ZX81+35 Clone

[mahjongg]

When I say "low cost", I mean low cost for the parts for the hobbyist who wants to build one for himself.
I don't see it as a viable "product". only very few people would be interested in them, but I would like the total price of the "kit" not to be a barrier. so cost considered is for the parts, not for labour cost of assembly.

Problem with the shell-less db-9 is that for some strange reason they are much more expensive, and more difficult to find, probably because they are much less used. the most used, and therefore far cheapest connectors are simple connectors like this:


my original idea was that using a simple single row header with 8 pins (the number of signals used) would allow detaching the above connector. and the pinout of a DB-9 connector would not. detaching it will be important during debugging the board, and handling it before you have a housing for it.

If I can fit it the 2x5 pin header seems the best solution, those cables/connector sets are cheap, and can be assembled by hobbyists. the single 8-pin row is a bit of a hack, and a directly mounted connector is either very expensive (shell-less) or doesn't fit on the PCB.

Still if I use the pattern of the connector pins of the shell-less DB-9 then you theoretically can use the connector pictured above, either with eight flat-cable wires, or directly with stiff wires.
it might also be possible to buy a cheap connector with a shell, and cut away the shell parts that conflict with the components on the board, but that would be a big hack.

Or you could put the connector shell on top of the parts below so lift the connector a few mm above the board. to be able to do that the pins of the connector must be long enough, and the shell should not have mechanical pins for 3mm holes.

[mahjongg]

An angled DB9 connector with plastic shell mounted on top of the SMT components is a cheap solution, and would look like this:

ZX81+35 revision 2.2 with angled db9 preview.png

(290.12 KiB) Downloaded 287 times

As long is the connector has pins long enough to solder to the board, even when lifted up a tiny bit, this would work.
It would block access to the components below which would make repair/exchange of the crystal (or other affected parts) difficult, so I would only solder on the connector after you have verified the system is working.
I will try to fit the 9 pins of the DB-9 to the board.

My decision is that this is the best solution.
If you dont want to do this, it is still possible to use a shell-less DB-9, or use the cheap DB9 with soldertags, and connect it with either stiff or supple wires

[mahjongg]

Well, I did some work, and managed to place the DB9 connector, had to swap the position of the FET's as pin 1 is now in the underside, rather than the topside, as it was in the past. All that remains of the connector are the nine solder holes.
also managed to pre-place the four new SOT23-5 devices, (U25, U26, U31 and U32) and locations for the new SOIC-14 IC's, (U8, and U21) but as you can see in the picture below U21 (the SOIC with the six inverters) is really in an unsuitable location, too many long connections are needed, so I have to throw away some wiring, and shuffle some of the SOIC IC's around for a more optimal configuration.

The bottom picture shows how the layout situation is at the moment, yellow lines are "ratnest" lines showing connections that need to be made.
Red circles indicate where there is an error (short) in the layout. Note that I removed the copper fills (ground plane fill) which caused Ultiboard to throw a hissy fit, and it declared a lot of my ground traces as unconnected, I probably have to find them, remove them and lay them again. Sigh...

Situatie_23_December_2015.png

layout situation yesterday.

(135.71 KiB) Downloaded 266 times

I probably will continue with the layout work next year.
Revspace has the latest schematics, with the DB9 instead of the 1x8 header.

[mahjongg]

Finished!

REV 2.2 FINISHED LAYOUT 29 DECEMBER 2015.png

(196.11 KiB) Downloaded 225 times

[mahjongg]

this is the finalized REV 2.2 ZX81+35 schematic, notice I swapped some inverters for optimal layout, added a few decoupling caps, and placed shorts over the pull down ( ROM section select and RAM halve select) so you don't have to put pull-down-resistors on the back of the board.
[attachment=0]ZX81+35 REV 2.2.png[/attachment]

[kubi]

Hi,

I'm a bit confused concerning the decoder for !OUT EF. Schouldn't A4 be used instead of A7? Or am I wrong?

[mahjongg]

you are right, its decoding 0x7F, not 0xEF, not sure if I meant to decode 0x7F, or that I should have used A4, but I will investigate my notes, thanks for the TIP!

[RWAP]

Hi,

I'm a bit confused concerning the decoder for !OUT EF. Schouldn't A4 be used instead of A7? Or am I wrong?

Well spotted - and welcome to the forums

[mahjongg]

Right, I went back to my design notes, and found that initially I wanted to select on the next available address line (A4) so the I/O address would have been 0xEF, and at that time I probably created the /OUT_EF label. Later I became concerned that using an obvious "next" address line would conflict with some expansion device, even though I could not find any evidence for that in several documents I read, but just to be sure I used A7 instead of A4, and forgot to change the label, so the labels should read /OUT_7F, not /OUT_EF.

Unless anybody sees any reason not to use I/O address 7F I will just change the labels used in the schematics.

for example, does anybody know if any recently designed device uses OUT_7F ?

[mahjongg]

Well, I repaired the error in the schematic, here is the corrected schematic:

ZX81+35 REV 2.2-1.png

corrected out_FE to out_7F

(672.63 KiB) Downloaded 169 times