Monday, June 03, 2019

I Forgot to Write This Post - Another A500+

I was working on this A500+ motherboard before Christmas (yes, over six months ago - oops). I swapped a working A500+ board for this board, plus a small contribution, with a chap who had bought an A500+ without realising the potential for disaster.


So here it is:


"What's in the box?!"

Oh. My. Goodness.

Let's have a slightly closer look at the board and the extent of that damage. If you are of a nervous disposition, look away now.

Argh! The Horror!

I can safely say that this is by far the worst A500+ motherboard I've had on my bench. The battery acid has spread from the end of the battery and engulfed U12, U10, Agnus and who knows what else. It really is pretty bad. Have I met my match?

First things first. Gary. What does the socket look like? Bizarrely, Gary was untouched. This is really unusual and could indicate that this A500+ was stored standing on its side.

OK. So, on to Agnus. Oh. Dear.

Looks bad.

Yep. It's bad.

So, a new socket required. The actual Agnus chip was OK and just needed a bit of a clean. The chip also had a chunk of the printing on it's top removed, presumably by the action of the battery acid...

Printing wiped from Agnus
(Taken before I removed the chip)

But, oh boy, that socket really is destroyed. I've not seen one with ALL of the pins corroded. Time to get the socket dispersion apparatus (a pair of cutters) on to it.

Socket nearly gone - damage marked with Sharpie

I removed the ROM chip and CPU to keep them safe. The corrosion had also hit the ROM socket. I took some time to clean this up as I didn't have any other sockets. The damage to the contacts was largely superficial so I think I got away with it other than a couple of pins. More on those later.

Socket begone.

With Agnus' socket removed I cleaned up the rest of the board. A dash of vinegar and then a gentle scrape with a screwdriver to remove the damaged solder mask, followed by a rub down with some IPA to get rid of anything left behind.

Clean and..err...bright.

I had one Agnus socket left. I bought a pack of six originally. So that means that, over the last few years I've saved at least six A500s from a fate worse than recycling. Crikey.

Anyway. 84 pins later, Agnus has a new home.

Action shot! Agnus being replaced.

Oooh! Fancy lighting...

Now, that ROM socket. Two of the contacts actually broke off as I was cleaning it. The others were fine so I don't know what made these two susceptible to the battery acid but, cest la vie. To repair these I de-soldered just the single affected pins and pulled them out of the socket. Then I cleaned the holes out and, using a couple of new pins that had been pulled from a good socket I just pushed them back down into the socket and re-soldered the pin. It was a surprisingly simple and relatively easy fix. Unfortunately I didn't take any pictures with my camera but I hope my description gives a hint as to what I was doing. Note that this would probably only work on dual wipe sockets where the entire spring contact assembly can be removed relatively easily. Moving on.

First smoke test. Does it work?

No.

Green Screen. Not a surprise.

So, the next thing. Replace U10 and U12. These provide a data path for memory access and, as mentioned in many previous posts, this would be a very likely source of the green screen. The corrosion on these two chips was very bad, to the point where I didn't even bother trying to de-solder them, I just cut them off the board and cleaned up the areas affect. I had a couple of spares from a previous repair.

It did not look good...

Disappearing pads and traces...

Out came the multimeter on continuity mode. Check. Beep. Check. Beep. Check. No beep. Note it down. Beep. Check. Beep. Check. No beep. Note it down. And so on for many, many traces. I ended up with a list of about nine missing traces.

To get the chips back on the board I installed sockets. The problem with this was that, as should be obvious from the picture above, there were several pads missing. To solve this I used fine single core wire and hooked a small piece down through each affected hole. Then I soldered it to the existing trace on one side of the board before gently pulling it tight and soldering it to the trace on the other side. Then, when a socket was inserted the leg had something to solder to. It worked quite well but was very, very fiddly..

Two repairs on this side.

Three repairs on this side.

And a close up..

So now I was confident that the replacement chips in their sockets would work but there was still a long list of broken traces to deal with. This time I used coated copper wire which I think came from an old electronics kit. It's pretty fine but stiff enough to shape if needed. Here's the first attempt.


Spaghetti repair.
I had to label the underside of the board as my simple brain kept losing track of which chip was which and which way around the pins are. It's harder than it looks when it's on the other side! And this was only about half way through....

Are we there yet?

With all those wires attached, I tried a second smoke test.

It's ALIVE!
That is a result. :)

I wasn't happy with the wires underneath though so I carefully re-worked them, taking extra care not to dislodge any of the pad/socket repairs. This was the final result.

Much better.
And now a few tests, just to prove it was reliable.

