Reasons for scarcity of working TI-58C machines

[John Garza (3665)]

10 years or so ago, there used to be plenty of working HP-25Cs and TI-58s on ebay. Now you don't see any. Any ideas on the reasons? Yes I know some people plug in the charger without the battery - but that shouldn't change over time - unless the people selling them are for some reason less aware nowadays of the issue - maybe the influx of junk peddlers or consignment businesses vs. HP users and collectors. Maybe. But could it be some sort of inherent weakness - maybe extra sensitivity to electrostatic discharge (ESD) ?

I recently bought a pristine 58C from a known reputable seller. Fully functional. Very well packed when it arrived. But I did notice the calculator itself was wrapped in clear plastic - not the pink or silver antistatic wrap. Everything works except it will not go into LEARN mode - meaning you can't program it. It can execute code in the CROM master library - so I'm guessing the RAM got zapped in transit.

[Garth Wilson]

Are you looking for one?  I have one, which I think is the one I originally bought in Dec '81 (meaning I'm the original owner) that I guess I'm ready to let go.  I guess I kept thinking that I might someday want to use it again, but at this point I'm sure I won't.  I used it a lot for a short time (until I got a 59), and it looks nearly new.  The battery pack will have to be rebuilt though.  The battery pack has not been in the calc in probably 20 years; but I just tried the calc with 3.6V from my bench power supply, and as far as I can tell from a quick test, it all works as it should.  I also never had the key problems others tell about. I have the original case, a few modules, with manuals, and I have one power supply, but I don't know if I want to keep it for my 59.  I have not tried the power supply to see if it's still good or not. Actually, I have a PC-100C printer too, which I quit using because a couple of the springy contacts that contacted the inside of the calc were broken and I got tired of sticking little pieces of solderwick or something in there to make it work; but it did work, last time I powered it up, which was decades ago.

[Steve Simpkin]

I found it interesting that TI chose to use an off-the-shelf CMOS RAM chip from RCA, Toshiba or NEC in the TI-58C instead of making a CMOS version of their existing memory chip used in the TI-58/59. To use this commercial RAM chip they had to make an additional chip to interface to it. Because of this there is quite a difference in the circuitry between a 58 and a 58C. I also wonder if these early CMOS RAM chips were less durable over time than the PMOS chips used in the non-C version.

http://www.datamath.org/Sci/WEDGE/TI-58C.htm

[SteveC]

The TI build quality of the 58C generation was not as durable as some others in my experience I had a TI 58-C in high school. It lasted through my senior year of HS and then 1 year of college. It just stopped powering up. My brother's 58C lasted maybe a semester longer. I discovered HP calculators were much more durable several purchases and a field of study change later. I have my second HP from the mid 80s (A US made 12C) and it still works for daily use. I did wear out the 10B that was my first HP. Few electric powered technologies have that durability.

[John Garza (3665)]

I would think a simple battery backup for the PMOS RAM chips in a regular 58 or 59 would work. Anybody know the standby current drain on those TMC0598 chips?

[John Garza (3665)]

Update: Since my original post, I've repaired 3 TI-58Cs, all with dead Toshiba CMOS memory chips. It seems these chips are prone to fail.

I also noticed a functional HP-25C sold for over $500 on the auction site this week. Prices definitely seem to be going up.

-John

[brouhaha]

(02-11-2023 11:33 PM)John Garza (3665) Wrote:  I would think a simple battery backup for the PMOS RAM chips in a regular 58 or 59 would work. Anybody know the standby current drain on those TMC0598 chips?

They're probably dynamic, and would need the two-phase clock running in addition to the power supplies.

Most PMOS chips are dynamic, because dynamic logic and memory need fewer transistors. Typically about 1/2 for logic, and 1/4 to 1/6 for RAM. A lot of NMOS chips are dynamic as well, although static NMOS RAM was used for relatively low density memory systems, to avoid the need for extra circuitry for refresh.

[blackjetrock]

(06-13-2023 01:09 AM)John Garza (3665) Wrote:  Update: Since my original post, I've repaired 3 TI-58Cs, all with dead Toshiba CMOS memory chips. It seems these chips are prone to fail.

I also noticed a functional HP-25C sold for over $500 on the auction site this week. Prices definitely seem to be going up.

-John

You could probably replace the RAM interface chip and RAM chip with a modern microcontroller like the RP2040. Non volatile storage can then be done by using the flash of the microcontroller. No battery required. You do need some 'scheme' to interact with the mcrocontroller and tell it what flash to load and save to.

Andrew

[Garth Wilson]

(06-17-2023 08:54 AM)blackjetrock Wrote:  You could probably replace the RAM interface chip and RAM chip with a modern microcontroller like the RP2040. Non volatile storage can then be done by using the flash of the microcontroller. No battery required. You do need some 'scheme' to interact with the microcontroller and tell it what flash to load and save to.

