Another ZX Spectrum's near-death experience

19.06.2007 22:23

Approximately one year ago I got two Sinclair ZX Spectrums (48kB versions) from Cyberpipe's computer museum to repair. I was quickly able to put together an undamaged circuit board and a working keyboard into one working machine which was then on display in the museum for some time. After that I was left with enough parts for another Spectrum which I declared unrepairable because it seemed that every imaginable component was burned out.

However after tinkering with it every once in a while when I found some free time I got it back to a perfectly working condition. Looking back it was quite an amazing experience. Here's a list of all things that needed repair in the order I repaired them:

There was a big hole in the 5V regulator. Replacing that was the easy part.

Seriously burned out 7805

The ferrite core of the little transformer on the circuit board was also shattered, probably because of over heating. The wiring also appeared to be damaged. When I couldn't find a similar new core I ended up gluing back together the original core piece by piece with epoxy resin and then making new windings.

Continuing with the obvious stuff, there was also a resistor with a crack along it's body. Perhaps the transformer core shattered because of a mechanical not thermal shock after all?

Since the transformer of the DC-to-DC converter was obviously shorted at one time it was easy to guess that the switching transistor TR4 (ZTX 650) was also burned out. Getting a replacement however wasn't that easy because this NPN transistor has to withstand 2A of pulse current but also has to be in small a TO-92 or TO-18 case in order to comfortably fit on the circuit board and under the keyboard. I tried a bunch of transistors from a basic 2N2222 to a fancy 400V ST13001 I got from an old fluorescent lamp. They all eventually burned out (with ST13001 holding a record at several hours of operation). I only found a satisfactory replacement two weeks ago when I noticed that one of the local electronic component shops had an old stock of ZTX 651 transistors.

2N2222 on its way to destruction

Now it was finally time to fully power up the computer and see how extensive was the damage to the digital part. As expected the first time only four characteristic black and white vertical bars appeared on the screen. I suspected that the low 16 kB of RAM was to blame and it took me a couple of months to find a supply of chips that could be used to replace the original 4116 ICs (4164 ICs by the way and I only got four for quite an outrageous price). After experimenting a bit with RAM chips however I couldn't find specific chips that were bad.

Suspecting the CPU, I took the Spectrum to my Faculty's lab and removed all the big chips from the circuit board and put them into sockets. Replacing the CPU with a known-good chip however didn't help. After having some wacky ideas how to diagnose this problem I got some good advice from Winston on the World of Spectrum forum to suspect the ROM.

After confirming that the ROM chip was certainly acting strangely (I definitely wouldn't suspect it myself, thinking that a mask ROM was the most robust IC in the Spectrum) I've found an old 27C128 EPROM, programmed the original ROM contents in it and modified the circuit to accommodate an EPROM instead of ROM.

Those wires look dangerously close to the coil

Success! Spectrum now boots to the "Copyright by Sinclair" screen, however the keys don't work. As usual the keyboard membrane broke in several places. I first tried to repair it, but soon gave up (there's a funny story here about how the membrane worked properly each time I tested it on the table but was dead when I connected to the machine).

Luckily these guys offer cheap new (manufactured in 2007!) replacement membranes. Now it was just a matter of undoing all my previous keyboard-related modifications.

Keys now work, but there's no familiar clicking sound from the speaker. Interestingly one last component needed replacement - the protection diode on the audio output pin of the ULA chip. This also made me wonder if the damage to this board was made by connecting the power supply with a wrong polarity. This diode would form a direct short circuit in that case and I can't imagine any other fault which would damage it.

final result

So, now I have two Spectrums in a working condition. Maybe there's a game that supports multiplayer on two connected machines?

bunch of Spectrums

Posted by Tomaž | Categories: Digital


This is great! I`m trying to fix a zx spectrum myself.
Same problem, computer dead, tested voltage regulator giving out 6.5 volts, changed that, as you say the easy part.
My suspect is now the coil I have resistance readings of 0.001 ohms across one coil and 0.007ohms across the other, is this faulty?
TR4 I think has definately blown.
My question is how did you make new windings on the coil?
Thank you for an informative post.
All the best.

Posted by Ade Kavanagh

The coil resistance looks fine to me. If there is no short circuit between the two coils and the core is still in one piece then the transformer is probably OK. In my experience TR4 is by far the most common thing to go.