Memory test

Picture test

Another A500+ lives to see another day.

Fin.













Friday, April 05, 2019

Spectrum+ Refurbishment & Repair - Part 2

Story so far. Cheapo Spectrum+, replaced the Z80, replaced an upper RAM chip. It works. Keyboard works a bit but there is a large portion in the middle that doesn't work.

So, I have a brand new membrane, the one that was recommended on several websites. I should also point out that several websites also noted the potential difficulty of getting Spectrum+ keyboards working correctly.

Pah. How hard can it be?

On the standard ZX Spectrum keyboard, there are several key combinations that allow different modes to enter different commands. For example, 'Extended Mode' requires the pressing of SYMBOL SHIFT and CAPS SHIFT. This turns the cursor into a flashing 'E' indicating 'Extended Mode'. Simple.

For the 'upgrade' to the Spectrum+ the keyboard was improved by adding specific keys to cover these modes so instead of having to remember a key combination, all I need to do is now find the right key. In the example above, there is now an 'Extended Mode' key. Nice.

Extra special keys. Yum.
To support this the keyboard membrane in the Plus is different to the standard rubber keyed models. It has an additional layer to cover those extra keys. Remember that the function is still the same but the membrane now has to mimic two keys being pressed at the same time. The membranes are bigger and, because of the extra layer, the tails need to be clamped together to make sure that the three layers are making good contact.

Dismantling the Plus keyboard is also much more involved than the standard model. There are two clamps which need to be removed. This frees up the tails. Then there are ten screws that fix a metal (or could be plastic) back-plate. Underneath this could be a cardboard spacer (if the back-plate is metal) followed by the membrane and then a 'bubble' mat which acts as the return spring for each key.
In my case, I have a metal plate and a very, very saggy bubble mat. More on that later.

So. Many. Parts.
The old membrane tails didn't look in too bad condition but, with a bunch of keys failing to work, it seemed obvious a new membrane was needed. There are several reproduction membranes on the market but the one that kept being recommended was those supplied by RWAP, certainly a well respected supplier in the Retro community. Having read that the most excellent people at Mutant Caterpillar assisted with tweaks to its design, and also seeing that Retroleum recommended them too (notwithstanding he won't stock any '+' membranes due to issues with making Spectrum+ keyboards work properly), I ordered one of this type.

On arrival it's obvious that the membrane is made from very different materials to the original. The plastic is smooth and transparent and the overall membrane is slightly thinner than the Sinclair original.

A sidenote. If you are lucky enough to end up with an Issue 4S then you have a Spectrum+ manufactured by Samsung. Yes, Samsung. Unsurprisingly, the keyboard membranes on this variant of the Speccy do not suffer the same degradation problems that the Sinclair models do and most are reported to be still going strong.

Anyway, back to my Issue 4A. I have watched several videos on replacing the Spectrum+ membrane including the excellent Mark Fixes Stuff. I also read up as much as I could on Google given the apparent difficulties that people encounter when carrying this out.

Fitting the membrane is a reversal of dismantling the keyboard. Firstly, the bubble mat sits behind the keys. Then, the membrane is laid on top of the bubble mat, ensuring that the location posts protrude through the locating holes and that the screw posts line up correctly. Then, the cardboard is placed on the back to help keep the membrane flat. This also has a couple of location holes. Then, finally, the metal back plate is attached with many, many screws securing all of the layers to the keyboard.

When this is done, it's important to remember that the tails need to be re-fitted in the claps. This is make sure that all three layers of the membrane have their contacts pressed together properly. 

Finally, the tails are inserted into the two connectors on the main board. And it just works! Right? 

Wrong.

First major problem. The mid portion of my keyboard was sunken quite badly. I had a to order a new bubble mat which, fortunately, was in a much better condition than the one I had and was a very reasonable price from our friends at Retroleum. 

Next problem. None of the keys actually seemed to work properly at all. Certain keys on the left side of the keyboard would produce numbers but that was about it. After re-installing the clamps and tightening them a little more the situation improved.

A lot of the keys now gave the correct response. Space, Enter and the two variants of Shift worked with most keys. But there was another big problem. The extra keys wouldn't work. Neither would certain key combinations, most notably the symbol shift + 'p' combo to give the " symbol. Any Speccy afficiandos will know that it's a bit tricky to load anything without the " symbol.

Also, and more annoying, most of the middle row of letters would trigger an additional character. For example, pressing 'g' would give a 'g' (or appropriate keyword) on the downstroke of the key, but on releasing the key, an unexpected 'f' would appear. In fact, all of the affected keys generated an extra 'f' except for 'f' which generated an extra 'g'.

