TELEPIC - a replica TELEKIT

[Phil__G]

Wow. I'm envious of your proficiency with Kicad Ian! Very nice.
A thought - would it be sacrilege to make the "shift" slide switch a button that looks like a slide switch?
I dont expect that would need any code changes to the standard Telekit image...

[Realtime]

The current layout has 2 buttons in the space occupied by the original slide switch, allowing the CTRL and SHIFT functions to be used in the same way as a traditional keyboard. Hopefully that will keep the look of the original but make it more practical to use.

[Timbucus]

This looks amazing - well done.

[Realtime]

A quick update. The PCBs have arrived (amazing quality as always) and the first ones are built ready to start testing.

[Mark1960]

It's looking really good.

[Timbucus]

Wow looks amazing well done.

[Realtime]

Things have gone very well and TELEPIC is communicating reliably with the LCDS at 110 baud (20mA loop) and with a PC using the TTL interface. I made a mistake when replacing an unobtainable shift register with a 74LS device and inverted the clock, so there's a track cut and wire link to rectify that. Also, the FLAG0 output from the PIC floats during a RST and causes garbage to be output on the serial line. A pull down has fixed that.

You can see in the photos that the left most digit is showing ghosting from other digits. This also happens on the original TELEKIT. The reason being that the SC/MP serial line (SOUT) is also used as the column select for the 9th LED to avoid the need for a 9 bit shift register. Maybe there's a fix I can put in without changing the SC/MP code? Any suggestions welcomed (draft schematic attached).

The PIC code needed a few tweaks, but no changes to the original SC/MP code. It was great having the PIC programming interface wired off board, allowing in-system programming of the PIC.

The area that was worrying me most was the keypad design. As we saw in an earlier post the key mapping is not a simple matrix. However, I'm happy to say it's working fine. This is also good because the very expensive replacement I bought for the original TELEKIT keyboard has now started to fail - the 'A' key no longer responds - so I'll be able to replace that if I wish.

Next task is to design a 3D printed frame to assemble the new keypad into the Novus shell.

[ceptimus]

Maybe a 74LS30 or similar (8-input NAND gate), to select digit 1 only when all the other 8 digits aren't selected? You could still gate the signal with SOUT, so that the display can be completely blanked when the TELEPIC is too busy to drive it.
That would need a new PCB, of course, so a bigger shift register might be a simpler option.

[Realtime]

Thanks that's a good suggestion. It got me wondering whether an 8 input diode-OR gate could feed into Digit1 cathode to provide the blanking signal. So I cludged it on the rear of the board using schottky diodes and it works well in removing the ghosting, although SOUT now modulates each of the LS164 outputs (about 0.4V amplitude) as the diodes conduct.

Should there ever be a Rev B PCB then a CD4078 OR/NOR device could be added.

[SiriusHardware]

Where did you find that 0.1" pitch 'stiff' clear ribbon cable in post #29, image #2? That's almost identical to what was used on original MK14s and would obviously also be great for MK14 replica builds.

[Mark1960]

Can the ghost be fixed in software by turning off the segment drive before scanning the keyboard?

[Realtime]

Originally Posted by SiriusHardware

Where did you find that 0.1" pitch 'stiff' clear ribbon cable in post #29, image #2? That's almost identical to what was used on original MK14s and would obviously also be great for MK14 replica builds.

That's an original Novus calculator display and flexi (from a 6025). However, if you search ebay for 2-1437186-9 you'll find 15 way strips at a very good price. I used 10 ways on my Iss VI MK14. Alternately search for "FST-21a cable".

[Realtime]

Originally Posted by Mark1960

Can the ghost be fixed in software by turning off the segment drive before scanning the keyboard?

Good question. I need to go back and look at the FLAG1/FLAG2/SOUT timing used in the Gleichmann production Telekit code. Perhaps FLAG2 was introduced for this purpose. It is released after FLAG1, which is used to enable/disable the segment drive to the LEDs in the prototype Telekit (and TELEPIC).

[ortek_service]

Originally Posted by Realtime

That's an original Novus calculator display and flexi (from a 6025). However, if you search ebay for 2-1437186-9 you'll find 15 way strips at a very good price. I used 10 ways on my Iss VI MK14. Alternately search for "FST-21a cable".

Yes, I've used those sold-core Flexi-strip 0.1" pitch Jumper-cables before. With Mouser etc stocking a few FST-21A 'Teflon' types: https://www.mouser.co.uk/c/?q=FST-21A

A long-time ago I bought some cheaper versions (from Electrospeed / Verospeed ?), that the strips appeared to be wrapped in paper with adhesive to hold the strips aligned. And looks like these are the same: https://www.mouser.co.uk/c/?q=FSN-21A
- But the strips tended to move / paper discoloured, if you applied heat from directly-soldering to the ends.

[SiriusHardware]

The ones I bought for my issue VI build have embedded flat flexible metal strips rather than round wires as in Realtime's old-school ribbon from a Novus, as such they are more prone to flapping about with the weight of a display teetering on the top end, not quite as stiff. The ones in Owen's second Mouser link above appear to be the same ones.

The PCBs in washing machines, printers, even Hameg scopes often used to be interconnected by very tough, but still bendable, typically grey or green 0.1" flat ribbon with round, rather than flat embedded conductors but I never did find out where to get those.

[Realtime]

Originally Posted by Realtime

Good question. I need to go back and look at the FLAG1/FLAG2/SOUT timing used in the Gleichmann production Telekit code. Perhaps FLAG2 was introduced for this purpose. It is released after FLAG1, which is used to enable/disable the segment drive to the LEDs in the prototype Telekit (and TELEPIC).

Well, I've disconnected FLAG1 from the segments shift register output enable (IC5) and connected FLAG2 instead (with the longer duration pulse) and it does indeed remove all ghosting . In fact it makes the whole display look cleaner.

So thank you Mark1960 for asking the question. I don't think I would have gone down that road otherwise.

[Realtime]

The first TELEPIC is assembled. I've posted a short video here. The case is from a 6025 Novus calculator. I'll be 3D printing a replica in due course.

[Phil__G]

Brilliant Ian, did you have to do many changes to Karens emulation code? I know you would have had to implement SIN & SOUT (I think I did SOUT?) and maybe more....

I'm most impressed that you resisted any changes to the original Telkit code, that makes it a true hardware replica in my book

.

[SiriusHardware]

Lovely job Ian, and I confess I am, despite earlier scepticism, actually quite impressed by how readable the pseudo-alphanumeric output is on the 7-segment displays with the ingenious use of the DP segment in some cases to 'suggest' the down-right sloping leg of 'R' or 'K' and the convention of using a 'tall inverted U' for 'M' and a 'short inverted U' for 'n' so your brain soon gets used to which one is which.

This really does look to have been a labour of love.

[Timbucus]

That is amazing - really looking forward to making one of my own now! Well done