Tiny SC/MP system !!!

[ortek_service]

Originally Posted by Mark1960

If you read the spec on those FRAM you will probably find they need some time between memory cycles. You won’t have that gap with the MK14 address decoding. You could try to gate the chip select with either NRDS or NWDS, as per Slothie mod for 65X61 ram.

This might also limit the speed a Z80 can run with these devices.

Thanks for the info. The speed of the FRAM's shouldn't be too much of a problem as ones I have are 70ns - Which I presume is write as well as read?
And I don't think they made much-slower than that ones.
So it does look like maybe some timing of the signals from the SC/MP is the issue, and may try adding the extra AND-gate that Slothie used to support the 65X61, to see if that works - Should be able to fit an SMD Tinylogic one, on a breakout board tucked somewhere under this.

[ortek_service]

Originally Posted by Phil__G

Hi Owen, I've collected all mine into one zip, with a folder per project.
Each project has its own zip with the files and text describing the project and to show how the equations were derived.
For now they're at the bottom of my SC/MP page
When I get a mo I'll do a table with a description of each, presently the descriptions are with the files in each project folder.
In fact I might do a full page on "Using GALs the easy way without having to learn WinCUPL"

Cheers
Phil
>>

Thanks for that - I'll take a look.
And maybe try re-compiling one, sometime, as it's been > 30yrs since I last did this (at Uni, with an Intel(=Altera) UV-Windowed Ceramic DIL PLD, making a BCD to 7-seg display decoder IIRC)
And would probably try to include as many comments as possible into the source files, to make it clear exactly what memory-map is being implemented.

Originally Posted by Phil__G

>>

Cheers
Phil
PS there are some fiddly mods to do to run the Martin.cz board standalone - I'd recommend Kris's PCB as it incorporates
all these mods and has an FTDI header, or if you like, my already-modded Martin.cz is redundant now so I could
send it to you, just plug in the chips?

Thanks for the offer, but I'd already got Kris's ones from PCBway. And had assembled a couple - One for use on MK14, without the crystal, and one without the SC/MP header pins for stand-alone use with kitbug(+) etc.

Note: for anyone assembling these for MK14-use, you have to solder things in the right order (double-ended round pin SIL header strips in first, ensuring that thicker-end goes into the PCB) before the 40way DIL socket and also use a turned-pin one for that to have thin-enough walls / standoff from top-side of those pins (even when cutting them flush to joints). Then use a stack of 40pin DIL sockets to space this board high-enough above the MK14's axial electrolytic.
Or you may be able to use the special DIL long double ended pins, designed for this type of plug-in board, but harder to find

[Mark1960]

Originally Posted by ortek_service

Thanks for the info. The speed of the FRAM's shouldn't be too much of a problem as ones I have are 70ns - Which I presume is write as well as read?
And I don't think they made much-slower than that ones.
So it does look like maybe some timing of the signals from the SC/MP is the issue, and may try adding the extra AND-gate that Slothie used to support the 65X61, to see if that works - Should be able to fit an SMD Tinylogic one, on a breakout board tucked somewhere under this.

The access time is only 70ns, but so is the minimum time between write cycles. This would be ok on 8060 if chip enable is qualified by NADS or NWDS. On a Z80 there is not enough delay between MREQ cycles when writing 16 bit registers to ram, it might be possible to clock MREQ to delay the start of every write cycle but I gave up at that level of complexity.

[Phil__G]

Originally Posted by ortek_service

Or you may be able to use the special DIL long double ended pins, designed for this type of plug-in board, but harder to find

I got mine from here Owen: https://www.ebay.co.uk/itm/254388299710




(Graham Station X has said its ok to post a link to a requested item)

[Mark1960]

When using those male SIL header pins on a daughter board I solder a turned pin socket or female socket strip onto the daughter board then plug the male SIL pins larger diameter side into that socket. These are a tighter fit into the daughter board socket than the target board socket, but still better to tie the male header pins to the daughter board socket using wire wrap wire to stop them separating. This raises the height of the daughter board above any other socketed ICs on the target board. Also means you don’t have to desolder the header pins from the daughter board when you break one of the pins.

[ortek_service]

Originally Posted by Phil__G

I got mine from here Owen: https://www.ebay.co.uk/itm/254388299710


(Graham Station X has said its ok to post a link to a requested item)

Thanks for the link - and co-incidentally I've just used that seller on there for the 4.3mm-high 'Omron B3F' compatible flat-top push-switches for use with Realtime's keyboard frames/overlays.

However, these appear to be the same low-height SIL types that Farnell / Rapid etc. sell upto 32 / 40way strips I used on this, that you need to stack 40way DIL sockets under to get enough height (and also help protect the brittle pins).