This extra character problem wasn't limited to the middle row either. Other random keys suffered but the middle row was by far the worst.

By now I was getting frustrated and contacted the seller. Their response was almost instant and very helpful and detailed. I tried a lot of the suggestions but some I knew wouldn't make any difference as I was having the problems even before the keyboard assembly was installed back onto the case. 

I ranted a bit on the Spectrum FB group too where the seller also happens to hang out (not surprising really!). Even more helpful suggestions came my way to try and I tried them all. Nothing worked.

Eventually, I tried stripping the membrane out of the keyboard assembly and plugging it into the main board but leaving it flat in front of the machine. By pressing my thumb and finger against the tails on my desk to replicate the clamps I could then see if I had a (rare) faulty membrane.

And...stretch, 2, 3, 4 and.. reach 2, 3, 4..
The results initially seemed to indicate the membrane was fine. I could press the 'keys' on the middle row and no extra characters appeared. I could hit the extra keys and they would 'just work'. Except one. The " key. This would still not work, nor would the symbol shift + 'p' combo. 

One of the FB group referred me to the Spectrum service manual. An addendum to manual detailed instructions on how to upgrade a 48K rubber keyed spectrum to a Spectrum+. One of the steps listed was to check that 'STOP' worked on the keyboard i.e. symbol shift + 'a'. On my membrane, this also didn't work even when presented naked on my desk. The suggestion to improve this was to add a 22K ohm resistor across R68 . This has the effect of reducing the resistance from 10K ohm to 6.8K ohm. This I duly did.

Annnnd, it made no difference to the 'STOP' or " symbol issues. 

At this point I decide to re-install the old membrane, just to see if I could make the 'STOP' work at least. The 'a' key was not in the group of faulty keys.

Having read so much and done so much fiddling with the new membrane I was a dab hand at re-installing the old one. Imagine my surprise when, after completing the re-installation I discovered that, not only did 'STOP' work but so did the " symbol and every other key too, even those in the previously faulty area. (Well, I say 'all' but one key didn't work - the 'Inv Video' key but who uses that? And more importantly, the 48K rubber keyed combo for 'Inv Video' still worked if I needed it.)

From this I can only assume that a previous owner had been fiddling, buggered up the clamps when putting everything back together and assumed that the membrane was dead. It's not. At least not yet.

Big clamp....

Little clamp...
By now, I'd pushed and pulled the new membrane so much that I'd worn away the metal conductors at the end of the tails. I was prepared to just chalk it up to experience and put the membrane away in case I needed it at some point in the future and I'd deal with it then.

This is getting a bit wearing...

Tails, tails, tails, you can swing them to and fro...

But then the awesome Mutant Caterpillar offered to test the membrane for me, on behalf of RWAP.
They found that the membrane was fine. They trimmed the ends of the tails very slightly to get rid of the worn ends and it just worked. I wasn't surprised by this given my finger stretching exercise, but I was very puzzled as to why it didn't work in my Spectrum+.

So I received the membrane back and decided it give it another try. And.....


It.

Just.

Worked.


I have no idea what I did differently, but the thing just worked. The only problem is the 'STOP' keyword doesn't work (which is the SymbolShift + 'a' key combo) but this is a known issue that I can probably solve with tuning the resistor across R68 - as per the ZX Spectrum service manual.

Perhaps I wore away one of the contacts on the tails on the very first install resulting in the erratic behaviour I saw all the way through, or maybe I just didn't hit the sweet spot before in the Molex connectors on the main board. Or maybe the screws on the backplate were uneven or too loose or too tight. Who knows. At this point the membrane now works. I'd highly recommend them despite the problems I had which were obviously not down to the actual membrane. You can get them here.

So, after all that the only things I have left to do are add a proper composite mod and, if possible, try and sort the STOP keyword. :)

Sunday, March 24, 2019

Spectrum+ Refurbishment & Repair - Part 1

I don't normally have a lot of luck when it comes to picking things up on the sale groups in Facebook. Thanks to the FB algorithm, I normally see posts about bargain stuff four or five hours after it's first posted meaning that the item has already gone.

This time, though, I saw something almost the second it was posted. A Spectrum+ for a very, very reasonable price. I bought it immediately. It was sold as needing a new keyboard membrane (no surprise there) but condition otherwise unknown - to be honest I didn't ask at that price!!

After a brief wait, it finally arrived. A quick inspection showed that it needed a bit of a clean but otherwise looks in good condition. It was the unit only but, as luck would have it, my rubber keyed buddy's power supply will provide the power I need.

