Thursday, November 27, 2008

ZX Spectrum Composite Video Mod

The original 48K spectrum was pared down to the very bone to keep the price low as possible, as a result many features that became standard on later 8bit machines were severely lacking, no joystick ports, no printer ports, no RS232 ports and no RBG video port. Many of these deficiencies were addressed by a multitude of 3rd party add-on interfaces over the subsequent years, yet there is still one problem facing folk who chose to use the original rubber keyed hardware today that was never properly addressed, its the fact that the only video output option is the RF (the aerial socket).

This is pain in the rear for many reasons, firstly the video quality is not as good as it could be, secondly TVs only have 1 aerial input so you have to fiddle round the back of the TV and disconnect your main aerial feed if you use it. Thirdly you have the problem of PAL type, the majority of spectrums were made with PAL I (UK PAL) modulators, if you happen to live in a non PAL I country and don’t have a TV smart enough to auto detect you are stuck with black and white.

Wouldn’t it be great if you could hook up the Speccy to one of the multitude of AV sockets modern TVs have? The Speccy didn’t have composite out because TVs of that era didn’t have composite inputs. Fast forward 5-10 years and most TVs did, but the Speccy is stuck in the past.

There is a way round this - and it is so simple it’s laughable; you need about 10 minutes, a soldering iron, and 4cm of thin wire. This modification also involves no visible changes to the case, no drilling, no cutting, and in fact the RF socket will still be the output method for the video - except it will be giving composite video out instead of RF. Happily the RF output socket is an RCA socket so you just hook up your yellow AV cable to the Speccy’s RF socket and you are away.

The trick is that the Speccy actually feeds composite video into the RF modulator in the 1st place, it modulates this to broadcast RF signals, TV is tuned to the output signal and dutifully converts it back in to composite for display. Naturally all this processing degrades the image and nowadays is totally unnecessary.

Heres how to do it - some of these photos are quite big, its hard to see what's going on when I shrank them, so I unshrank them.

Unscrew the 5 screws underneath that hold the case together. Lift the upper portion of the case off the lower and bring it towards you until you see the 2 ribbon cables attaching the keyboard to the main board. You don’t have to disconnect these but I would, you are more likely to strain them by leaving it connected while you work on the system than by removing and reconnecting them afterwards. By the way - this is the only risky part of the procedure, these keyboard membranes were never the most robust component and aged ones are probably more fragile - don’t strain or bend them ribbons.

There is one screw in the centre of the board that holds it to the lower case, unscrew this and remove the board.

In the rear left hand corner there is a metal box, this is the RF modulator, the lid is only clipped on - pop it off.

Now you can see what’s behind the RF socket, you will see a resistor standing on end in a small plastic tube, the upper leg is soldered to the central pin of the socket.

you need to unsolder that and bend the it away from the pin (Crap camera angle coming up... - you can just about see it in this shot. Tis easy to see what I mean if you have one in front of you tho).

You will also see 2 wires going into the RF box on the left hand side - one of these wires goes through a plastic insulator - that’s the video feed, clumbsily labelled V on the image.

The other one goes through a silvery nub is the +5V power feed for the modulator - we need to disconnect both of those wires. You could leave the modulator powered up but there is little point, also all the RF frequency circuitry could in theory interfere with the signal you are passing out the back in close proximity, so you might as well leave it powered off.

The 5V wire is just soldered to the upper of the board, unsolder it and bend it up out of the way as shown above. Just the video wire to disconnect, this is soldered through the board to the composite video signal line.

The video feed is soldered from beneath.

...unsolder this wire and bend it out of the way too. Make sure these 2 wires don’t touch anything or each other - you could put a piece of heat shrink insulation on both if you are enthusiastic enough.

Now Sinclair very handily chose a modulator with 2 holes through the plastic insulator on the side of the modulator, so you don’t even have to work out a way to get the wire to the socket. Pass your 4cm of thin wire through the second hole and solder the end to the central pin on the output socket. Solder the other end to where the RF box got its video input originally.

You can see the new yellow wire going from the old feed to through the second hole, also the old 2 connections detached and bent out of the way.

And that’s it - the ground on the Speccy is all common so the ground on the RF box and sleeve of the video socket is already connected up.

Double check you have wired everything as per the photos and put the Speccy back together. Connect "Der Computor" up to the telly.

Et Voila - a Speccy connected to my TV on input AV2.


By the way – I didn’t invent this, it’s not even a recent discovery – it was described in some detail in a 1986 Crash magazine. There are also little FAQ text files from the bulletin boards of the early 90s floating around the web – but they all seem to make it sound harder than it is, and none had photos of the process which shows how simple it really is. Plus they all seem to advocate leaving the RF connected up and running a flying cable out the back of the unit – and or drilling a big hole in the case to have the feed dangling out the back – both are no-nos in my book, if I want an old system it has to be physically intact, cutting big holes is not an option, especially as it can be so damn neat as the above shows. It’s another 10 min job to reconnect the RF later if you decide you really do need it – but I can’t see why anyone would ever go back again.

Saturday, November 22, 2008

Repairing Panther's Oric

Eventually I will be migrating all my old articles from my backups to the new site, but I thought I would start with the most recent/active tales of my fettling. Another member on the old site had bought an untested Oric Atmos and found that it was rather dead...

