Sinclair ZX Spectrum repair

George rang... 

"Got a ZX Spectrum here that doesn't work. Care to have a quick look?"

Yeah... why not?

It duly arrives in the post. 

I have to be honest, and say I haven't so much as seen one since the 90's... and only fixed very few back then. 

It's the earlier sort, with the rubber keyboard. 

The power supply is tested (Centre negative!) and it's output is around 14V DC off-load, so that's OK. The supply is put to one side, and the unit is powered up from the workshop supply. Something is loading the supply heavily.

Five screws in the bottom of the case gives us access to the PCB. 

Carefully remove the two keyboard ribbons, and remove the single screw in the middle of the board, to remove the PCB. 

This machine is fitted with a version 4 PCB, and the chips are all dated around 1984. 

Attention is focussed on the power supply. 

It consists of an LM340TS (7805) linear regulator, and a small self-oscillating supply, which provides + & - 12V, -5V (not at 20 amps!!) and a 12V "AC" rail

A few cursory checks, and it looks like our 7805 has failed, thankfully short circuit to ground, rather than short circuit from input to output... having ~14V on something that's expecting 5V can really ruin your day.

The heatsink and offending part are removed from the board (carefully, the double sided print is fragile) and a Traco TSR1-2450 switched mode IC fitted in it's place. This is pin-for-pin compatible with the 7805 it replaces, but barely gets warm to the touch, and requires no heatsink. 

The rest of the board is in really nice condition. The modulator's a bit rusty, but it's getting on! 

Powering up the unit, and we have life.. but just a black screen with a white boarder. No copyright message, no keyboard clicks... Removal of the socketed IC's, and a quick clean up with some servisol contact cleaner .. and we can try again.

Bingo, the familiar (C) 1982 Sinclair Research Ltd message appears on my (rather small) workshop test monitor!

The obligatory test programme is typed in ... 

... and it runs, but all is not quite right ... 

Keys 1 to 5 don't work. :( It'll need a new keyboard membrane. 

Never mind, we can at least play Manic Miner... 

So a .wav is obtained of Manic Miner, and loaded onto my aging MP3 player, and connected to the Spectrum's EAR socket... 

LOAD ""

The boarder changes colour as it's supposed to, but no amount of amplification persuades it to pick up the pilot tone, nor can I hear the tone through the speaker. My heart sinks at the thought of a faulty ULA ... 

The tape and sound interface all share a common pin on the ULA (28) ... audio comes in to our EAR socket, is loaded by R37, fed via R36 and C32, with a bit of filtration by C35. D13 Clamps the signal, and it's passed to pin 28 of the ULA. Also hanging on pin 28, is an output to the MIC socket for saving programs, and, via D9 & TR7 out to the speaker. Possible candidates for failure are R36 & C32 open circuit (unlikely), C35 & D13 short circuit. D9/TR7 and possibly even the speaker faulty, or, of course the EAR socket itself. The ULA could be faulty, and loading the whole thing down. So, we have key clicks, which shows that the speaker and TR7/D9 are working, and that the ULA is at least outputting something (whether it can input, is undecided at present.) and that it isn't being loaded down by a short circuit D13 or C35. We also can't hear the signal through the speaker, so that leaves us C32 or R36 open, a duff EAR socket, or a ULA that goes into a low-impedance situation when it's set to receive audio. What's the betting it's the socket ... after-all, that's the bit that had all the abuse in the 80's! 

After tracing the audio through the circuit, it's not the EAR socket! It is in fact, and open circuit C32! 

As an aside, it appear my 4B board differs from this circuit diagram, as TR7 is fed from the 12V rail, rather than 5V as shown. 

LOAD ""  


Bingo! 

Now onto the keyboard. Yesterday I ordered a replacement membrane from the fabulous  https://www.retroleum.co.uk/  . It arrived this morning. What fantastic service! Phil who owns the store also manufactures a replacement ULA, called a Nebula for sensible money, although stocks are short at the time of writing.

The kit comes well packed, with detailed instructions, and a plastic razor blade tool.

Changing the membrane is simplicity itself... 

Flip over the keyboard, and carefully bend the brass tags straight. Later computers apparently have the front plate stuck down with some double sided tape, which the plastic razor blade will assist in removing.

The plate can then be lifted off. Some wiggling and tweaking of the brass tags was required...

The old membrane can now be lifted out, and the new one slots right in. I have to say I'm now a bit worried, as the numbers 1 to 5 have a conductor in common... what happens if I've missed a broken connector, and the membrane's not at fault? I suppose I've just spent a few pounds for nothing...

Thankfully, my diagnosis of a faulty membrane was spot on..

Another saved from landfill!