I removed the old wire from the core and took note of the number of turns, the direction and connections at the base. I then took the same gauge transformer wire and made identical new windings. Note that the base is symmetrical, so it's best to make a mark somewhere.

Posted by Tomaž

Thank you for your advice, You are right the coil is fine and TR4 had blown. I`ve replaced that and got the spectrum powered up again, but it does not boot to the sinclair research page, and suspect the 16k RAM ic`s as 3 of them are extremely hot. I just hope none of the main ic`s such as the ULA, CPU and ROM are faulty, because finding spares is extremely rare!
Thanks again for your help, and I`ll keep you posted.

Posted by Ade Kavanagh

I have 3 spectrum 48k that only give me squares on the screen
the CPU and ROMs are ok but I dont know how to test the memory chips
and the infamous ULA
I waiting for a working 48k to come so I can mesure the diferences
and maybe test the ULAs as they are removable

The squares on the screen are usualy a fault in the lower 16K RAM(IC6-13) chips(4116 2N).Power the spectrum with the keyboard removed and lightly touch each of these ic`s in turn to see if any are getting hot quicker than the rest, if so these are the ones short circuit, and causing the fault.(This is common if an interface device has been connected or removed with the spectrum still powered up! This is the problem with my spectrum,after diagnosing the fault with the power supply side of things.
The problem these days, Nuno, is finding replacements. The USA seems to have these ic`s in abundance, but getting hold of them is a different matter!

Posted by ade kavanagh

thank you for your advice Ade
there are 4116 ram chips for sale on ebay at this moment they cost 2 pounds each

Hi Nuno,
Many Thanks, managed to get hold of some.
I installed them in the spectrum, but it still is not initialising to the copyright sinclair page. On to the next culprit now, the ULA. Do you or anyone know if the piggyback ic spider can be used with the ULA5C112E-2 chip? That is there is two wires soldered to two pins of the ULA not connected in the DIL socket.

Posted by ade kavanagh

Hi! Great that you got the ram, they will come in handy for sure
I'm not very familiar with the ULA to talk about it but as I've read the 5c112 was that one
with the timing problem wasn't it? that's why there was the spider , so do some research
because I think that ULA definitely needs the spider, but dont trust me on this

I managed to fix one of the 3 48Ks I have, it didnt gave any signal and there wore only
lower voltages present in the +5 -5 and 12 measuring points
I found that the 7805 was shorted and replaced it and now the speey came back from the dead !!!!
now I'm going for the 2nd one that has all the right voltages but only shows squares, all ICs are soldered to the MB - just my luck
but the CPU seems to get really hot fast, so that will be the first one to go as soon as I master the art of unsoldering a 40 pin IC :)

WoW! I think I might be a little out of my depth here :) I have just got hold of a ZX Spectrum, in fact I will have 2 of them. I got them to see if I could get them going but it’s been many years since I’ve done any electronics (Built a ZX81 28 years ago). I do understand most of above, but I’m beginning to realise that I don’t stand much chance in getting them working.

All I’ve done so far is take the board out and connect it to a 9 volt PSU (I don’t have a original PSU, but the one I’m using is a large 1200mA 14.4VA PSU with an overload LED) and connected the video to my TV card in my computer and searched for a signal (is there a default channel I should be looking for).

I’ve noticed that the voltage drops a lot when the power is connected, the PSU puts out 9.5V when disconnected but when connected it drops to 3.4V, so I assume I have a component shorting somewhere.

I know you can’t tell a novice like me how to fix a Spectrum but if you can offer me any tips to spur me on it would be much appreciated.

I've just ordered a 7805 5v Voltage Regulator to replace the 78005AP fitted, because that's the only component that gets hot ;)

Does anyone have a good circuit diagram for the Spectrum?

I've got them both working enough to get the Sinclair logo on the screen, but the picture isn't clear, but I'm wondering if that is because the power supplie I'm using isn't smoothed.

Congratulations Steve on successfully bringing your Spectrum back to life!

About fuzzy picture: the stock Sinclair power brick doesn't provide regulated voltage either. It's possible some electrolytic capacitors in your Spectrum went bad - check if any are obviously leaking fluid or are warm to the touch.

There are a lot of really good Spectrum troubleshooting guides around the internet. A good place to start is the documentation section of

Posted by Tomaž

Thanks for the advice Tomaz, I now have a Spectrum + with corrupt video, it has loads of vertical messy stripes down the screen and the screen is dark, I will check the capacitors as you suggested.

