Sunday, July 6, 2014

Juno-60 battery replaced but draining too fast

Juno-60 battery replaced but draining too fast

Replaced NiCd battery with lithium cell

TL;DR? Scroll down to the "Fix" header at the bottom to see how I solved it.

So this is one of those things I regret not documenting. I actually did this back in 2011 but I was too lazy to document it and of course too lazy to write any post about it.
It's not really a complicated process so I'll just describe it with words.
You see the old battery in a previous post about the Juno-60 here from May 2011, though I didn't put a new battery in at that time.

In October 2011, I just took a small stripboard, placed a CR2450 holder on it and attached it to the inside of the Juno case using a screw. I decided to use CR2450 instead of the usual CR2032 due to the increased capacity. I used two wires to connect it to where the battery pins should be soldered.
This worked well and I was able to save sounds again. Yay!

Battery draining too fast

...or maybe it didn't work that well.


According to my notes (I did make a note of the date, even if I didn't document the process) I installed the CR2450 in October 2011 and in February 2013 the battery was dead. That's 16 months. Is that reasonable? Probably not...
I put a new CR2450 in and for some reason hoped it was just a bad battery.
In March this year (2014) it was dead again, only about 11 months later this time.
This really needs to be investigated.

Time to look at some numbers...

What is the battery supplying current to when the Juno is powered off?

Current goes to the RAM, which consists of two TC5514AP-3 4kbit SRAM (IC 11 and IC12).
The battery also powers a TC40H000P quad NAND (IC20), which provides the /CS signal to the RAM, thus keeping it in stand-by mode.

According to datasheets, the TC5514AP has stand-by current consumption of: Typical 0.05uA. Max 20uA.
The TC40H000P has quiescent supply current of: Typical 0.001uA. Max 2uA.

So, totally we have an expected total battery drain current of:
Typical 2*0.05 + 0.001 = 0.101uA.
Max 2*20 + 2 = 42uA

That's quite a difference between max and typical, though you'd have to be really unlucky to get the max current on a device, and even more if you get it on all of them.
In any case, let's see for how long the factory installed battery is expected to last.

The original battery according to the service manual is a Lithium CR1/3N. I found the specs for Varta CR1/3N which claims a capacity of 170mAh. How long would it last?
Typical 170mAh/0.000101mA = 1683168h = 70132d = 192 years
Max (or min for battery life) 170/0.042 = 4047h = 168d = 5.5 months

So under normal circumstances the battery should last until it self-discharges from aging. Seems reasonable.

The CR2450 I installed should have a capacity of 540mAh which would give:
Typical 540/0.000101 = 5346534h = 222772d = 610 years
Max (or min for battery life) 540/0.042 = 12857h = 535d = 1.5 years

This looks awfully a lot like the max current consumption case...
Time to find out where all that current actually goes.

Investigation

I removed the battery and hooked up my bench PSU and put my multimeter in series to measure current.
I measured an amazing 90uA current draw! That's not even worst-case (42uA). It's just wrong.
What is causing this???

After the battery, there's a diode (D21) preventing the battery from charging, but it's just in series and can't "consume" any current. After D21, there's a capacitor to ground (C15). Mhmmm... Leaky cap?
I lifted one of C15's legs off the board, but there was no change in the current consumption. Bah...

OK, so it's down to the RAM and the NAND.
I removed the first RAM and the current dropped to around 85uA. Hmm?
I also removed the second RAM and the current jumped back up to 90uA. Eh...
It's gotta be the NAND then. I removed it.. and... still 90uA. What?!
With all the ICs powered by the battery, something was still drawing 90uA.

About 30 minutes later, I realized there's one more component in the circuit: D26.
D26 is the diode that prevents the battery from providing power to everything else on the +5V rail.
I lifted one leg off the PCB and... ZERO current draw from the battery!!
Now is a good time to look at the photos from my previous post again.
Look for D26. Location is in the text below the photos.

D26 is in the top center, right of the two ceramics, by the PCB edge.

D26 is in the top left corner
I didn't really clean up around D26 when I replaced the ceramics back in 2011. I guess it was current leaking through the nasty gunk underneath it.
So now I cleaned up with some isopropyl alcohol and removed all the gunk and soldered it back.
I soldered sockets in for the NAND and RAM ICs and put them back in.
After checking the current with everything back in place it showed near zero current (my multimeter can't reliably show nA numbers, but at least it was less than 1uA).
... for a while ...
Suddenly the current jumped back up to around 90uA. What?!
I removed the ICs from the sockets but there was no change.
So, it wasn't current leaking in the nasty stuff under D26. How much could I expect the diode to leak anyway?
D26 is a 1S2473. Checking the datasheets it's supposed to have a max Ir (max reverse current) of 0.5uA.
Not even close to the 90uA I was getting!

Fix

Finally, I decided to remove D26 and replace it with a 1N4148.
Now I got ZERO current draw and I checked several times. 
I used my uCurrent (not the new Gold version) and measured the battery current with the uCurrent set to mV/nA.
Using my crappy Voltcraft meter I got 4.7mV (so 4.7nA). Using my Fluke 87 III it read 2.5mV (so 2.5nA).
So, the two RAM ICs and the NAND actually only used 2.5nA. Hah! That is WAY less than the 101nA I calculated above using the datasheet numbers.

OK, so the CR2450 should last for... 
540mAh/0.0000025mA = 216000000h = 9000000d = 24657 years...
Right. That's probably not gonna happen, but I think we can conclude that a regular CR2032 would have been just fine as well :)

Feels good to have this old problem finally taken care of! Aaaahhhhhh.....

Conclusion

When your memory backup battery is draining too fast, check/replace the diodes !!!

No comments:

Post a Comment