MK14 Rev VI PCBs

[SiriusHardware]

IanB_63, as stated at the thread beginning forum member Realtime is just now about to have another 5 issue VI PCBs made, so if you are interested as well, you might just make up the numbers. Try dropping him a PM, I don't know how often he visits the forum.

Yes, the Issue VI (a replica, there never was a real issue VI) is unique in having the system buses and control lines tracked back to the underside of the rear edge connector making it much easier to connect additional hardware like original or substitute VDUs. It also allows for the use of an obscure but currently cheaper type of RAM* and it has the RESET input brought out to one of the two upper connections on the keypad edge connector (originally both 0V connections). Like the original issue V and clones of the issue V, the unwanted images of the PROMs which were present in the address range 0200-07FF in issues II through to IV have been removed on the issue VI leaving that space clear to map other circuitry, typically more RAM, into.

*The RAM in question is not compatible with the original VDU, due to the peculiar way in which the SOC VDU handles the NRDS signal.

The problem with 'spotting' parts which are available to buy (and posting about them in public posts here) is that other interested parties soon hoover them up, so I think it's better to keep any news about new supplies of parts on a behind the scenes basis.

[Slothie]

Gerber files are at https://bitbucket.org/IanKRolfe/mk14/downloads/

Kicad sources are at the same site, and need the libraries in the "kicad_ikrlib" repository.

[Realtime]

The PCBs have arrived . They look great (better than in the photo!)

The green ones are accounted for but there is one spare blue PCB available. As I said before, it's a non-profit making exercise - it will be £10 including UK postage. PM me if interested.

[Mark1960]

That was fast, I guess green was a shorter manufacturing time.

I prefer blue myself, it just seems a bit easier on the eyes.

[SiriusHardware]

Is that five already spoken for, then, Ian?

Now that there seems to be a clear and present danger of some more of these being built, can I offer a hint about heatsinking of the regulator? You can get and fit a heatsink which is small enough not to encroach on the rear edge connector, but I feared that would be too small and I fitted a wider / longer heatsink which looks (image #1) as though it overflows onto the left hand side of the edge connector.

It doesn't, though.

I left the legs of the regulator full length and put a teflon spacer with flared ends between the heatsink and the PCB, the height of it chosen so that the ends of the uncut regulator leads only just pass through the holes in the PCB with a mm or so poking out through the other side. (Image #2).

The resulting gap allows an edge connector to slide onto the end of the PCB unobstructed, it also has the useful side effect of allowing the air to flow around the top AND underside of the heatsink.

[ScottishColin]

All - I'm being offered parts from various people - I need to collate the offers to see what I have left to buy; thanks to all for the offers. I'll sort myself out over the next day or so to collate a list. Work is getting in the way right now.

Colin.

[Realtime]

Originally Posted by Mark1960

That was fast, I guess green was a shorter manufacturing time.

I prefer blue myself, it just seems a bit easier on the eyes.

Actually all the PCBs went through manufacture together according to JLCPCB’s progress tracker. Quite amazing - 9 days from placing the order to receiving them. I had 5 green and 5 blue made. I wanted blue for myself so it matches the RealView PCB

[SiriusHardware]

So that is NINE, five green, four blue, already spoken for with just one remaining blue PCB available as of your post #63? I think I know where three of them are going but I am surprised there has been so much additional interest.

I just hope the parts can be found for them all...!

[Realtime]

, well there's a little smoke and mirrors in the sums. I've kept 2, another has taken a few but I'm expecting at least 5 of them to be built in the near future.
My aim is to set up a multiprocessor system: 2 x MK14 + SOFTY1 all accessing a shared 'RealView' module. (At least, that's the plan. Time will tell).

[SiriusHardware]

Mark1960 has form in this respect, as he has already tried multi processor setups. He's also had his MK14 clocked up to above and beyond 8MHz, I don't know if you found that thread yet...

One of my finds when trawling through Computing Today in 1980 was a short article describing a twin processor setup on an otherwise standard MK14 allowing the possibility to run both an application program and the monitor at the same time...

http://www.flaxcottage.com/ComputingToday/8009.pdf

(Page 45).

