MK14 schematic revisions

[SiriusHardware]

The regulator on my original MK14 has been mounted offboard on a much bigger finned heatsink for, well, more or less forever. It still gets quite warm. When I had the VDU connected to it originally I ran everything on the one regulator but I already had it on a chunky heatsink by then.

I think with this one I'll try running it from a variable bench supply on between 7.5-8.0V, whatever the lowest input voltage consistent with good 5V regulation is, and if the regulator runs sufficiently cool that way I will make a dedicated 8V(ish) linear regulated PSU for it. I think I actually have a 7808 regulator somewhere.

[Timbucus]

Originally Posted by SiriusHardware

RH one has MK14 review, Middle one, SCRUMPI 3, I am defeated by the LH one however.

I didn't know Gilsoft, purveyors of fine Text Adventure Authoring Software for the ZX Spectrum etc, had ever produced an in-house magazine. How many of those did you do?

Ah the LH one is a trick really as it has part two of what is on the October cover with the SCRUMPI3 the Tolinka Chess Recorder - well worth a read as its VDU system is very well described. I include it as it is also the edition with the first edition of Computing Today and the launch of the Triton... a sea change moment in home computing with the NASCOM as well - I also remember drooling over those two. Someone has found a Triton so it should form an interesting You Tube series...

https://www.youtube.com/watch?v=0cSRgJ68_tM

I think we only ever did the one magazine mainly for distributors etc...

[SiriusHardware]

Thanks for the link.

Regarding regulators, on a modern replica maybe one of these would be acceptable. No heatsink required.

https://uk.farnell.com/murata-power-...-5a/dp/2102101

[SiriusHardware]

Anyway, I expect you're all wondering whether the board works?

[SiriusHardware]

...Of course it does. Well done Slothie, a magnificent effort, one you can really be proud of. Starts up cleanly every time, manual reset works every time. I'm going to leave it there for now and tomorrow I'll try running some actual code on it. But looking great so far...

[Timbucus]

Well done - sir just finished mine - mad scramble to find an 74LS20 missed that extra chip - spotted it when I moved over all the working ones from my other machine to remove one possible source of problems. I also had to scavenge a power in jack from a recent binned breadboard PSU!

I have nothing on the display though so some work needed for me...

[Timbucus]

I made Link1 the 2111 type as well (and Added R16! - I used 5K though not 4.7K due to stock levels) but, I would not think it was that critical nor for R1/R11? And of course a 4Mhz crystal ready for the CPU.

[SiriusHardware]

Mine has a 4.00MHz crystal in at the moment although I have a 4.43MHz crystal on the way as I'm curious to see if this one can work with the uploader running at flat-out speed, or whether I will have to dial the speed down a bit like everyone else has so far.

I can't believe you have yours built already if you only got it today. I suppose aside from the ICs there aren't really too many other components on the PCB.

It's good that we both kicked off at the same time - with one of them definitely working at least we know that they both, in theory, should work.

[Slothie]

Originally Posted by SiriusHardware

Anyway, I expect you're all wondering whether the board works?

Great work, and very reassuring. It'll be interested to see how it works with the VDU. It looks great in black

[Timbucus]

Well mine now works. When I was testing the bare board for power and cross shorts on the Data/Add lines I noticed that I had Zero across the reset CAP - ignored it as a fluke.

Well two minutes with the scope and a strong clock made me start at the top and PIN 7 RST was Low.... hmm checked the routes via IC11 etc - swapped the LS00 as I had a batch of 10 with some faulty gates... nope - de-soldered the cap no still short.

Button seems to be OK up and down - continuity test shows it is permanent on - de-soldered it - which is VERY hard - took it apart, can't see any reason - put it back together works fine...

duhh seems they have an orientation as it is square pins... obvious once you see the inside of the switch...

[SiriusHardware]

Ian, yes, at least you know now that you have those five good PCBs waiting for you when you can finally get at them.

Tim, just to confirm, I also have the 2111 memory link made (as you might expect as my RAMs are AM9111s) and I have the JP2 (Sense-A?) link shorted, although the latter would normally only stop you from running entered programs if you missed that out.

Your display, has that been tested by any other means?

Edit: Aha! the old reset-switch-90-degrees-out-routine? We use that same type of switch in an old CPU PCB we still repair from time to time, so I'm used to not getting caught by that.

[SiriusHardware]

I predict that Tim will probably have his VDU PCB built and connected to the new PCB by this time tomorrow.

[Timbucus]

Cool I did not notice that link I was going to use my little adapter - which has that and my cassette interface plug on it. I had wondered what to do when I put the DIN connector on ... that is easier.

By display do you mean my Large LED one or my VDU PCB - I would hope to have time to build it tomorrow... not sure if that will happen though.

[SiriusHardware]

I meant the LED display, disregard. We cross posted so I didn't realise you'd already fixed it by the time I hit 'submit'.

