Nat Semi SC/MP Low Cost Development System

[Realtime]

I've recently become the proud owner of a National Semiconductor Low Cost Development System (LCDS). It's SC/MP based and intended for the "development and testing of SC/MP hardware and software designs...". Currently it's non-working so this thread will be capturing the progress to get it up and running again.

If that's not exciting enough, it came supplied with something I've not found in any of the on-line LCDS descriptions and photos. It's a TTY emulator called TELEKIT in the form of a modified NOVUS calculator that plugs into a 7 pin DIN socket (again not a standard LCDS feature). It's intended as a low cost replacement for a genuine teletype, which in the late 70's cost around £1000.

I've found 3 references to TELEKIT in old electronics and computing magazines. One from John Miller-Kirchpatrick (Sept '76, attached) states that only around 25 of these were produced before Nat Semi gave up on them. A second one (Nov '76) states that TELEKIT has been produced as an add on for Introkit, which is obviously something different to the Keyboard Update kit sold for Introkit. In 1977 the Introkit sold for £66.33, the Introkit Keyboard kit £66.50 and the LCDS with one processor card £349.30 (+VAT @8%? on all of those). Prices taken from a Bywood Electronics advert. The TELEKIT price was intended to be in the region of £150-£200.

What I'm hoping for is that someone else has a TELEKIT and perhaps some documentation they would be willing to share. But even if you remember seeing one or using one in the past that would be great to know about.

I've scoured the internet for every reference I can find to the LCDS. There are some in museum collections but the on-line info is generally quite scant so it would be nice to build up a fuller picture. My LCDS came with a full user manual, so that's a big bonus (although it is available on line too).

[SiriusHardware]

Fantastic find, we look forward to hearing more about it as time goes on. If you haven't already found them there are a couple of threads over on VCFED which you might want to take a look at. Incredibly one of the contributors to one of the threads is 'realtime' (no capital) ...but he is not you.

https://forum.vcfed.org/index.php?th...-needed.66412/

https://forum.vcfed.org/index.php?th...sting.1238428/

[Realtime]

Thanks for the links. I had come across VCFED but hadn’t seen user ‘realtime‘
I’ve registered on the site as RealtimeUK. I was going to use TheRealRealtime, but didn’t want to appear rude

I see those threads have been running for a few years now so hopefully I can tap into their experience.

[Realtime]

So I've made some good progress with this. Normally when the SC/MP CPU exits reset it would start executing from address 0x0001. However, there is a state machine in the LCDS that takes over the address and databusses at the end of an instruction cycle forcing the execution to jump to subroutines in page 0x7000. At this point the contents of the SC/MP registers are read and stored in RAM. This happens for every program instruction executed and so provides a single-step mode with the ability to interrogate and modify the SC/MP registers.

What I found is that there is a DM8544 Quad Switch Debouncer device (actually just a quad Set/Reset latch) used in that state machine whose outputs were neither floating nor driving - almost an open collector behaviour. Definitely not right so needed to be replaced. Apart from one source on eBay (that I won't use) I couldn't find any DM8544s, so have instead re-wired a 74279N quad S-R latch to look like the DM8544.



Now when I power up I get a display and the ability to interrogate the SC/MP registers . That was a big surprise as I expected that if one device was damaged there would be others. So far so good.

I would like to fit a DM8544 at some point if possible. Does anyone out there have any spares they would be willing to part with or know of a reliable source?

The top right segment of the LED display is not currently working so that's the next thing to investigate.

[Timbucus]

Originally Posted by Realtime

So I've made some good progress with this. Normally when the SC/MP CPU exits reset it would start executing from address 0x0001. However, there is a state machine in the LCDS that takes over the address and databusses at the end of an instruction cycle forcing the execution to jump to subroutines in page 0x7000. At this point the contents of the SC/MP registers are read and stored in RAM. This happens for every program instruction executed and so provides a single-step mode with the ability to interrogate and modify the SC/MP registers.

What I found is that there is a DM8544 Quad Switch Debouncer device (actually just a quad Set/Reset latch) used in that state machine whose outputs were neither floating nor driving - almost an open collector behaviour. Definitely not right so needed to be replaced. Apart from one source on eBay (that I won't use) I couldn't find any DM8544s, so have instead re-wired a 74279N quad S-R latch to look like the DM8544.

Attachment 285021

Now when I power up I get a display and the ability to interrogate the SC/MP registers . That was a big surprise as I expected that if one device was damaged there would be others. So far so good.