The article is by one G. Philips who I think may have been the gent who ran the UK MK14 owner's club which I sadly didn't know about at the time. Tim tracked him down a while ago and he was kind enough to upload all of the user group newsletters (that he still had copies of) to archive.org. Tim will know more.

[Realtime]

Thanks that's really interesting. I had had thoughts of creating a dual processor card to plug into the 8060 socket, which is exactly his approach.

Mark1960, I would be very interested in any info you have, particularly the pitfalls of doing it. I'll have a search for the past thread.

Edit: Just found the thread. 4 processors on an extension card!! I'll have a proper read of the thread but perhaps everything that can be done on this front has already been demonstrated?

[IanB_63]

Originally Posted by SiriusHardware

One of my finds when trawling through Computing Today in 1980 was a short article describing a twin processor setup on an otherwise standard MK14

I remember that article and did consider adding the second CPU to my MK14 but my ZX80 had recently arrived so that got my attention instead.

I ended up permanently converting my MK14 into a digital capacitance meter for some years using a 2716 EPROM to hold the code. It was surprisingly accurate as far as I can remember.

That was based on this project in Practical Electronics page 68:
https://worldradiohistory.com/UK/Pra...cs-1980-06.pdf

There is also a rather ambitious backplane project for the MK14 here on page 63:
https://worldradiohistory.com/UK/Per...0-04-S-OCR.pdf

[SiriusHardware]

Good finds, Tim resurrected the 'Minefield' game from the same article in the first link but I don't recall anyone ever mentioning the nearby capacitance meter project in the past. SOC did forsee that the machine might be used for semi-permanent fixed applications the way we use Arduinos now - that's why a PROM programmer was one of the peripherals made available, so you could do exactly that. Of course, blank DM74S571 PROMs were widely available and moderately cheap in those days.

Single units were quite often reprogrammed as single-task units, I think CfCH has one such MK14 cased up in a tough wooden enclosure with a proper keypad - some sort of medical research, I think that one was used for.

I saw a comment under an MK14 related youtube video in which someone said they had used MK14s commercially as dedicated controllers at the heart of (stage?) lighting control units, in much the same way as you might use a Raspberry Pi Compute module or Arduino for the same purpose now. It made me wonder how many original MK14s may yet be lurking inside dusty, disused lighting rigs in the backrooms of theatres, music venues etc. Unfortunately the commenter failed to mention the make and model of the units involved, so we may never know if the story was true.

[Mark1960]

Originally Posted by Realtime

Thanks that's really interesting. I had had thoughts of creating a dual processor card to plug into the 8060 socket, which is exactly his approach.

Mark1960, I would be very interested in any info you have, particularly the pitfalls of doing it. I'll have a search for the past thread.

Edit: Just found the thread. 4 processors on an extension card!! I'll have a proper read of the thread but perhaps everything that can be done on this front has already been demonstrated?

Not sure if we should start a new thread or request the previous one to be reopened. There is still quite a lot that could be done.

A couple of pitfalls I remember. As you add processors you increase the latency between busrq and busak through the nenin/nenout chain. First processor is ok, second and third take an extra cycle but the fourth takes two extra cycles. If using the atomic instructions to implement a mutex the MK14 vdu’s method of booting the 8060 off the bus would break up the read/write cycle of the atomic instructions and possibly allow a different 8060 to take control of the bus. I made a start on trying to replace the arbitration but didn’t complete that or try to build hardware. I later found a datasheet for the 74F786 that could possibly be replicated in discrete logic, but I wasn’t convinced the logic diagram was correct.

I didn’t do much on the software side. I designed the 74HCT109 circuit to make it work with the standard SoC monitor. On start up only one processor runs, when you use G to start a program the active processor goes to that target and another starts the monitor. A simple OS could be developed with a process list where any 8060 could run any thread, but more memory is needed. I made a board that adds 32k in 2k pages and replaces the SoC monitor with ram but still not tested this with the multiprocessor.

[SiriusHardware]

I gather the new batch of issue VI PCBs have made it into the hands of at least some of those who want them. I've just been looking at prices for the INS8060 and INS8154 today.

