Sinclair ZX80 / ZX81 / Z88 Forums

I am in the process of building more of my reproduction Ram packs for the ZX80 and recently received 3 more PCB's from OSH Park of the same design as the previous order. I am populating the board with the same components as before (18-pin dual-wipe sockets, 5 VOLT 1KX4 450 NS DIP-18 STATIC RAM, 74LS138 1 of 8 Decoder/Demultiplexer IC in a 16-pin socket, and the two capacitors.

The first board I built worked perfectly, just like the previous batch. The two subsequent boards, however, refuse to work. No magical "K" prompt appears. I have swapped chips from board to board without success. I next printed out the PCB layout files and meticulously did a point-by-point check of all connections, finding no shorts or dead spots.

Right now, I am dazed and confused as to what to try next. I do not have either the equipment or knowledge to do further testing.

Suggestions will be most welcome.

Regards,
Mike

Hi Mike

mhudson52 wrote: ↑Thu Nov 22, 2018 9:09 pm I next printed out the PCB layout files and meticulously did a point-by-point check of all connections, finding no shorts or dead spots.

Did you check/measure on the PCB or on the contact inside the socket (maybe soldering is the problem)?
Soldering all components again often solves such problems.

Siggi

PS: could you show some pics of the populated board (both sides)?

@Mike - try your boards with NO RAM chips installed, just the 74LS138 decoder fitted.

The machine (ZX80 or ZX81) should start up using it’s internal RAM controlled by the 74LS138 decoder on your boards.

I’m assuming that your boards are similar to the boards in this thread and this thread.

Mark

Mark:

I tried your suggestion using only the decoder installed and everything looked rosy (got a nice "K" prompt. The board I am using is based on the Markus Brenner design. The same board as discussed in the first link but corrected by Charlie.

Siggi:

All the connectivity checks I did were with the chip sockets soldered to the board. Also checked to make sure there were no shorts to adjacent connections and found none.

Photos attached.

Thanks, guys!

Mike

So the next easy step, is to fit just two known good SRAM chips, keeping in mind that as per the discussion in one of those threads, that one has to be suitablely fast. Then try it again.

Which Sinclair machines do you have? And if you have one or more ZX81s, which type of RAM chips do they have, if they use 2114 type, are these socketed?

Mark

The only machine I have to test on is my TS1000, which has shown to be completely compatible with all my previous Ram pack builds.

Due to availability issues at the time I ordered the last batch of RAM chips, what I am using now are 450ns chips which have shown to work in past builds. The 200ns have become available once more and I have another order of them on the way. Should I just wait on them before any further testing?

Which sockets should I use if fitting only two chips?

Mike

Maybe it does not solve the current problem, but usually each chip should have its own capacitor (around 10-100nF), connected across its power pins to stabilize its power supply. If those C are missing, the board may run (because the logic is OK), but not very stable (because the power supply of a chip breaks down for a short period, when the chip draws a lot of current during an internal state change) ...

Siggi

mhudson52 wrote: ↑Sat Nov 24, 2018 2:21 pm Due to availability issues at the time I ordered the last batch of RAM chips, what I am using now are 450ns chips which have shown to work in past builds. The 200ns have become available once more and I have another order of them on the way. Should I just wait on them before any further testing?

Which sockets should I use if fitting only two chips?

@Mike

See the extract of the Sinclair instructions shown below: Note also the two different speed rated memory chips, 2114L in the 'A' positions and the 2114L-1 in the 'B' positions.

Here is a photo I pulled from the internet archive of a Sinclair 3K RAM Pack: 2114 SRAM timing / speed information

2114, 2114L ............. 450ns max. access time (tA or tACC)
2114AP-12 ............... 120ns max. access time (tA or tACC)
2114AP-15 ............... 150ns max. access time (tA or tACC)
2114-1, 2114L-1 ........ 150ns max. access time (tA or tACC)
2114-2, 2114L-2 ........ 200ns max. access time (tA or tACC)
2114-20, 21L14-20 .... 200ns max. access time (tA or tACC)
2114-25, 21L14-25 .... 250ns max. access time (tA or tACC)
2114-3, 2114L-3 ........ 300ns max. access time (tA or tACC)
2114-30, 21L14-30 .... 300ns max. access time (tA or tACC)
2114-4, 2114L-4 ........ 450ns max. access time (tA or tACC)
2114-45, 21L14-45 .... 450ns max. access time (tA or tACC)

The type marked "2114L" or "21L14" are lower power versions (typically 70mA, 370mW rather than 100mA, 525mW). The "2114AP" are also a low power type.

I hope this helps

Mark

siggi wrote: ↑Mon Nov 26, 2018 10:20 am Maybe it does not solve the current problem, but usually each chip should have its own capacitor (around 10-100nF), connected across its power pins to stabilize its power supply. If those C are missing, the board may run (because the logic is OK), but not very stable (because the power supply of a chip breaks down for a short period, when the chip draws a lot of current during an internal state change) ...

@Siggi, yes, one decoupling capacitor per IC is good practice. This RAM board is however a recreation of an existing Sinclair design. So it "should" work if suitable 2114 SRAM chips are used.

Mark

1024MAK wrote: ↑Mon Nov 26, 2018 5:00 pm

siggi wrote: ↑Mon Nov 26, 2018 10:20 am Maybe it does not solve the current problem, but usually each chip should have its own capacitor (around 10-100nF), connected across its power pins to stabilize its power supply. If those C are missing, the board may run (because the logic is OK), but not very stable (because the power supply of a chip breaks down for a short period, when the chip draws a lot of current during an internal state change) ...

@Siggi, yes, one decoupling capacitor per IC is good practice. This RAM board is however a recreation of an existing Sinclair design. So it "should" work if suitable 2114 SRAM chips are used.

Mark

The first new board worked, so number 2 and 3 also "should" work
The issue 1 board above has a big ground plane next to the connector and maybe below the ram chips. Does the issue 3 board also have such a ground plane?

Siggi