I would like to fit a DM8544 at some point if possible. Does anyone out there have any spares they would be willing to part with or know of a reliable source?

The top right segment of the LED display is not currently working so that's the next thing to investigate.

Attachment 285022

Wow well done nice progress. As regards the chip Cricklwood list them and I have never had a fake from them.

[Realtime]

Originally Posted by Timbucus

Wow well done nice progress. As regards the chip Cricklwood list them and I have never had a fake from them.

Thanks for the pointer to Cricklewood. I'll definitely put an order in. They have quite a good range of parts from the earlier logic families.

[circuitryboy]

Interesting to see this LCDS version with two heatsinks at the rear. Wiki has a picture of a version with taller boards plugging into DIN connectors.
If JM-K's 'only about 25 TELEKITs produced' is correct then user guide will be elusive.
The calculator used was one of a small range of specialist Nat Semi models from 1975, a 6025 'Financier' (the -5 and 'PR' is programmable version).
Only that model had those keys with those colours.
(Search '203599533974' on e**y for a good US specimen - shipping is +144%!)
At present I can't see how one emulates a Teletype with an 8-digit calculator display.
And what does the silver knob do?

[Mark1960]

Found the schematic for the LCDS motherboard on bitsavers.

http://www.bitsavers.org/components/...atic_Aug76.pdf

[Realtime]

Originally Posted by circuitryboy

Interesting to see this LCDS version with two heatsinks at the rear.

Yes, it's the only one I've seen with two. Originally they were sold with no internal power supply and connections were made to the terminal block at the top of the motherboard. I've attached a PDF showing the internals of this LCDS. The Power Supply mod seems to have been professinally installed. I guess the heatsink was added on -12V regulator as their application needed more capacity. The second heat sink definately is an afterthought as it covers the equipment label on the rear.

Originally Posted by circuitryboy

And what does the silver knob do?

Someone has added a small audio amp, speaker and volume control attached to the TTY data pins. I suspect this was added as part of testing the TELEKIT so you know when messages are being sent / received?? I've attached an audio clip of a TTY burst output on initialisation. Anyone like to tell me what characters are being sent ?

Originally Posted by circuitryboy

The calculator used was one of a small range of specialist Nat Semi models from 1975, a 6025 'Financier' (the -5 and 'PR' is programmable version). Only that model had those keys with those colours.

Nice find. I hadn't been able to find that model.

Originally Posted by circuitryboy

At present I can't see how one emulates a Teletype with an 8-digit calculator display.

Me neither It' s actually a 9 digit display but time will tell how usable it is. Once I've sorted the dead LED on the 7 segment display then the TELEKIT is next on the list.

[Realtime]

The LCDS display is a 6 digit 7-segment display. As noted before, one segment of the right most digit was not working.


On removal of the display diffuser I found each digit has it's own set of removable light pipes, and under each pipe is an LED about 0.25mm diameter. Just a dot really.


The existing LED was not recoverable so would have to be replaced. The smallest thing I could find was an 0402 surface mount LED with a 1.5x0.5mm footprint. Much bigger than the original but workable with a modification to the light pipe above that LED. After some amount of swearing I got it installed but it was (of course) massively brighter than the original. I ended up fitting a 5K1 resistor in series with the new LED (the originals have a 100R series resistor, showing just how LEDs have improved). It was a trade off between enough brightness when on and minimising ghosting when off due to leakage current. You can see it's dimmer than the original but it's working well enough.


I feel inclined to replace the axial resistor with a hidden surface mount part under the display difusser now that I know what value is needed but I know I'm asking for trouble, so I'm just going to leave it be .

[Mark1960]

It looks like that display might have been modified from a display for a desktop calculator. It seems unlikely they would design something like that assembly for what was probably a low volume development system. Curious if anyone has any suggestions which calculator may have used that type of display.

I am impressed you were able to make an effective repair on that type of display assembly, without damaging more the the working segments.

[circuitryboy]

Nat Semi made clock modules with the displays constructed in that way.
The light-guides were in a single white block. The red filter was a 'casing' that fitted over it.

[SiriusHardware]

Funny you should mention that, I once repaired a LED clock-radio alarm with failed segments by replacing all of the LED elements in the display with modern tiny SM LEDs. I had the same problem as Realtime re: brightness though - I ended up putting a pot in series with one element and adjusting that until the segment brightness 'looked right', then replaced all of the series resistors with the standard value closest to the measured value of the pot.