Good idea, to use a microcontroller, to keep lots of these calcs alive where otherwise there'd be no hope.  I seem to have one of them.  I'm sure the nonvolatileness was from the battery continuing to provide power though, with CMOS memories' current being negligible when they're not selected and the power-supply voltage is in the retention range, below the operating range.  In 1986 at work I observed an 8KB SRAM holding its memory overnight on only a 10uF capacitor.  I did not try to see how much longer it could go.  Flash memories have a limited number of writes before they're worn out though.  The number has gotten super high for file storage, but not for use as RAM.  They also take much longer to write than they do to read (although that might not be a problem for a calculator as slow as the 58c), and you might have to write a whole sector just to write a byte, and even if it has a byte-write instruction, it might put the whole sector in a buffer, change just that one byte, and write the whole sector back, adding that wear to the whole sector.  I've worked with flashes that do that.  Non-flash EEPROMs don't.

[blackjetrock]

(06-17-2023 04:20 PM)Garth Wilson Wrote:  

(06-17-2023 08:54 AM)blackjetrock Wrote:  You could probably replace the RAM interface chip and RAM chip with a modern microcontroller like the RP2040. Non volatile storage can then be done by using the flash of the microcontroller. No battery required. You do need some 'scheme' to interact with the microcontroller and tell it what flash to load and save to.

Good idea, to use a microcontroller, to keep lots of these calcs alive where otherwise there'd be no hope.  I seem to have one of them.  I'm sure the nonvolatileness was from the battery continuing to provide power though, with CMOS memories' current being negligible when they're not selected and the power-supply voltage is in the retention range, below the operating range. 

Yes, I'm working on a casio and the non volatile memory is a low power CMOS. very low power, unlike the RP2040 I'm using which definitely is not low power.

Quote: In 1986 at work I observed an 8KB SRAM holding its memory overnight on only a 10uF capacitor.  I did not try to see how much longer it could go.

i was fiddling with an HP41C recently and that seemed to have the ability to retain its program memory for long periods of time when off. This was running off a USB supply too, so when it was off there were no batteries.

Quote:  Flash memories have a limited number of writes before they're worn out though.  The number has gotten super high for file storage, but not for use as RAM.  They also take much longer to write than they do to read (although that might not be a problem for a calculator as slow as the 58c), and you might have to write a whole sector just to write a byte, and even if it has a byte-write instruction, it might put the whole sector in a buffer, change just that one byte, and write the whole sector back, adding that wear to the whole sector.

Yes, I'm using the flash as a file storage sort of thing and using RAM to do the emulation on the bus. The RP2040 is dual core so one core does the emulation with interrupts off while the other core does everything else, USB CLI, flash reading and writing, wifi etc.

Quote:  I've worked with flashes that do that.  Non-flash EEPROMs don't.

A side effect of this work is that you can look at the internal representation of the programs and sometimes do things that weren't intended, like extra characters in numbers or synthetic programming.

Andrew

[Garth Wilson]

(06-18-2023 08:58 AM)blackjetrock Wrote:  A side effect of this work is that you can look at the internal representation of the programs and sometimes do things that weren't intended, like extra characters in numbers or synthetic programming.

I did minor synthetic programming on my 58c & 59 (accessing HIR registers comes to mind, but that was almost 40 years ago so I don't remember any details, and I'd have to review), but I have no idea what would happen if you could enter non-BCD numbers into the memory.

[dmh]

I have a working 58C and am wondering, based on this thread, will it definitely die sooner or later or if it's still working now it may continue for some time?

Will usage bring on the failure sooner?

Maybe I should get rid of it while it's working, I have a 59 too.

Thanks, dmh

[John Garza (3665)]

(06-28-2023 01:10 AM)dmh Wrote:  I have a working 58C and am wondering, based on this thread, will it definitely die sooner or later or if it's still working now it may continue for some time?

Will usage bring on the failure sooner?

Maybe I should get rid of it while it's working, I have a 59 too.

Thanks, dmh

Number one... throw away the charger.
Always charge batteries externally. I have reason to believe the 58C is sensitive to charger use without a battery or a bad battery or bad electrical contact with the battery. Similar to the HP-25C.

I use an HP classic charger box (Reserve Pack) and place the battery in with the plastic tab facing the LED. Place a little bit of foam near the hinge in the well to support the battery and move it's contacts in position over the charger contacts. Works great.

-J

[rprosperi]

(06-28-2023 01:16 AM)John Garza (3665) Wrote:  I use an HP classic charger box (Reserve Pack) and place the battery in with the plastic tab facing the LED. Place a little bit of foam near the hinge in the well to support the battery and move it's contacts in position over the charger contacts. Works great.

Thanks for this tip, it never occurred to me to use the HP charger for the TI packs. Beautiful tip, HP to the rescue once again!

[dmh]

Thanks for the advice.

I'll keep an eye out for an HP classic charger box!

(06-28-2023 01:16 AM)John Garza (3665) Wrote:  

(06-28-2023 01:10 AM)dmh Wrote:  I have a working 58C and am wondering, based on this thread, will it definitely die sooner or later or if it's still working now it may continue for some time?

Will usage bring on the failure sooner?

Maybe I should get rid of it while it's working, I have a 59 too.

Thanks, dmh

Number one... throw away the charger.
Always charge batteries externally. I have reason to believe the 58C is sensitive to charger use without a battery or a bad battery or bad electrical contact with the battery. Similar to the HP-25C.

