Roland CR-78 non-rechargeable Lithum battery replacement mod

Replacing NiCad with Li.

So, I was servicing a CR-78 for a friend and the old leaking battery had already been removed when I received the unit. After I was done with replacing all the damaged components I was going to put a non-rechargeable battery in there. It's probably possible to replace the NiCad with a NiMH battery, but they recharge slower than the NiCad (so you need to keep your machine running longer and more often to keep it charged) and I'm not sure exactly how they work when they reach their full charge.
As I'm not an expert in rechargeable batteries, I decided that the safest option was to use a non-rechargeable Lithium battery.



For the short story and how-to, scroll to the bottom of this wall of text.


NOTE:
As I was servicing this particular unit due to extreme NiCad electrolyte leakage corrosion, I had removed a lot of the original components.
D109 had already been removed when I got it, as recommended in the service manual to increase the charge current to the NiCad battery.

The parts relevant for the RAM backup circuitry that I had removed are:
R237, D111, D112, D221, C115
 I replaced D111 with a 1N4148
Also, C115 was replaced with a Nichicon KL-series low-leakage current capacitor (UKL1C470KEDANA), because it's kept charged by the battery when mains power is switched off.
The KL series have a leakage current of either 0.2uA, or 0.002CV of rated capacitance and voltage, whichever is greater.
For this particular capacitor (47uA, 16V) that would make it 0.002*0.000047*16 = 1.5uA

3V or 3.6V?

The RAM is speced to hold the contents down to as low as 2V Vcc.
Most Lithium batteries are 3V, but as there is a diode after the battery that will drop the voltage a bit, I think it's safer to choose a 3.6V battery if possible.
In the end I bought a Xeno brand 3.6V 1200mAh lithium-thionyl chloride battery with solder terminals.

How long will it last?

OK, this section is a bit of a guess, but I did a test with a bench power supply and connected to where the battery terminals should be connected. I then tried to measure the current using a couple of multimeters.
I really really wish I had a proper microampere meter...
After the current had settled (C115 needs to charge), I could read 0.05uA, so 50nA, which is about the minimum value my meter could show.
Considering the quality of my instruments I think there's like 1000% tolerance on that reading :)

I looked at some datasheets for the 5101 RAM and it said maximum stand-by current consumption is 10uA. Another datasheet said typical current consumption @2V = 0.14uA

Let's assume the worst: 10uA per chip

1. We have 2 ICs so that will give us 2*10uA=20uA.
2. The replaced (see above) capacitor C115 leakage current is speced to around 1.5uA.
3. Diode D110 is the original 1S1588 with a speced reverse leakage current of 0.5uA (at 30V...)

Summing the three together gives us 20uA+1.5uA+0.5uA = 22uA
Given the battery capacity of 1.2Ah we get:
1.2Ah / 0.000022A = 54545h = 2272 days = approx 6 years and 3 months, worst case.

Let's assume typical current consumptions

Typical 5101 stand-by current consumption according one datasheet claims 0.14uA.
The Nichicon capacitor leakage current is probably around the same, but let's make it half as much: 0.75uA.
The max reverse current of the 1S1588 was speced at 30V.
I think we can assume it's lower at 3V. Let's say it's also half as much, so: 0.25uA.
I know... A lot of assumptions, but what do you do?

1. We have 2 ICs so that will give us 2*0.14uA = 0.28uA
2. C115  may have lower leakage since it's not charged to full capacity. Assume half, so: 0.75uA
3. 1S1588 max leakage current at 30V was 0.5uA, but let's assume that at 3V it's half: 0.25uA.

Summing the three together gives us 0.28uA+0.75uA+0.25uA = 1.28uA
Given the battery capacity of 1.2Ah we get:
1.2Ah / 0.00000128A = 937500h = 39062 days = approx 107 years (in a perfect world).
That's a ridiculous number.

In any case, I think it's safe to say that the battery will last AT LEAST 6 years.
It'll most likely die due to self-discharge before being drained...

So, what did it look like in the end?

It looked like this, with the schematic and original layout as reference:

Battely :)

Note: I hadn't removed R161 when I was cleaning the PCB. Once I started to look at the RAM backup battery replacement, I didn't bother removing R161 as it didn't make any difference, since D109 and D112 were already removed. But, if you want to remove any unused components, you could remove that too. That's why I marked it with a red X in the photo.

The main goal is to prevent the battery from being charged, while still allowing it to supply power to the RAM ICs.
To stop the battery from being charged you have to cut connection to the power supply coming from +5V via R237 and D221. You also have to cut the connection to the +15V coming via R161 and D112.

So you need to remove:
R237 and/or D221
as well as
R161 and/or D112

If you want to be sure, remove the RAM ICs and check the voltage coming through to them both with and without the battery in place, both with the power switched on and off.
It should never be more than 5V and never less than 2V (preferably a bit more).


Here's a more complete part of the photo showing the battery as well:

 

Enough blabla. What do I have to do if I want to do this myself?

Simplest: Remove R161 and R237.

More thorough: Remove R161, R237, D221, D112 (and D109 if it's not already gone).

Replace NiCad battery with 3.6V Lithium battery.


Hope this may be of help to others who are thinking about replacing the battery in their CR-78.

Disclaimer: I take no responsibility for any damage caused by errors in this procedure or the description :)