Speccy Plus! Yeah!

And then I made a mistake.

There is an excellent video by Joules per Coulomb on YouTube (see HERE), which runs through the basic tests to perform prior to plugging in a ZX Spectrum, the idea being that any major faults can be intercepted and corrected, thereby avoiding unnecessary additional faults being caused. For example, the 4116 lower RAM chips require three voltages to operate (5v, -5v and 12v). If any are missing then the lower RAM can easily be fried.

So, of course, I followed standard procedures and carefully ran through all the required tests, didn't I?

Nope. I was a dumbass. I plugged it straight in. This was the result:

Oh s**t!!
In addition to the red border and garbage on screen, there was a random screech from the Spectrum's speaker, just to add to the excitement.

At this point, I opened him up and decided to run through the tests that I should've done at the start. (I noticed the keyboard membrane was borked but that's for later.)

Issue 4A. Nice.
The board is a 4A with a 6C001E-7 ULA, made in August 84. Out came the multimeter.

Test one: Voltage regulator left terminal resistance to ground. PASS
Test two: Lower RAM ground rail resistance. PASS
Test three: Lower RAM 5v rail resistance. PASS
Test four: Lower RAM -5v rail resistance. PASS
Test five: Lower RAM 12v rail resistance. PASS
Test six: Inductor coils end to end resistance. PASS
Test seven: Inductor coils cross coupled resistance. FAILED

Fsck. So reading resistance through each coil was fine i.e. the resistance was very low. But testing across the windings results in low resistance too. JPC suggests that this is down to TR4 failing with a short across its collector and emitter.

Test eight: Diode test base to collector on TR4. FAILED
This test should show 0.6 to 0.9v but instead the meter read 0.00v.

Test nine: Diode test base to emitter on TR4. FAILED
This test should show 0.6 to 0.9v but instead shows O/L i.e. no connection.

Test ten: Diode test base to colelctor on TR5. PASS
Test eleven: Diode test base to emitter on TR5. PASS
Test twelve: Diode test (reverse polarity) base to collector on TR5. PASS
Test thirteen: Diode test (reverse polarity) base to emitter on TR5. PASS

After all that, TR4 is toast. I removed it from the board and re-ran the tests and got the same results i.e. base to collector read 0.00v and base to emitter read as open circuit. So, I need a new TR4. These were promptly ordered from RS Components for the princely sum of about 35p each or so in a pack of five. They were not the original ZTX650 but actually ZTX651 which are an appropriate drop in replacement.

A day later, they arrived and I set about soldering a new one. This was straightforward as the ZTX651 has the same pinout as the ZTX650. Some other replacements that could be used may not have the Base, Collector and Emitter in the same order.

ZTX651s and some extra caps too.

With a new TR4 installed, I went through the initial tests again and, this time, they all passed. Back to the desk for switch on number 2.


Bugger.
With the Smart Card V2 plugged in, this was all I could get. Vertical black and white bars. The only thing that changed was the colour of the background which would sometimes be blue and sometimes black, very occasionally it would be white. Now, at this point, I'd like to say that I started to channel my inner Sherlock Holmes and deduced that, if the display is working but nothing will start then it is most likely that the CPU is toast. I'd like to say that but I can't as I had to ask for assistance from Twitter and the ZX Spectrum Facebook groups. Many suggestions were made but the one that kept coming up was that the CPU was toast so I decided to look at it more closely.

Rather annoyingly, the Z80 is this Spectrum was soldered to the board. In my Issue 2 the Z80 was in a nice, convenient socket. So I would have to remove this one from the board if I was to have any chance of proving the problem. I did not need to order a new Z80 as my CPC has (had) a socketed Z80 which I could 'borrow' for a while. What I was missing was any good way of de-soldering a 40pin DIP.

I looked through RS Components website and found decent quality solder wick but I have never really had much luck with it. The good stuff can also be expensive and is, obviously, single use.. Then, on Amazon, I spotted a de-soldering soldering iron for a fiver. What could possibly go wrong? :)

Cheap - but will it make me cheerful?

It's essentially a standard mechanical solder sucker but bolted into a soldering iron. And although there are plenty of reviews and YouTube videos suggesting these are a bit, well, crap, my experience so far, is that they are massively easier than solder wick and a standard solder sucker. I practiced on a scrap board and the components were just falling out of the board into my lap. Which was nice.

To cut a long story short, I successfully de-soldered the Z80 and carefully removed it from the board. I was lucky to have a 40 pin, turned pin socket that would drop straight into the empty space, closely followed by my borrowed Z80.

