I Crashed My Fiesta: A Catalogue of woe...

After looking over my unexpectedly large collection of cassette recorders, I decided it was finally time to test which ones would be suitable for use with the faithful old Speccy. There are a couple of reasons. First, I like things to be original where the can be - although I'll never shy away from a faster loading mechanism if it's a good price - and second, if they work with the Speccy then there's a good chance they'll also work with my recently acquired Electron (as long as I can get a cable).

So, down to business, I went and pulled my Spectrum+ out of the box upstairs where it had been stored for the last eight months or so. Main unit? Check. Power supply? Check. Smart Card V2 from Retroleum? Check.

And then, the first problem. Powering up I was greeted by the happy beep of the Smart Card but I couldn't move the cursor up and down to select a game. How odd. I switched off, removed the SC and powered back up. Much to my disappointment and annoyance a large proportion of the keys had stopped working. This should not have been a surprise since it was still the original membrane but I had spent a long time 'tuning' the keyboard to keep it going and it was actually 100% working when I put it away. Sadly, the tails are now too far gone and it's game over for this membrane.

Tails, tails, tails

I hate this keyboard

But no problem. I already have a new membrane that I bought a while ago. It's been sat in a drawer for about a year. Out it came. BUT. This one has corrosion on the tails too! What. The. Heck. For some reason, this membrane is suffering the same fate as the original despite only being a year or two old. I put it into the Speccy to see whether I might be lucky. I wasn't. Almost the same amount of keys just did not work.

Corrosion on modern membrane

Look at the two outer tracks
Brown, not gold..

At this point it was time to give up on the Spectrum + so I removed the Z80A and re-installed it back into the CPC that it had originally came from. See the results of that, here.

Plan B. Get the 48K rubber keyed Speccy out of storage and use that. So, out it came. It's an Issue 2 but we won't hold that against him. The 3D printed case I have for my Smart Card V2 (as provided by my awesome brother) lifts the rear of a standard 48K case off the desk, but that's fine. It's only for testing stuff and it's fairly stable anyway.

Turn it on. Oh, for fudges sake. It won't boot. It's stuck on a black screen with a white border which screams of a RAM issue. So I switch the SCv2 to Diag mode. But this also struggles to boot although, fortunately, this was just down to a dirty edge connector (we've all been there!). A quick rub down with some IPA and cotton buds and the diagnostic ROM boots successfully..

Dangnabit.

...and promptly identifies IC 18 as suspect.

Curse you IC18!

So out comes the de-solder station....

I have a distinct Hal feeling.

Anyway, the de-solder station makes light work of the suspect RAM IC. In a few minutes it's removed and a socket installed in it's place.

Heath-Robinson RAM Tester

But this is a good excuse to get my RAM tester out of the drawer and check that it actually IS the problem. And sure enough, the RAM tester says it's fine. Wait? What?

That was unexpected. The RAM tester thinks this chip is OK but if I put it back in the Speccy I just get a black screen. Something is clearly affecting the result. But that's for another day. For now, let's just get a new RAM chip in the socket and see if it will boot. And then I realised the only RAM I had is a couple of chips of the infamous 'MT' variety. Known for being crap, but I had no other choice and it would only be for a few minutes of testing. Urgh, I feel dirty putting this chip in there but, sorry buddy, it's only for a short time. In fact, I was so embarrassed about having to use these pariahs that I didn't even take a photo of it.

But does it now work? Yes, yes it does. :)

Testing, testing.

Annnnd, we're back in the room.

There is no way that I'd ever leave one of those MT chips in there so I needed to find a solution. Fortunately I had a stash of Sanyo 33256 chips which came from an old and decidedly broken A501 512Kb Amiga expansion. The pinout for these chips is virtually the same although they have a higher memory capacity, and consequently an extra address line at Pin 1. I thought long and hard about whether I should try to use one of these. Eventually, my ancient, addled brain decided that I could either pull Pin 1 out of the socket or connect it to ground and it should work. And would you believe, pin 1 on the upper RAM chips is connected to ground by default.

The only other difference was pin 9 which is A7 on general RAM chips of this type but is labelled as 'AR' on the Speccy. Basically, the original chips in the spectrum were specifically chosen as they had a fault on one bank of 32K of their 64K. As a result, Sinclair apparently got them cheaper but it means that the original chips have to match which bank is faulty. In this case, it's irrelevant as the chip is fully working with 64K (well, 256K really). It just means the Spectrum will only use one half (or quarter) of the chip.

Here goes:

Sanyo to the rescue

Yay! It works!

Boo! The picture is, to coin a phrase, complete garbage. It's difficult to convey how back the picture really was on this thing. Imagine the worst RF output you can, then make it a bit fuzzy, lose all colour (except a big smear of dark blue across the whole screen) and you'd be somewhere close.

Sadly, this is an Issue 2 which has a deserved reputation for this due to it having manual picture controls. Later revisions had an automatic circuit built in so the variable resistors next to the ULA (pic above) are not in there from Issue 3 onwards.

Fortunately I have a shiny new sillyscope - which is actually a generous donation from my Boss. Let's see what we can see.

Composite TV signal ahoy!

So the big dips are the 'back-porch' which tell the TV, "I'm about to start a new field'. Then the squiggle to the right of the dip is the colourburst. As the name suggests, this contains the colour information for the field. Finally, there's the main part of the field. Compare this to the display at the time:

Bad colour. Smudgy. Urgh.

To improve this the variable resistors need to be tweaked to get the video signal as 'flat' as possible. This is easiest with just the white boot screen, rather than the busy screen I had up. Once I took out the SmartCard and booted up to a plain white screen, it was much easier to see the fuzzy nature of the signal.

There is a very, very narrow sweet spot for VR1 and VR2. Once it's there, it's great. You can see the result on the sillyscope below: