Texas TI-58C: (in)famous constant memory issues

[jebem]

I have this Texas TI-58C that is not able to store in any of the registers or program codes.
Otherwise, the calculator operates normally and all the math and scientific functions gives the expected results.

The previous owner left the batteries inside and it leaked badly.
I have dismantled the machine piece by piece, put everything in the kitchen sink in warm water and dish soap for 15 minutes, then used a soft brush to remove the battery remains from the PCB and plastics.
Two days later I powered it on and it worked normally, except that it doesn't store in the registers or in the program steps. Just get zeros when reading.

I am not able to find the service guide or the schematics, so I'm looking into the PCB traces and try to figure out its architecture.
There are some information in the Internet concerning the TI-58 and TI-59, including the service guide, but these share little with the TI-58C at the hardware implementation, despite being very similar at the architecture level.

Does anyone has this documentation on the TI-58C and willing to share or sell?

1) For this TEXAS TI-58C from 1981 (ATA3381), the main PCB components are:

• TOSHIBA TC5047AP-1 (SRAM 4-bit 4K 1024X4 550nS)
• TI TMC0591NL (SRAM I/O, DoM y1980 w18)
• TI SN27882N (Display Driver, DoM y1981 w22)
• TI SN27882N (Display Driver, DoM y1981 w22)
• TI TMC0573NL (ROM1, DoM y1980 w18)
• TI TMC0501-NL (CPU, DoM y1981 w27)
• TI TMC0580NL/CD2400NL (ROM2, DoM y1981 w22)
• TI TMC0580NL/CD2401NL (ROM3, DoM y1981 w31)
• NICHICON 47uF/10V (NiCd battery bypass, DoM y1981 w15)
• NICHICON 470uF/10V (SRAM battery backup, DoM y1981 w16)
  

It seems that the registers and program area are sharing a common storage sub-system based on two IC's (TC5047AP-1 and TMC0591NL).


2) A few things I have checked:


2.1) I have recorded the current consumption to be about 220mA, which seems on the high side for me, but I do not have another machine to compare.

2.2) The voltages from the DC-DC converter power supply for the Texas IC's are fine:

• Vdd = -10VDC (Pin 2 at CROM socket);
• Vgg = -16VDC (Pin 1 at CROM socket);
• Common point = Vss at +Vbatt. (Pin 8 at CROM socket).
  

2.3) The power supply for the SRAM TC5047AP-1 is fine (reading taken while powered ON):

• Vdd2 = -4.9VDC (pin 8);
• Common point (pin 20);
  

2.4) The SRAM backup power supply (470uF cap.) looks fine as well (reading taken while powered OFF):

• Vdd2 = -3.6VDC (Pin 8) --> the voltage will drop with the passing time;
• Common point (pin 20);
  

2.4.1) This large 470uF capacitor near the DC-DC power supply is a energy storage for this 5VDC SRAM IC supply, allowing the battery to be removed without losing the SRAM contents.
Basically, the Capacitor positive terminal is connected to the +Vbatt (this is the common return), while its negative terminal is fed by two power supply sources:

• When in power OFF state, the -Vbatt (-3.6V) feeds the SRAM thru a Germanium diode;
• When in power ON mode, the -Vdd (-10V) is feed thru a transistor/regulator (I had no time to investigate what this component is and how it is driven).
  

3) So, I believe that the SRAM chip may be faulty. Why?
Well, I do not have the required equipment to do the circuit analysis to be sure, but:

• The SRAM IC is basically in parallel with the battery;
• The battery is fed thru a full wave rectifier bridge from the AC power adapter;
• The battery was dead and in short circuit, so probably it was removed at some point;
• This means that if someone powered On the calculator without the battery pack in place just to test it, the SRAM chip did receive an unregulated high voltage at their power supply terminals, killing the chip.
  

I'm ordering a replacement SRAM chip from China. Wait and see from now on.