Socketed Z80

Back to the desk for switch on number 3.

Almost there....
The excellent Smart Card V2 started up this time so the Z80 was clearly the original problem. But just as my excitement grew, I heard the deep beeps that told me I had a RAM problem. IC17 to be precise.

Out came the de-soldering iron again and the offending chip was quickly dispatched and socket installed. Next problem, I wasn't sure I had any appropriate RAM chips to fill the gap which needed to be type 4532 or 4164. I didn't have any of either but I did have a few 4264s. Google was my friend and it confirmed that a 4164 can replace a 4532 but a 4264 could replace a 4164. By my logic, this meant that a 4264 should work.

4264 RAM Chip Installed
And, by jove, it did. :D

Lower RAM Done - What about the upper??
GET IN!

The main board is repaired but still needs to be re-capped. But for now, let's get the top on and try out that keyboard. The unit was sold as having a keyboard fault, presumably the membrane so I had already ordered and received a new one. Sure enough, the original membrane looked a bit crappy and a portion of keys in the centre refused to work.

And then the real problems started....(to be continued)




Thursday, February 07, 2019

Replace the word 'Song' with 'Pong' in any film title...

Some classics there. The controversial 'Pong of the South', the obscure 'My Pong goes Around the World' or the immortal 'The Pong Remains the Same' from Led Zeppelin.

Anyway.

The old Interstate console has been sat waiting patiently for some attention. Well, the donor Interstate machine has now arrived and, despite being in great condition, will soon be giving up its parts for the collective good. Comrade.

And kudos to the eBay seller who provided the best packaging I think I've seen from anyone on eBay.

Great Packaging

Super Clean Donor Unit

My console works without any issue other than the joystick and so the first thing to do is remove the right stick from the donor machine. On these 'cheaper' models the sticks are wired directly into the console rather than having a handy DIN plug so there's no other option than to remove the top.

First, pull off that giant knob. (No sniggering at the back.)


Insert Schoolboy Joke Here.

It should come as no surprise that the case of the donor machine is suspiciously like the case on mine, even down to the holes that are drilled in the back, but covered by a stick on plate on the donor machine.

Holy moley Batman!

It did come as a surprise at how 'cost reduced' this unit actually was. The video cable was soldered straight into the back and the 'modulator' for RF is basically a tiny circuit board with a few resistors on it...

Nothing Fancy Here

There are also large gaps in this circuit board where components could have been installed, presumably for other models.

What would go here I wonder?

Donor Unit Top case - spot the speaker...
There's no speaker in this one, or rather, there is, but it's a tiny piezo thing in the top case rather than a proper speaker in the bottom of my soon to be repaired unit.

My unit for comparison - it's still not exactly an XBox..
My unit, in comparison, has a proper speaker and an attempt at a separate RF modulator unit. Note the green wire which is the composite mod. Also note that the front 'dome' type buttons are identical between the units.

Looking left and right, the joysticks are connected to each far end of the board and I just cut the wires for the right stick and removed it. I did leave the grommet in the case, just in case this one ever gets put back to its original condition (unlikely, but possible).

It's very, erm, basic?
So now I have a joystick with a cable. What I planned to do was to desolder the cable at the potentiometers by disassembling the joystick. Then the block with the potentiometers will be soldered to the end of my cable which has the DIN plug on the end. At this point, I noticed something I hadn't expected. My joysticks had connections to all three lugs on each potentiometer. However, the donor unit only has two lugs connected on each pot.

One, two...oh, where's three?

Hmmmm. If I have time I may investigate further. I suspect that the donor console games may not have the complete range of movement that mine have. For now, I shall just swap the donor stick to my console.

This was easier said than done...

The wires in the original cable are quite short and the thick 'Vcc' cable was actually badly corroded when I opened it up originally. As a result of re-soldering, the wires are a VERY tight fit now. The end result was that the joystick worked perfectly. Almost.

For some reason, there was a noticeable 'jitter' from left to right on the repaired stick. Sometimes this would make the 'bat' jump from one side of the screen to the other then back again. Opening it up I realised that a strand of the 'Vcc' wire was loose and flexible and was probably making contact with one of the other potentiometer contacts. I removed it from the centre leg and re-soldered it to the other cable connected to that point. After another squash and a squeeze to get the thing back together, the bat was rock solid. But now it won't move left or right. Arse.

Pass me the screwdriver. Again. This time, I will solder longer wires to the pots then solder the original cable to them, making sure I use heatshrink to prevent any shorts. There's not a lot of room in them sticks!

This is taking longer than I thought...