Freeowww!

I hope the guys building them are getting them for better prices (but if you are, don't tell me where - not on here, anyway).

[Slothie]

Getting INS8154's is becoming a real problem now. I'm lucky a kind forum member got me the one I needed for my MK14, and I have one for the Acorn 1 I will someday build. I do have one in my storage and some SC/MP's but not sure when I'll get access to that and I bet its right at the back on the bottom!!

[ortek_service]

Originally Posted by Slothie

Getting INS8154's is becoming a real problem now. I'm lucky a kind forum member got me the one I needed for my MK14, and I have one for the Acorn 1 I will someday build...

Yes, the only place that I was aware of still stocking these at a reasonable price (especially for the ceramic package version they had), now seem to have finally run out (They had around 150off a couple of years ago, but were down to 7off a couple of weeks ago, before these all got snapped-up to complete a last-few Acorn System 1 replica kits). Although, maybe they might unearth some more as they do say to contact for availability:
https://www.rockby.com.au/SearchresS...rchkey=ins8154

They were never the most-widely used PIA IC (and Acorn seem to have hedged their bets on some of their system cards like the VIB, also using the 8255 & 6522 - In addition to using an 8154 for Keyboard / display / cassette etc elsewhere on the System 1 etc.)

If the internal 128-Byte RAM isn't required, then it seems the 6821 / 6522? (maybe even the 8255) might be register compatible for the I/O, and I'd seen some Acorn system replicas using these. But would need to make a pin out adapter board-module (and may be some firmware changes might be required, if registers aren't quite identical).
Might also be able to use a 6810, to provide the internal 128-Byte RAM.


It should also be possible to emulate one using an RPi-Pico uC, with its built-in PIO 'Co-Pro' module (I saw someone had managed to do 10Mbps Ethernet by using this to bit-bang, to avoid need for a proper Ethernet I/F Controller)
And may also be able to get it to emulate the internal 128-Byte RAM.
But it may not have quite enough pins to fully map to all of 8154's pins, so some external port-expansion would be required - Probably would need to have some level-converters for operation on 5V systems, anyway.

[SiriusHardware]

It's a shame they didn't choose the 8155 - 2 8-bit ports, one 6-bit port, 256 bytes of internal memory and an independent internal timer peripheral all in a 40-pin chip.

In the early nineties I worked on oldish equipment which made extensive use of 8085-based microprocessor control boards with at least one 8155 on them so I have quite a few in my 'chip museum', along with a number of 8085s of course.

Only problem is that the address / databus are multiplexed which makes it ideal for the 8085 but probably less so for the SC/MP.

Given that the issue VI has the system signals on the rear edge, there is nothing to stop anyone from coming up with the mother of all multi-i/o expansion boards with their choice of parallel, serial, counter-timer etc. peripheral ICs on it which just plugs into the rear edge connector.

[ortek_service]

Yes, I'd previously looked at the 8155, initially thinking it might be a good substitute.

Although the 8155's are maybe now getting quite expensive, as may be obsolete too, never seeming to be as popular as the the 8255 (that I've a few of / I used a couple in my own Beeb fully software controlled Universal EPROM-Programmer design back in the 80's).

However, the unusual multiplexed bus (In order that they had some spare pins to give an extra 6/8th's of a full 8bit I/O port ?) does rather rule-out its use, as by the time you've made something (with a uC?) to convert the different protocols, it would be easier to just get a uC to fully-emulate an 8154.

Or, as a PIA is relatively simple, use a CPLD/FPGA to implement one - Maybe using a cheap Lattice device (And might find one with enough cells to give the 128Byte RAM as well, Without resorting to now mostly much larger & expensive Alteras & Xilinx devices )

[Realtime]

Chris Oddy has a couple of 8154's for sale at the bottom of this page. They are the beautiful purple and gold ceramic ones (yes, I know it doesn't make them run any better but still nice to look at!).

And if anyone feels the need to pay proper money for a white and gold ceramic SC/MP II look here. There's no mention that it's been tested though.