I had a BBC Computer FDC-adaptor daughter board with a 40w DIL thick round pin standoffs section spaced-off double-header on it, as shown on picture I found on the 'net. I've dig mine out, and found it had "Scott" (Who 30years ago we used their similarly 'green' coloured PLCC-socket extractor tools - were quite expensive, but didn't damage the Socket / IC that many other thicker puller pins did) stamped in the moulding, as shown on the attached.
But I can't find any trace of "Scott" on the 'net, so not sure what happened to them.

I have discovered Harwin made D9xxx series "DIL Through Headers", that Farnell sold 14 & 16 way ones many years ago. But could also get 40w.~
However, many of these were obsoleted by Harwin in 2014 (Although maybe Gold etc. ones are still made, as only some type numbers on the obsoescence list). However, these are just DIL versions of those low-height SIL ones, and so you need to stack IC sockets onto (Although that at least protects the brittle pins, when not plugged into something).

So it would be good to find anyone still selling 40way pre-spaced off double round pin headers (I had to bodge the broken end pins one on my Watford Electronics DDFS board, by plugging a "dual-wipe" IC socket onto end, then dropping solder down to connect broken-off pins into socket contact.

Originally Posted by Mark1960

When using those male SIL header pins on a daughter board I solder a turned pin socket or female socket strip onto the daughter board then plug the male SIL pins larger diameter side into that socket. These are a tighter fit into the daughter board socket than the target board socket, but still better to tie the male header pins to the daughter board socket using wire wrap wire to stop them separating. This raises the height of the daughter board above any other socketed ICs on the target board. Also means you don’t have to desolder the header pins from the daughter board when you break one of the pins.

I think I had considered that, but couldn't force the thick-end of the pins into those sockets.
Also, fitting a DIL socket one side, can make soldering access for the one on the other side a bit more difficult, if solder joints are now under the socket although thinner walls and slight spaced-off on shoulders of the turned-pin ones may help a bit.
Always keeping an IC socket on the ends of those double-pin headers, does help prevent their pins getting broken (with the socket going first if also using fairly-brittle pin turned-pin types that don't normally survive being straightened after getting bent).
And is also generally essential when using separate SIL strips, to keep them aligned to 0.6" width (especially when soldering these in, to start with)

BTW, I've just discovered these IC-like pins, for when trying to make smaller IC-replacement modules (There was a previous-thread somewhere, about using FPGA's etc for this and best pins to use):
https://shop.evilmadscientist.com/productsmenu/855
https://www.digikey.ca/en/products/filter/contacts/leadframe/416

[Phil__G]

Originally Posted by Owen

I used on this, that you need to stack 40way DIL sockets under to get enough height

I have nothing under the daughterboard so its as low to the PCB as it will go (8mm?), get rid of the big electrolytic
On a dev board I'm not concerned about aesthetics, if I was I'd mount the daughterboard underneath and use corner stands on the main pcb

[Realtime]

Originally Posted by Phil__G

I got mine from here Owen: https://www.ebay.co.uk/itm/254388299710




(Graham Station X has said its ok to post a link to a requested item)

Same here. Been using Switch Electric for a number of years - always good quality and fast delivery.

[ortek_service]

Originally Posted by Mark1960

If you read the spec on those FRAM you will probably find they need some time between memory cycles. You won’t have that gap with the MK14 address decoding. You could try to gate the chip select with either NRDS or NWDS, as per Slothie mod for 65X61 ram.

This might also limit the speed a Z80 can run with these devices.

Originally Posted by Phil__G

I found the Ramtron FRAMs were absolutely perfect on the PICL where access is slow because of the emulation, but the very same chip wouldnt work on any of my three Grant Searle Z80s (7.37 Mhz, no waits).

Incidentally, RS still have loads of bytewide M48Z08 NVRAMs in stock, and confirmed deliveries of M48Z35 later this year... Farnell are a bit more expensive but have over a thousand M48Z08 and hundreds of DS1245's, 1230's & 1225's. You might sniff at the price but I bought a Nascom 32k memory board in 1980 so I know these chips are excellent value at £20-£25

Or, buy an AS6C1008 128k x 8 cmos static ram for £2, a DS1210 for another £2 and make your own!
this is what one of my three GS Z80's uses, gives two switchable nonvolatile 64k pages!

Well I do have a few proper NVRAM's - With dead batteries by now, but could probably resurrect, converting to coin-cell holder etc.
But it would be nice to use a maintenance-free FRAM.
- Inc. with Z80's, even if needing to add some wait states that PC's etc have historically relied on / I recall early ones having fixed delay-lines.

