preserving old Spectrums?

[animaal]

So I'm about to embark on refurbishing my old grey +2, swapping out the 30+ year-old capacitors. This got me thinking. Theoretically, what's the longer-term outlook for preservation of Spectrums (Spectra?)

Given that I intend to use similar caps to the ones I'm removing, I assume they'll need replacing again in 30 years time. I hope I'll be here to do it, and capable

How would the rest of a spectrum be expected to age? Is it ridiculous to imagine a 60-year-old ULA or Z80 still working? Is the PCB itself likely to deteriorate at some point if not stored in a temperature/moisture controlled museum?

I suppose I'm wondering how soon time itself is going to overcome our best efforts to preserve these great old machines.

[beanz]

Well there are ULA replacements and the Z80 is still made (I believe) as well as modern memory replacement solutions, so I don't see an availability problem in 30yrs...all you'll have to worry about in 30yrs is what time they'll be having the chocolate pudding social in the nursing home.

[1024MAK]

It’s impossible to answer your questions with definitive answers. But this may help...

Reliability is generally considered to follow the ‘bath tub’ curve. Keep in mind that this applies to a batch of production units, not to individual units. Where when new, there is a relatively large number of failures as substandard components and manufacturing defects affect operation. This then calms down and the failure rate curve levels off and flattens out. Over the following years, the failure rate stays low. Then as components wear out or start to reach the end of their design life, the failure rate starts to creep up. The failure rate will then continue to increase with the failure rate curve becoming steeper.

However, that only applies if no servicing work is done. Replacing components that are known to have a limited life (electrolytic capacitors for example) extends the flat part of the ‘bath tub’ curve.

Everything that is man made degrades eventually. But semiconductors, ceramic capacitors, low power film resistors that are operated well below their power rating and PCBs have a surprisingly long life.

Where I work, until three years ago (when it was replaced by a more modern system), a communication system operated continuously (apart from failures) for about 45 years. It was serviced every year though, as well as when it failed. Similar systems are still used elsewhere.

Currently Z80 CMOS microprocessors are still made, so that’s not likely to be a problem. The biggest problems may be the ULAs or the DRAM chips.

But I would not worry about any of that until much further into the future.

Mark

[Ralf]

Actually Spectrum are keeping quite well I believe.

As far as I know some "better" vintage machines like e.g. Amigas require full replacement of capacitors before you even try to turn it on. And floppy disks after all these years are mostly unusable.

Spectrum on the other hand is a simple but tough guy - cassettes usually work all right and most often you don't have to replace any capacitors at all.

How long will they last? I don't know Obviously Uncle Clive didn't plan them to be used for 50 years or so Your question may be also too general as it may depend on the model and the manufacturer, after all Spectrums were produced in many countries.

[Ast A. Moore]

The biggest problems may be the ULAs or the DRAM chips.

The later 48K boards have a custom glue-logic IC. It is also found in the original 128K machine and the +2. In addition, the +2 has a custom Sinclair Joystick (buffer?) IC. The PAL chips that were used in both the 128K and +2 are still available, but I’m not sure if the originals can be easily copied to program a blank one.

[1024MAK]

The biggest problems may be the ULAs or the DRAM chips.

The later 48K boards have a custom glue-logic IC. It is also found in the original 128K machine and the +2. In addition, the +2 has a custom Sinclair Joystick (buffer?) IC. The PAL chips that were used in both the 128K and +2 are still available, but I’m not sure if the originals can be easily copied to program a blank one.

• The (main) ULA (IC1) used in the 16K/48K/+ machines can be replaced with a module (using a CPLD).
• The (main) ULA used in the 128K toastrack and +2 grey machines was still available when I last went looking. But if needed, it too be replaced with a module (using a CPLD).
• The DRAM multiplexer custom chip (ZX8401 / PCF1306P / Amstrad 40058) - IC27 (48K/+ issue 5 and issue 6A boards, 128K toastrack machines), IC10 (+2 grey machines) can be replaced with standard 74xxx series logic chips on an adaptor board.
• Sinclair Joystick chip (MCE HT62001 / MCE MT62001 / Amstrad 40057) IC1 (+2 grey machines), IC1 (Sinclair ZX Interface 2) has no current replacement that I know of, but is not thought to be particularly complex, so should be able to be replaced with either a CPLD or standard 74xxx series logic chips on an adaptor board. Note that the MCE HT62001 / MCE MT62001 and the Amstrad 40057 look very similar in pin function, so I am assuming they are indeed functional equivalents, but have not done any testing to confirm this.
• For both the the PAL used on the 128K toastrack and on the +2 (grey), all the programming details are available for GAL chips that do the same job.

However some of the replacement items are only available currently in limited quantities. So in years to come, there may have to be replacements for the replacements!

Mark