4) Some pictures from the internals:

[jebem]

Power supply voltage sources:

• Common: +Vbatt
• Vbatt (-3.6V)
• Vdd (-10V)
• Vgg (-16V)
• Vdd2 (-5V) --> when powered Off, Vdd2 falls to -Vbatt (SRAM backup)
  

Master Library 1 diagnostic results:

• Library Module Check (SBR 2nd R/S) result: 1
• Run Diagnostic (SBR =) result: 0 (flashing)
  

Apparently I have at least the Master Library 1 ROM in good condition.
As expected, the Diagnostic fails.

[jebem]

External Power Adapter/Charger and Battery Pack information:


According to the official TI site, the TI-58C should use:
- Battery Pack to use: BP-1A
- Adapter/Charger to use: AC9131.

The "Appendix A" from the official Texas "Personal Programming Owner's Manual for TI Programmable 58C/59", 1977, 1979, PN:1034157-1, ISBN 0-89512-027-5 also confirms the above information.
Additional information taken from the above manual:

- CAUTION: The calculator can be damaged if the charger/adapter is connected without the battery pack installed.
- The battery pack needs at least 4 hours to a full charge when the calculator is powered off, or about 10 hours when it is powered on.
- A fully charged battery pack allows for about 2 to 3 hours of continuous operation.
- Under normal use, the rechargeable battery pack have a life of 2 to 3 years or about 500 to 1000 recharge cycles.
- The battery pack can lose its storage capacity if it is not allowed to discharge occasionally. For maximum battery life, it is recommended that you operate the calculator as a portable device at least twice a month, allowing the batteries to mostly discharge, then recharge accordingly.
- Repeated occurrences of excessive battery discharging will permanently damage the battery pack.

The BP-1A battery pack includes three Ni-Cd rechargeable battery cells, for a total of 3.6VDC and a capacity of about 600mAh.
My battery pack contains three cells, each one printed with the reference "188-8134 ASM IN MEXICO".

- The Adapter/Charger AC9131 to be used with the TI-58C, should have an output of 3.3VAC at 500 mA, accordingly to various sites in the Internet.
Here and here and here and here
I do not have one, so I can't confirm that output value, but it seems that a value between 3.3VAC RMS (4.6VDC open circuit) and 5VAC RMS (7VDC open circuit) should be acceptable.
The adapter should indeed output an AC current because the calculator includes a full wave bridge rectifier and a bleeding resistor to limit the batteries charging current.

[d b]

RE: "- CAUTION: The calculator can be damaged if the charger/adapter is connected without the battery pack installed."
About 10 years ago Joerg Woerner confirmed to me that this is true.

[jebem]

(11-07-2014 03:45 PM)Den Belillo (Martinez Ca.) Wrote:  RE: "- CAUTION: The calculator can be damaged if the charger/adapter is connected without the battery pack installed."
About 10 years ago Joerg Woerner confirmed to me that this is true.

And I have just confirmed that my TI-58C RAM memory chip was also another victim of such irregular procedure performed by the previous owner. Details below.


Toshiba TC5047AP-1 Memory chip replacement arrived today.

After waiting for almost one month for the replacement memory chip, it finally arrived today. I am surprised to be able to find these old chips for sale in the market, so I have ordered a few, instead of keep buying additional TI-58C machines hoping to be able to find/build a working one. I like this electronics hobby after all.




Time to extract the old defective memory chip.
I still remember the technique we used to apply to remove the TTL and MOS DIP style chips in the good old days - just use a proper cutter to remove the offending component with zero damage to the PCB.



The underside of the memory chip reveals some nasty effects, probably caused by the high current passing thru the pins, thanks to a combination of high voltage from the power adapter used without the battery pack and the battery electrolyte leaking.




Removing the old chip pins and prepare the PCB for the new chip.
I like this audio grade Wonder Solder by TRT (rosin flux core leaded solder). It is very easy to work with, making very good joints, and shows excellent conductivity as well.



Time to insert the new memory chip and assemble the machine.
As I do not like the keys to jingle in the slots, I have applied a new thicker foam as well.




Testing time....

I was rewarded with a working machine. It can store data and programs once again!