I have recently obtained some 128K SRAM's, to complete a Z80 MBC2 board, I've been assembling. But will have to check if they have a very low power standby mode.

[Mark1960]

I don’t think a wait state would be needed for FRAM unless running the z80 at a higher clock frequency. In fact a wait state would not help as it will not increase the time between Chip enable going high and the next Chip enable low. Minimum of 60ns precharge time with chip enable high between each access. Also a maximum of 2,000 ns chip enable active, so could not be used with simple single step debug.

[coolsnaz2]

I have modified a Tiny SC/MP board to accept NVSRAM (DS1245Y) in place of the RAM/ROM chips and the GAL16V8 has been replaced with the GAL22V10 version.

2 fly leads are required to give the MK14 access to 54k of NVSRAM. My current configuration in the GAL gives me 4k ROM and 50k RAM.
The GAL is being used to decode A12-A15 lines from D0-D3 using the NADS pin on the SCMP.

The 2 fly leads are used to select only page zero for the Keyboard/Display D00, and the I/O chip at 800. The shadow memory locations for the Keyboard/Display and the I/O chip have been maintained on page zero.

Slight modification are required to the MK14 board, two IC’s require a pin to be removed from their socket and the appropriate fly lead inserted into the socket. Issue V and VI MK14 replicas require slightly different locations for the fly leads to be connected.

Currently my VDU card cannot select just page zero, so the VDU memory at 0200-03FF is shadowed across all pages. Unfortunately this restricts the amount of continuous memory to 3k on any one page. This limitation may be fixable in the future.

Next step is to create a new PCB for the modified board.

[Phil__G]

Sounds good Ian, its similar idea to the Max, Ian (Realtime) tells me you were one who taught us how to do address latching, thank you, that makes such a difference and will probably feature in most designs from now on!
So much quality scampery happening at the moment!
What next, pageable ram and a disk operating system?

[Phil__G]

I forgot to say (timeout again) that this makes a full-house SC/MP possible with 3 chips!
The 8060, a DS1245Y for ram & rom, and a gal for latching, decoding and R/W gating.
A three chip, fully-expanded system. Amazing!

[Mark1960]

Originally Posted by coolsnaz2

I have modified a Tiny SC/MP board to accept NVSRAM (DS1245Y) in place of the RAM/ROM chips and the GAL16V8 has been replaced with the GAL22V10 version.

2 fly leads are required to give the MK14 access to 54k of NVSRAM. My current configuration in the GAL gives me 4k ROM and 50k RAM.
The GAL is being used to decode A12-A15 lines from D0-D3 using the NADS pin on the SCMP.

The 2 fly leads are used to select only page zero for the Keyboard/Display D00, and the I/O chip at 800. The shadow memory locations for the Keyboard/Display and the I/O chip have been maintained on page zero.

Slight modification are required to the MK14 board, two IC’s require a pin to be removed from their socket and the appropriate fly lead inserted into the socket. Issue V and VI MK14 replicas require slightly different locations for the fly leads to be connected.

Currently my VDU card cannot select just page zero, so the VDU memory at 0200-03FF is shadowed across all pages. Unfortunately this restricts the amount of continuous memory to 3k on any one page. This limitation may be fixable in the future.

Next step is to create a new PCB for the modified board.

If you have enough gates left in the GAL you could possibly gate the NRDS and NWDS to the 8060 socket on the MK14. I did something similar using a 74hct257, either 4k of the MK14 board enabled, or top 2k and low 2k is battery backed ram on the daughter board. This would avoid the flying wires and any modification on the MK14.

Not sure this would work with the vdu, can the GAL outputs be tristate?

[coolsnaz2]

Originally Posted by Phil__G

What next, pageable ram and a disk operating system?

I have adapted the Fast Loader to load and put a hex file into a page of the MK14 NVSRAM, and another adaption of the Fast Loader takes it from the page of NVSRAM memory and puts it in page zero.

Originally Posted by Mark1960

If you have enough gates left in the GAL you could possibly gate the NRDS and NWDS to the 8060 socket on the MK14. I did something similar using a 74hct257, either 4k of the MK14 board enabled, or top 2k and low 2k is battery backed ram on the daughter board. This would avoid the flying wires and any modification on the MK14.

Not sure this would work with the vdu, can the GAL outputs be tristate?

Mark, thanks for the suggestion, I may try out your idea on Issue II of the board, Issue I already in production. I was thinking I could interrupt A11 going back to the main board unless we have page zero selected. It would be great not to need the fly leads. I don't think the GAL can do Tri state outputs, so another chip would be required.