[Realtime]

Originally Posted by circuitryboy

Nat Semi made clock modules with the displays constructed in that way.
The light-guides were in a single white block. The red filter was a 'casing' that fitted over it.

That does look very similar. On the LCDS display there is a NS part number 980305030 A . Only reference I found to that is on the Computer History Museum website where it's included as one of the equipment part numbers.

Originally Posted by Mark1960

I am impressed you were able to make an effective repair on that type of display assembly, without damaging more the the working segments.

Thanks. I was very aware of the fragility of those tiny gold wires attaching the LEDs to the PCB and was glad to eventually put the cover back on and call it a day !!

[circuitryboy]

In eager anticipation of investigation of the TELEKIT itself, this might be of interest:
http://www.vintagecalculators.com/ht...ator_chip.html
An enthusiast administers TLC to a Novus 4525 'Scientist' PR (brother to the 6025 'Financier').
[I would be In Trouble if I did that on the Dining Table!] The PR versions were NiCd rechargeable.
Shows the keypad is self-contained and connects by clear flexible into connector block on the pcb.
Note the mention of red Fresnel augmenting the 'bubble' displays.

I've been assuming that NS Germany ran the DIN plug cable where the charging-socket was and the NiCds discarded.
And that there's new circuitry inside to handle serial signals through the DIN plug?

[Realtime]

The time has come to open up the TELEKIT and see what's going on. Currently when it powers up it generally has 8's on 8 of the digits , but sometimes on 9 digits. Other than that no obvious life.

Originally Posted by circuitryboy

In eager anticipation of investigation of the TELEKIT itself, this might be of interest:
http://www.vintagecalculators.com/ht...ator_chip.html
I've been assuming that NS Germany ran the DIN plug cable where the charging-socket was and the NiCds discarded.
And that there's new circuitry inside to handle serial signals through the DIN plug?

I expected there would be either a dedicated logic chip, designed for that purpose, or a SC/MP. Guess what?


Wow, a white ceramic SC/MP. That looks like the exact same keypad and flexi, so yes very interesting to see that repair on the VintageCalculators site. Generally all looks good. No burned components and the such. The flexi to the keypad looks to be in excellent condition and hasn't deteriorated and cracked like ZX81 flexis do.
On the back of the card we learn a bit more about prototyping at NSC.


Quite a few mods made there. Looks like they may have corrected the mapping on the keypad, but the one that made me laugh is 3 wires reconnecting signals that were drilled through for a fixing hole.
And here's a breakdown of the IC's on-board. Date codes are 1975/1976 but with a 1977 EPROM.


From what I can tell with a scope the processor core is running and it does alter it's behaviour when a TTY message is sent by the LCDS base unit, so I'm going to focus on the keyboard/display side of things to start with. First off though, I'm going to make a copy of the EPROM before I do something stupid. I spent yesterday putting together a MM5204 to 2716 adapter so I can read back the EPROM on my DATAI/O 29B.


I'm going to put Phil_G's SC/MP Disassembler to good use once I've recovered the code and work out what the processor should be doing.

[Mark1960]

From the components listed it seems likely all interface to keyboard and display is being done through flag, sense and serial io. A single line held high or floating high might be enough to light all segments.

[Realtime]

Thanks that's useful to know. Do you know of an example design that uses serial for the keypad? I've seen a number of MK14 type parallel key matrix implementations but not serial. When I was 'scoping around the board I did see that the serial output had an alternating pattern on it. Didn't see anything on the flag outputs that I remember. What I did see was that the TTY from the LCDS base unit was received at SENSEA and its inverse at SENSEB. Perhaps SENSEA is used to trigger an interrupt subroutine?

Once the EPROM is backed up then I can start having a more detailed look.

[Mark1960]

No I haven’t seen any similar keyboard interfaces, just a guess from the number of shift registers on the parts list.

Surprised they have both sensea and senseb for the serial input. I expected maybe sensea and serial in, similar to nibl. Maybe serial in is being used for keyboard input.

Hopefully the firmware survived and isn’t degraded. It might be an interesting project for a replica.

[Phil__G]

Originally Posted by Mark1960

I expected maybe sensea and serial in, similar to nibl

er....
NIBL uses the same principle as Kitbug, just senseA for soft serial input, no SIN or SOUT

This calculator-based add-on, presumably it reads ascii async serial from the Introkit and converts it to 7-segment in the same way as the Science of Cambridge 'Message program' ?

And reads the keypad matrix & converts it to ascii async serial back to the Introkit?

Re the eprom, to read old ones I generally do at least half a dozen reads and compare all the results for a match