I use an HP classic charger box (Reserve Pack) and place the battery in with the plastic tab facing the LED. Place a little bit of foam near the hinge in the well to support the battery and move it's contacts in position over the charger contacts. Works great.

-J

[LEDfan]

I found this thread rather concerning; my every day calculator is a fully functional TI 58C (I know, blasphemous in an HP forum, but brought up with TI/ Sharp calculators) and lives in my laptop bag with no anti-static type protection.

It’s a bit of a rough model, with lots of knocks and dents, which had been adapted to take 3 x AAA batteries by a previous owner, so at least batteries get charged externally.

Thinking about taking it out of circulation now, if it’s a ticking time bomb. I’m converting across to RPN, and really appreciate the build-quality of the classics.

[Garth Wilson]

(07-08-2023 08:29 PM)LEDfan Wrote:  Thinking about taking it out of circulation now, if it’s a ticking time bomb.

After further testing, I found that the 58c I mentioned above, which I thought was still in perfect condition, does have the failed memory.  It went south even without being used, meaning there's something about mere time that ends their life.  It was fine the last time I used it, which was decades ago, and the charger was fine too, so I can't blame the charger.  When I tested it now, I did so with an adjustable, regulated power supply set to 3.6V since I don't have a working battery pack for it.

[blackjetrock]

(07-08-2023 11:50 PM)Garth Wilson Wrote:  

(07-08-2023 08:29 PM)LEDfan Wrote:  Thinking about taking it out of circulation now, if it’s a ticking time bomb.

After further testing, I found that the 58c I mentioned above, which I thought was still in perfect condition, does have the failed memory.  It went south even without being used, meaning there's something about mere time that ends their life.  It was fine the last time I used it, which was decades ago, and the charger was fine too, so I can't blame the charger.  When I tested it now, I did so with an adjustable, regulated power supply set to 3.6V since I don't have a working battery pack for it.

I've just had a search on the internet and there is a service manual for the TI58 and details about the multi-register IC that provides the RAM storage, so it looks like it would be feasible to replace the RAM with an RP2040. It seems to run off 9V but that's not a problem with some voltage dividers and a driver IC to do the outputs.

It seems to be an unusual device so dropping a RAM chip in doesn't look to be simple.
Best not to throw a broken 58C away. I'll have to try to get a broken one so I can have a go.

Andrew

[brouhaha]

(07-09-2023 04:41 AM)blackjetrock Wrote:  I've just had a search on the internet and there is a service manual for the TI58 and details about the multi-register IC that provides the RAM storage, so it looks like it would be feasible to replace the RAM with an RP2040. It seems to run off 9V but that's not a problem with some voltage dividers and a driver IC to do the outputs.

It seems to be an unusual device so dropping a RAM chip in doesn't look to be simple.
Best not to throw a broken 58C away. I'll have to try to get a broken one so I can have a go.
Andrew

If the TMC0591 isn't damaged and you just want to replace the RAM, a normal 5V CMOS low power SRAM should work, e.g.6264L. They don't run the SRAM on 10V; the whole point of the TMC0591 was to interface to normal CMOS RAM.

Because the calculator chips are PMOS, everything is referenced below ground. The calculator ground goes to the SRAM Vdd, and the TMC0591 Vdd2 goes to the RAM Vss.

Tie the extra SRAM address lines to Vdd2. Tie the extra SRAM data lines through individual series 22Kohm resistors to Vdd2.

I haven't actually done this, as I don't have a TI-58C.

[blackjetrock]

(07-09-2023 05:36 PM)brouhaha Wrote:  

(07-09-2023 04:41 AM)blackjetrock Wrote:  I've just had a search on the internet and there is a service manual for the TI58 and details about the multi-register IC that provides the RAM storage, so it looks like it would be feasible to replace the RAM with an RP2040. It seems to run off 9V but that's not a problem with some voltage dividers and a driver IC to do the outputs.

It seems to be an unusual device so dropping a RAM chip in doesn't look to be simple.
Best not to throw a broken 58C away. I'll have to try to get a broken one so I can have a go.
Andrew

If the TMC0591 isn't damaged and you just want to replace the RAM, a normal 5V CMOS low power SRAM should work, e.g.6264L. They don't run the SRAM on 10V; the whole point of the TMC0591 was to interface to normal CMOS RAM.

Because the calculator chips are PMOS, everything is referenced below ground. The calculator ground goes to the SRAM Vdd, and the TMC0591 Vdd2 goes to the RAM Vss.

Tie the extra SRAM address lines to Vdd2. Tie the extra SRAM data lines through individual series 22Kohm resistors to Vdd2.

I haven't actually done this, as I don't have a TI-58C.

I don't have a 58C either. i might look out for one though, preferably broken.
I've done lot of RAM replacement with the RP2040 recently, it must be 5 or 6 by now. One was a CMOS RAM that ran with a similar supply regime. That actually ran off 6.4V and that, I think, killed it eventually. I'll have a more detailed look at the circuit.
One advantage of the RP2040 is that you can store RAM images in the onboard flash, giving you NV RAM and also a catalogue of images.


Andrew