I bought this card a couple of years ago as a spare and back then it was working 100%.
Having recently developed an Arduino2560 based test rig to run memory & battery tests on the Sector101 Datablade card I brought this card out of storage to give it a try in the rig. Although the card passed all the bit-pattern tests the battery was completely flat. I’m sure I had put a fresh battery in when I got the card so that was unexpected. I put another in recently and tested the current pulled from the battery (< 0.2 uA) and thought nothing more of it.
A week later the battery was down to 1.4 volts. Hmmm… what’s happening here?
As documented in my other faulty MCD card post these cards have a tendancy for the internal insulation to break down and cause shorts between the PCB and the inside of the metal cover. Time to whip this open and take a look!
It took quite a bit of time to find the fault as the damage was tiny. The photo above shows trace damage where half the trace has been eaten away. Before I removed the soldermask from the trace there was a tiny dot of what looked like crystallized flux residue just over the damaged area. It is possible this tiny dot of flux has been gently eating away at the trace and insulation layer for the past 26 years. The other areas of soldermask in the photo were removed for continuity probing.
The photo below shows the matching area on the inside of the metal cover. There is a tiny darkened area which directly faced the damaged PCB trace. It is safe to assume that this trace was shorting to the metal cover at this location. Further inspection of the PCB showed no other shorts.
Why the flat battery..?
The trace that was shorting to the metal cover did not directly connect to the battery circuit so why would the battery go flat so quickly? Let’s take a look..
The arrows in the photo above show the points on the circuit where the damaged trace connect. The signal source is Pin 15 on the HC138 and the destination is Pin 20 on the memory IC.
The MCD 64 card uses two 32K ICs to simulate a 64K memory area. The HC138 looks at the memory address location (the SY / TG sets this) and chooses which IC to activate for addresses 0 thru 32767, and 32768 thru 65535. It is possible to add a further two 32K ICs in the unpopulated locations to create a 128K MCD card but the SY / TG series do not look for this extra memory so that would be redundant.
Pin 20 on the memory IC is the Chip Select signal (/CE). Basically this tells the IC to wake up and do something – either output a byte or record a byte depending on other control signals. If the /CE signal is High (5V) then the IC sleeps. If the /CE signal is Low (0V) then the IC wakes up.
What’s been happening is that the /CE signal has been intermittently shorting to the cover forcing the signal to Low (0V) and waking the IC up from its low-current sleep state. As well as the memory ICs the HC138 runs from the battery when the power is off. The output from the HC138 has been grounded causing a further drain on the battery. Luckily there is a 470R resistor between the battery and the ICs so a direct battery short was avoided.
The card trace has now been repaired and a thick layer of Kapton tape placed over all the PCB traces to create a robust insulation layer. The card covers have been fixed in place with Dow Corning 3140 flowable silicone. I have used this silicone for all the MCD Sweet16 modifications and card repairs in the past as it is designed for electronics and is kind to copper. Regular silicone adhesives have a pungent acid smell like vinegar but this silicone has no smell at all. I tried other methods of attaching the metal covers in the past like using double-sided tape but this created extra width to the card and it would not fit into the card slot without a lot of force.
As well as the trace repair the 2 ceramic 100nF capacitors were replaced. Ceramic caps have been known to leak DC over time and as new caps are inexpensive it was worth doing this when the card was open.
Now the card is back together (with another fresh battery installed) I expect it to be working for another 26 years 🙂