[Phil__G]

16v8 does tristate Ian

[ortek_service]

Originally Posted by ortek_service

Thanks for the info. The speed of the FRAM's shouldn't be too much of a problem as ones I have are 70ns - Which I presume is write as well as read?
And I don't think they made much-slower than that ones.
So it does look like maybe some timing of the signals from the SC/MP is the issue, and may try adding the extra AND-gate that Slothie used to support the 65X61, to see if that works - Should be able to fit an SMD Tinylogic one, on a breakout board tucked somewhere under this.

Well I've now tried doing that.I found that nWDS goes to pin27 (nWE) of the RAM when using this SC/MP RAM/ROM replacement board, so all signals required were on the RAM's DIL28 socket and could just stack 3 sockets and break connection on pin 22 (nOE) on the middle one / wire to pins on top & bottom sockets.
And I used a TC7S08F SOT-23 single AND-gate, as had some to hand, on a small SMD breakout board that fitted in the centre of the sockets with a bit if filing.

But unfortunately this still didn't work on the MK14 + SC/MP RAM/ROM replacement board (or the MGH8060 I also tried it on).
The AND-gate I'd used was a High-Speed CMOS type, rather than HCT / TTL, so a little bodged. But I did try it with the original 6264/62256 RAM IC through the adaptor and that still worked OK, so must have been happy with logic-levels.
And I also tried adding some pull-ups to the inputs, in case it wasn't quite being driven high-enough voltage for logic-high by the existing TTL-compatible circuitry (although SC/MP is NMOS, so may be a bit higher).


So it looks like something a little more involved is required.
I wonder if some extra gates in the GAL, possibly with an R-C delay circuit. might be usable?

However, the existing circuit I'd added was between mRDS and nOE of the FRAM, as with the 65X61. Whereas IIRC, it is the nCS pin (selected only by gating some of the address lines), that requires cycling for a short period for writes to work. And may need to also gate this nCS enable with nWDS / nRDS, to make this happen?

This will probably need a bit more-thorough investigation with a 'scope / looking at FRAM's datasheet's timing spec's.

[Mark1960]

On Slothie’s schematic the pin labelled output enable is actually chip select on the 65X61.

Selecting output enable on 6264/62256 type ram when write enable is active is probably not a problem on most of these ram types, as write enable will override output enable.

If the FRAM is the 32k type in 28 pins it only has one chip enable, so to qualify with nWDS or nRDS in a single chip would need 74x1g0832, or in two chips a 1g08 followed by a 1g32.

[ortek_service]

On Slothie’s schematic the pin labelled output enable is actually chip select on the 65X61.

Well on the Slothie v1.2 one previously linked-to / attached on a 2111/2 thread, pin9 (Output Disable on 2111) is simply labelled 'E' on Slothie's MM2111 symbol. But (like CE1 and CE2 inputs) it has an 'Inversion' circle on this pin.

So that does basically match what the datasheet calls them: https://www.tvsat.com.pl/pdf/I/IM65X61.pdf
With pin 9 'nE1' and the (n)CE1 & (n)CE2 called 'nS1' & 'nS2'
(Although they've made an error in the description, saying 65X61 doesn't have nS2, when they really meant nE2)

However, nE on the 65X61 isn't nOE or a chip select, but a strobe input (more of an 'E' clock input as on 65xx/68xx systems) that on a falling-edge latches the address lines (which aren't really edge-triggered themselves, as datasheet description indicates), but needs to have a rising-edge back to high before data is output (/ input for writes) on the RAM IC. And data remains at the output until nS1/2 are taken high, so the 65X61 doesn't really have an nOE input.

Therefore the Slothie added AND gate of nWDS with nRDS isn't controlling the Chip Select, but the 65X61's more unusal Strobe input that was OD (nOE) on the 2111.

If the FRAM is the 32k type in 28 pins it only has one chip enable, so to qualify with nWDS or nRDS in a single chip would need 74x1g0832, or in two chips a 1g08 followed by a 1g32.

Yes, that is the case on the FM1808, and also on the FM1608 8KB one / most 28pin Memory IC's.

But if the Chip Select is more f a strobe on these, then I probably need to try gating signals on that pin, instead of nOE, on this.

[coolsnaz2]

Originally Posted by Phil__G

16v8 does tristate Ian

Thanks for that Phil, still learning what the GAL can do.


Just an update on the MK14-NV board, which was originally based on the Tiny SC/MP system.

Boards arrived from JLCPCB, and one built and working, see attached photos.

I still have the 2 fly leads for 0D00 (Keyboard/Display) and 0800 (INS8154 chip), but may look at changing this in future developments as suggested by Mark.

I have a few boards available for UK based forum members, just PM me with your details.