You can fit a dinky SM pulldown resistor at JP2 rather than a link and that will make Sense-A available for user programs. Did you also notice that Ian made the reset line available, finally, on the keypad edge connector? Another thing SOC should have done from day one.

As you have the luxury of having keypad switches fitted, check out the contents of 0000, 0200, 0400, 0600 to see if the unwanted PROM images have gone. That's another improvement Ian made, in fact that's the reason for the extra IC, because he wanted to do that and still have the option to use the alternative type of RAM.

[Slothie]

The 2111 memory link is joined by default with a teeny trace because I assumed most people would use them. My prototype with the original board had this link broken and I tried the other kind of memory which fortunately worked. I wasn't able to test the RAM/io chip as it didn't fit with the heatsink so close! The chip has been moved to make space.

[Timbucus]

Good idea on the pulldown - I will have to buy some SMD - not something I use - maybe I can scavenge one off that board I got the Power connector from.

All those just return 0x1F now not ghosts so that seems to be working.

I did notice the RST was going to adapt my programmer to use it rather than the clips on the Cap - struggling to get that to work as is at the moment though. After a RESET it does not seem to enter the address to load at - maybe I am trying old HEX files.

[SiriusHardware]

Try leaving the reset clips off and just do a manual reset before each upload.

If that then works, it is probably still coming up out of auto-reset when the uploader starts sending the first address, in which case you need to increase the 'post-auto-reset' delay.

Also maybe worth just checking the keypad edge connections using key14 to see if the new board is reading individual external key presses correctly. I don't think Slothie ever had the opportunity or reason to test those connections on his original PCBs. I'll be connecting my own interface tomorrow, so Ill update with what happens for me.

Another thought, do your other machines have 4.43MHz crystals fitted? If so, this one will be running slower, so all delays will need to be lengthened a bit.

[SiriusHardware]

Originally Posted by Slothie

The 2111 memory link is joined by default with a teeny trace because I assumed most people would use them.

I expected that you might have made 2111 the default with a hair-link between the pads but, try as I might, I could not see it so I blobbed the pads just to be sure.

[SiriusHardware]

A quick update, I've had the new 'Slothie' Issue VI connected up to the pi-uploader, first of all I used the 'Key14' test utility to check the layout of the keypad connector and of course it was OK, all key presses individually sent to the PCB via the edge connector were coming up as the correct characters, commands or actions, including reset. This was to be expected as the timings for things like the length of a key press are much more generous in the key14 script than they are in the send14 script, where the delays were eventually pared to the shortest possible length consistent with everything still working - that was on my original issue II, 4.43MHz machine.

I used key14 to enter some test code, and naturally I chose the original test routine from #100 of this thread to honour the great man himself, image #1 attached.

The big surprise today was that the original 'send14' timings - including the original reset-hold and reset-release timings - work perhaps 98% of the time with this new PCB with a 4.00MHz crystal in it - I've uploaded the 'Message' program to it over and over and over again, and only once did I see a problem indicating that the timing was a bit tight, the letter 'J' of 'Jolly...' was corrupt on one occasion. I assume if I slow the timing a tiny bit that will then shift to 100% perfect entry. But otherwise, it's working better than I dared hope.

One way in which my PCB probably is different from Tim's is the four resistors near the upper left corner of the keypad and the four at the top right hand corner. The values originally given for these eight resistors is pretty arbitrary, all 1K2, but up to about 15K, I think.

My original MK14 has 4 * 4K7s for the four near the keypad and 4 * 6.8K for the four at the top right near the PROM, so that's what I have also fitted in the same positions in the new PCB.

[Timbucus]

Yes I had similar findings and had to slow send14 down a tad. Once I realised I was using the wrong keyboard connector (the one on the flying lead is for the JMP layout) I was away.

I have run MESSAGE (<256 bytes), LIFE(384 bytes in RAM/IO) and SEGTRIS (512 bytes BOTH RAM) fine on the board. One disappointment was that my INS8154 for the second MK14/SCRUMPI3 was dead as a doornail - too late to send it back now 10 months later... and I can't see anyone selling any either.

It has been a busy day with errands etc and not the least of which working out why my PC is resetting all the time - seems one of the disks in my array is failing... no spare to bring online either - so one on order and finding some spare spinning rust big enough for a backup run.

I did just finish the VDU and due to the risk with only one 8154 I rigged a few jumper wires in its socket to hold lines low as described in the manual. No video with or without the Char Gen - so looks like some work to do checking everything but, not tonight - I think I need to get out of this chair...

Looks like the V6 board is good though. One thing to watch out for is that the changes for NENIN (cutting link to IC10 etc) have been done on the board already but, it does need a pullup on NWDS to 5V - I fitted it on my VDU under the board as that comes through on the connector - I suspect others may find a place under the V6!