Sinclair ZX80 / ZX81 / Z88 Forums

So. The results are in. The Spectrum is now working correctly (it was sent to Alex to test the ULA and well.... I couldn't just send it back knowing it didn't work right, could I?!).
The first thing I did was do the mandatory modifications in the service manual, which I thought may fix it - but no. It was then - after far too long staring at it - that I realised what the problem was.

On page 1, there is a nice large photo of the board. Cast your eye on the top row of the upper 32KB RAM. And then scan just a little bit left...

When you see the fault, you'll wonder how you missed it....

Hurrah for Sinclair quality control! It was like this since 1982!

Missing resistor?

Yup. Whoever constructed this particular board in 1982 obviously felt that some of these resistors were completely optional. The vias were completely filled with original solder, so it hadn't been removed - they'd just never fitted it. This resistor is the 8K2 pull-up resistor on the D2 line (which is, of course, also the data line for the '8' key). As it wasn't pulled up, it was left floating, and caused havoc when anything else (e.g. a DivMMC) was connected to the data bus.

I guess some manufacturing defects take longer than others to come to light!

Well spotted! And of course many thanks.

I wonder if the factory (which of course was not actually Sinclair owned or operated) actually tested units with any kind of expansion before boxing them up?

Mark

Lardo Boffin wrote: ↑Fri Oct 20, 2017 9:29 pm I guess some manufacturing defects take longer than others to come to light!

It's worse than that! In the industry where I work (nope, can't say), sometimes construction/manufacturing defects/design issues are found with safety critical systems... Including logic systems dating from the 1970's that are still in use... through to present day 'modern' systems. Despite the standards requiring extensive system testing.

Mark

We find problems with "Seamens" software and controllers frequently. You learn to look for consoftwarefusion and adjust your assessments.

But the controllers are harder to assess because there may be engineer design problems, air handler problems, electrofusion needing reboots.

Sometimes the controllers just fence sit either side of temp suppose to be and techs just ignore since it works "twice a day" like a broken clock.

Just sayin'

At a hydraulic cylinder mfr they wanted us to cheat the customer from heavy duty seal and only include it in the replacement parts kit. I asked if it was unethical and told it is not an unethical action...hmmmm

Why I hated going into engineering...for someone else