Here I sit 12,000 miles away from Panther, and yet on my desk is his Oric. Tis a funny old world, what's even more strange - is that it is working away happily beside me, probably the first time it has worked in years, decades possibly, it seems happy.

I have had it here for about 2 months, nature conspired against me fixing it a number of times, firstly my trusty Commodore 1084S decided it didn't want to be a monitor anymore, it took me about a month to work out how to convince it otherwise - monitors aint my thang. Then yesterday when I was ready to tackle it once more my bench power supply decided it wanted a change of career too. That turned out to be a manufacturing fault, am surprised it had lasted as long as it had.

Anyway - Panthers Oric arrived on these sunny shores with a nice simple fault - it did this...

...lots of pretty coloured crap on the screen, somewhat inconsistent pattern tho.

Always a good candidate for a fix, first thoughts tend to be RAM RAM RAM. Oh actually the first thought is always power supply, but as that's external you really only have to check the voltage regulator and you can move one. Tangerine computers were renowned for buying really really cheap and nasty RAM chips, they were ones that had failed the manufacturers QC, but worked just well enough to go into an Oric that didnt drive them too hard. The failure rate was alarming in the early days and its quite uncommon to see an Oric with all its original RAM, Panthers had already had 2 chips replaced. So after a quick buzz round the board with a logic probe I removed the remaining RAM, soldered in a batch from my stock and prepared to bask in the glory of a quick fix.

The speaker is a pain in the rear as its only mounted on two tags and it flops about a bit due to its weight, so that came out for a while too to make it easier to work with.

I was disappointed - even with the new RAM it was ill, similar pattered effect as before, and as I had soldered in the RAM I couldn't even swap it around a bit as now seemed to be required, the Oric is a very very fussy machine and juggling the RAM chips on my old Atmos chased away some boot up problems.

So I had to desolder all the RAM again. :(

I am lucky to have another Oric here that I fixed ages ago, when that has all its RAM removed it puts up a messy black and white display on the screen that keeps resetting, presumably Panthers Oric should do the same without RAM, thus allowing comparison between the boards with RAM out of the equation. So I unsoldered the RAM from Panthers board to see what it did sans RAM - nothing similar to mine was the answer. Lots of crap on the screen, colourful tho.

I quickly dumped the eprom and compared the binary file with my Orics eprom dump, they were identical, no clues there.

At this point I was stuck, which way to go next? Orics are simple beasts and as such they either tend to work fully, or not at all, there is no half way house to give you any clues. Based on Speccy (similar architecture) faults I knew a duff ULA would give weird results, and that they are not uncommon failures, and indeed the lines on Panthers ULA were doing different things to my Oric, then again whatever was causing the fault could be driving that, and it may not be ULA at fault. Anyway - my ULA was socketed from birth, Panthers was soldered directly to the board, so I desoldered it and plopped it into my board, and was greeted by a fully working Oric - arse - so not the ULA then.

Having replaced the RAM the crap on the screen was slightly more consistent and this was reassuring, we will never know if the original RAM was good or not as I had to cut it loose to avoid damaging the board, but a consistent fault is better than a random one, RAM faults tend to look random so perhaps there was an issue there originally.

A static screen of rubbish now looked like a CPU fault - my repairs of Capcom A arcade boards has yeilded a small pile of dead Z80 CPUs and I tend to slap them into my Galaxians board to see what they do, the results are heavily dependant on what is in the associated EPROMs so the rubbish is not going to be directly comparable, but there was a fishy similarity.

So fired up the desolder station and whipped out the 6502 CPU chip.

While the board was CPUless I decided to see what the board did when powered up, surprise surprise it did a damn good impression of the main fault, it seems even with the CPU chip on board the machine is missing its CPU.

6502s were old school by the mid 1980s and none of my scrap boards date from then, any boards I do have that are older are all Z80 based. My Orics CPU was soldered in and I was not keen to subject it to a desoldering. So I hit google hoping to find a supplier or ebay seller, turns out I had one within ten feet of me, in my Commodore 64 1541 floppy drive. It also had the PIA chip that goes with the 6502 which was my other prime suspect. Both were luckily in sockets so I whipped the CPU out, fitted a socket on the Oric board and installed the CPU.

Flipped the power switch and got this....

....signs of life, the system was booting but freezing almost instantly. This was going to be either a faulty PIA chip or more RAM problems. I was planning to replace each chip 1 by 1 along the RAM bank, but infact on replacing the 1st RAM chip the system booted and beeped happily.

One working Oric, and an addition to Womble's knackered CPU pile.

Need to find a 6502 chip from my contact out West and this fella can wing its way back to Panther.

First Steps

10 PRINT "hello world"
20 GOTO 10

BASICally thats what this post is as I am starting my own page. I used to hang around an old site that long ago was called and was a hangout place for folk who buy the dusty old computer crap on ebay, and spend their evenings trying to get it working again, in a nutshell people exactly like me :)

That site ended its days with the smoking ruin of the hard drive at the hosting facility and the sudden realisation that although the admin had been paying for a backed up service no sod had actually been changing the tapes, or noticing that the tape in the drive was wrecked.

That site then became a sub site of, which was hacked by spammers in the last 24 hours which was the final straw.

So my articles needed a new place to live so here she is, Wombles Retro Repair Shack blog, the original format was one of a forum, which wasnt really the ideal format for my ramblings, a blog may well be.