I'll deal with it.  For something I use 1-2x/yr.