The 2 Spectrums did work fine (apart from one needs a new keyboard membrane) on a proper Sinclair PSU, I've noticed the Sinclair PSUs seem to put out 12V even though they are supposed to be 9V.

It depends how you measure it. My PSU will give out around 12 V peak unloaded. But with a working Spectrum connected to it (smoothing capacitor + load current) it falls down to 9 V. AFAIK the black box contains nothing more than a transformer and a rectifier bridge.

Posted by Tomaž

That is nothing short of fantastic.
Such time, effort and skill put into that. Brilliant.

Hi Tomaž,
I've just got my old spectrum out of my loft, and tried to power it up. All I get on screen is the white border, but then flashing vertical lines. I suspect that this could well be like your's in that the ROM chip has died.
I am not very technically minded about how to programme a new ROM chip nor how to work out how to alter the motherboard to cope with a ROM chip.
Would you be able to make me a modified ROM chip and tell me what components I would need and where to fit them to modify the board? I can cope with a soldering iron, I just don't have the knowledge (or equipment) to programme an EPROM.
I can send you however much it costs for the chip and your time.


I was cleaning out some old electronics stuff and I found the following chips..

HN613128P - has sinclair logo on it
ULA5C112E-2 - Ferranti
VLSI VL2333-QC - Acorn Arm 1 CPU
TC531000CP-008 - Acorn RiscOs ROM 1
TC531000CP-009 - Acorn RiscOs ROM 2
TC531000CP-010 - Acorn RiscOs ROM 3
TC531000CP-011 - Acorn RiscOs ROM 4
TMS1000N13228 - Texas Instruments Microcontroller (Speak&Spell)
SCN8049HCBN40 - Signetics Microcontroller
MOS8371 - Amiga Agnus - Address Generator Chip 512K
MOS8372A - Amiga Fat Agnus - Address Generator Chip 1M

I have no idea of the state of the chips, but they have been kept in an ic 'sponge' so _could_ well be ok.
If anyone wants to pay the P&P, they can have it all..

Posted by KR

Any idea what the guage is for the wire on the coil as i have broken on, well one sire and counted 39 twists but not sure what the guage is.

Graham, I don't remember the exact gauge that was used on the original coil. Probably didn't even measure it. I just used the nearest gauge I had on stock.

Posted by Tomaž

Hi KR,

If you still here those chips then I would like to have them. Please drop me a mail.

Great write up, could you possibly identify which diode it was you replaced for the beeper to get back up & running ?


Posted by Sandy

Sandy, probably D13 or D9 according to the schematic. Don't remember exactly.

Posted by Tomaž

Hey Tomaz, thank you for that !!

Posted by Sandy

Hi there

Just picked up a speccy but only get a large black square on the screen when powered up. Am I right in thinking this is probably the lower 16k ram

Any suggestions greatly received


Posted by Mark

Hi Avian

Wow this is so interesting and exciting. This was my first computer and have very fond memories and was basis on my best friend relationship that still exists today. So I've bought 2 part working ZX spectrums and the plan is to get at least 1 working so I can give it to my best friend as a present.
Do you know if people fix for a fee as I dont have the skills?
If not I will try my best using all the advice you have kindly given.
Many thanks
PS just started following you on twitter (I'm @docdisco

Posted by Kev


Bought a non-working ZX machine a while ago. Many parts in the power suply area was gone (probably because someone put inverted voltages to it) but I managed to get it up and running again. Much thanks to this page and the information you have posted.


Posted by Mats

My spectrum+ has the black and white bars aswell, where can i get replacement ram?

Posted by Arlen Moulton

Hi, can you please watch this video for the flashing screen on my 48K?

I´ve already recaped the board, and the lower ram IC's seems to have the right voltage values.

Thanks in advance for suggestions.

Helder, sorry, it's impossible to say without probing the board with a oscilloscope or a logic analyzer. If power supply is working, then it could be a bad ROM, RAM, CPU or ULA.

Posted by Tomaž

Thanks for your reply.
I have another working board. If I swap the ULA and ROM to test, can they be damaged in this faulty computer?

Yes, if you are not sure that everything else is working correctly there is a possibility that new chips will get damaged as well.

Posted by Tomaž

Ok. I’ll start by buying some lower RAM ic’s and see how it goes.
Greetings from Portugal.

Posted by Helder Simões

Add a new comment

(No HTML tags allowed. Separate paragraphs with a blank line.)