Trying to restart a 33 years old ZX81 [SOLVED]

[dinosaur]

Some Info on the different versions here.

It looks like I got the "best" ULA then... Let's hope it's fully functional...

[IanB]

I might have missed this in the thread, since you're using a modern TV (which may not be happy with the 81's poor quality output), have you put your oscilloscope on the actual signal to the TV to see what it's sending out? Also, switching to FAST mode to see if the screen blinks? It might just be that the modern TV isn't seeing the black levels the 81 is generating?

[dinosaur]

I might have missed this in the thread, since you're using a modern TV (which may not be happy with the 81's poor quality output), have you put your oscilloscope on the actual signal to the TV to see what it's sending out? Also, switching to FAST mode to see if the screen blinks? It might just be that the modern TV isn't seeing the black levels the 81 is generating?

Please, re-read the thread. Yes, I did all that already.

The problem is not with the TV signal, but with a Z80A which doesn't seem to start any more, even with the ROM and the RAM disabled (via the forcing of the CS lines to +5V).

I decided to try and unsolder the Z80A to test it apart. It'll be quite a challenge to extract it without destroying it, but there's little else I can do...

I'll let you folks know about my progresses, but I'll be absent next week, so it will take some time before I can come back with interesting news...

[1024MAK]

Take your time, desoldering is not something to be rushed

Mark

[dinosaur]

Take your time, desoldering is not something to be rushed

It took one full hour, but I managed to do it... I first used a standard 25W solder iron and desoldering air pump to remove as much solder as possible from the underside of the PCB, making sure each of the Z80A pins looked "free" from its PCB pad. Of course, at this point, I still had to deal with the upper side pads which still got soldered there, on quite a few pins (I'm still cursing Sinclair for that cheap PCB design... At least, and unlike the Sinclair QL, the ZX-81 PCB is quite thick and indeed pretty sturdy).
At this point, I used a hot air blower solder iron (at 350°C, which is very hot, but Sinclair's solder doesn't melt at lower temperatures... 250°C is usually enough for my self-made projects), still on the underside, as well as a small and flat screw driver inserted between the PCB (components side) and the IC and used as a lever, to (gently) part the CPU from the board as I was heating up the pins underside... Of course, I paused several times, to avoid overheating either the CPU (which still easily heated above 60°C, i.e. got 'burning' hot to the touch for a few seconds each time) or the board.

To my relief, the CPU finally came out, apparently (there's still a doubt about a potential overheating) unharmed, and so was the PCB (but for a single pad, partly destroyed, but thankfully not where it connects to its PCB track). The 'tropicalizing' varnish/lacquer is also unharmed (I feared to carbonize it at this temperature, but no, it's of good quality, obviously).

I could then test the Z80A... I used a test bench I made up one year ago or so, which, ironically, is using a PIC easily 100 times faster than the Z80A. That test bench got a power supply (5V and 3.3V), a bread board, a LCD display and 2 push buttons, as well as a DB-25 port to the PIC (a Micromite, for the ease of programming, in BASIC) and a serial port (connected to a computer, for programming the PIC via a simple treminal program). It's a nifty little device, which makes testing ICs a real pleasure and quite an easy task...
The test consisted in wiring the Z80A GND to the bench GND, pulling down Dn lines (via 1 kOhms resistors) to GND (so that the Z80A would execute NOPs forever when clocked), wiring Vdd, WAIT, INT, NMI, BUSRQ to the +5V, wiring the Z80A RESET pin to one of the PIC's output pin, the Z80A CLK pin to one of the PWM pins of the PIC, and three An lines (A1 to A3, at first, but I tested all of them with the scope as well) to three PIC inputs (these inputs being configured with their internal pulldown resistor). I then applied a PWM (50% duty cycle) frequency of 1KHz (and tested further from 100Hz to 100kHz) to the CPU clock, pulled down the RESET pin for 10ms to GND via the connected PIC pin before pulling it back up, and watched the An lines (both from the PIC readings and from my oscilloscope).

The result confirms what I finally (with your help, folks) deduced: the Z80A is brain-dead... This *really* comes as a BIG surprise to me, but this is just a fact: after 30 years of storage (in rather good conditions: no moisture, no extreme temperatures), a Z80A can become non-functional... Be it because of (in likeliness order) electrostatic discharge (when, how, inside a closed box ???... A mystery !), electromigration (doubtful at that engraving thickness and while not power supplied), internal solders breakage from the pins to the silicon chip (just an hypothesis of mine, but why not ?), cumulative cosmic rays damages (maybe I'll die from cancer because of them, then...), or even Gremlins (now, *that* might be an explanation !), the fact is that it's dead, nuked, kaput !

Next step: after I come back next week, I'll order a new Z80 (only found a Z84, i.e. a 10MHz Z80 at Farnell/Element 14... It should be fully compatible...), a proper socket (<rant>not the cheap type Sinclair used for the ULA</rant>), and will put that on the board... Let's cross fingers !

[IanB]

Maybe your house sits on top of some rather radioactive granite bedrock

I'm quite surprised as well that it would just expire like that. There you go, it pays to buy the extended warranty after all

[1024MAK]

Some interesting theories...
But maybe there was a small bit of dirt on the silicon die that made it into the chip, and over the years this has caused corrosion...
Or there was a defect with that Z80 chip and although it worked while new, over time, the defect has continued to degrade and now after 30 odd years it is now causing a fatal failure.

Mark

[IanB]

This sort of reminded me of the lighting control board we had in the theatre I worked at in the 1980s. It was a Strand Compact 120, very old tech by today's standards- it even had magnetic core memory(!)- and in general the service engineers, faced with an inexplicable glitch, would put something like "a spider might have crawled onto one of the cards" on the report

[1024MAK]

Talking about insets, in my job, every once in a while, a insect will get caught in between electrical contacts (used to detect the position of a mechanical machine) and cause a failure costing rather large amounts of time and money.

Mark

[Moggy]

Talking about insets, in my job, every once in a while, a insect will get caught in between electrical contacts (used to detect the position of a mechanical machine) and cause a failure costing rather large amounts of time and money.

Mark

I read in one of today's newspapers that we have apparently been invaded by Argentinian "crazy" ants which love nothing more than shorting out electrical contacts!

Eyes peeled then Mark.