Outrun saves some rudimentary data (credits, time on etc.) on the main RAM at IC 129 and IC 114 that is powered using a supercapacitor when AC power is off. The enhanced ROMset saves even more data and settings on the main RAM so it is important to retain those as long as possible so you don't have to reset them. For that ROMset, it is recommended to change the supercapacitor from 0.1F (original ) to 1F.
However, how long is the data retained from the original supercapactior vs. a higher capacity one?
I decided to figure this out and see what gains there are from switching to a 1F or higher supercapacitor.
First, I had to determine
what is the minimum voltage required to retain the data. For this, I used an OutRun board with the original capacitor and let it run for a few minutes, enough to charge the original supercapacitor and r
ecord a power ON time (as shown in TEST mode). I then disconnected power (but kept it plugged to the harness) and added a 100 Ohm resistor across the Vcc and GND terminals of the main RAM at IC 124 so that it will bleed voltage. Using a DMM connected in parallel to the RAM, I could measure voltage. I added a switch in series to the resistor that would allow me to enable/disable the bleed resistor.
The original voltage to the RAM, even with the power ON at 5V, was less than 5V, at 3.5. By bleeding voltage to predefined values and testing if the power ON time was retained (in TEST mode) I could determine what the minimum voltage for data retention was. The graph of multiple measurements below shows that you need at least 0.45V across the Vcc and GND of the RAM to retain the data.
Now that I knew the mimimum voltage for data retention, I set out the determine how long it takes to reach minimum voltage on the original 0.1F supercapacitor.
This is determined by the equation of t=RC*ln(Vo/V), where R is resistance (Ohms), C is capacitance (Farads), Vo is initial voltage and V is final voltage (In Volts). While I knew the values for C, Vo and V, I did not know the value of R. As a matter of fact, resistance at the Vcc and GND of the RAM was large and I could not measure it. So, in order to get obtain the t to 0.45V, I would have to experimentally determine it.
So, as a
second step, I used the same OutRun board that I played for 30min in or so. After a day (give or take a few hours), I took voltage measurements at the RAM IC 124. From the data and the equation above, asumming a t=86,400sec (24hours), I could also calculate R and even current (I, in Amps but shown in microA below).
The data are shown below:


Voltage and current decays were as expected. Resistance was little wobbly in the 3-9M Ohms, with an average of 5.22M Ohm. I think the wobbliness was due to me not taking measurements at exactly 24hs every day and having very small voltage values (and likely with high error in measurements) as time went on.
Initial voltage (Vo) in this setting was 3.54V. As you can see, it
takes 10-11 days to go from Vo to V of 0.45V. Using the t=RC*ln(Vo/V) equation, C=0.1F and using the average R of 5.22M Ohm, we get t=12.3 days, close enough to the experimentally determined value (10-11 days).
Since I determined the R value, using a 1F capacitor would allow the data to be retained for ~124 days (or 4 months). Using a 1.5F capacitor would extend this to ~186 days (or 6 months).
While I will need to get some experimental data with a 1F supercapacitor (I don' have any handy), at least to make sure the voltage decay follows the same pattern, I think we can conclude that:
0.1 F original capacitor: data retained 10-11 days
1F supercapacitor: projected data retention ~ 4 months
1.5F supercapacitor: projected data retention ~ 6 months
While I have
FRAM as the main RAM on my board (no need for supercapacitor, projected 150+ years of data retention), I think that 4-6 months is pretty good data retention with a simple component change!
This was a good experiment, I had fun.
p