CPC6128+ repair job

[ralferoo]

Just googled the Aoyue 8208. Looks like I should have bought that instead of the Chinese one I actually got. It's about 50% more expensive but I'd say probably worth twice the price...

[Bryce]

Although an SMD rework station is a handy thing to have, it usually gets used a lot less than the soldering iron. So I wouldn't spend too much on one. I would recommend spending a few hundred euro to ensure you have a good branded soldering station such as Weller / Ersa, because you need to be sure that the tips which (should) get changed regularly are still available in a few years, but an SMD rework station is just a glorified hairdrier and shouldn't cost more than €100 unless you really do use it daily.

Bryce.

[Badstarr]

I should really get one of those SMD rework things! Mind you I've removed a fair few ICs without any damage using my cheapo heat gun, that's how I made my BASIC/Burnin Rubber cartridge. I wouldn't want to risk it on a precious ASIC though. I have often wondered how precise ASIC emulation is on various software emulators and whether or not that emulation could be pushed into a FPGA that could be physically soldered in place of an ASIC. A rather expensive solution I know! But pluses are so damned rare and expensive and I would guess that all the ASICs that were manufactured went directly into a machine without a second manufacturing run being produced when it became apparent the Plus and GX4000 machines didn't have legs commercially. I could imagine the conversation between Lord Sugar and who ever was in charge of the Plus manufacturing went something along the lines of "So these new computers with the expensive custom chip we designed aren't selling well then?... Well I aint throwing any more money at it... and for that reason... YOU'RE FIRED!"

[Bryce]

Yes, but unlike an electronic hobbiest, large companies usually sign off supply contracts before they've sold the first unit, so depending on what their marketing people were saying, they might have signed a deal to take a few thousand ASICs per month for a set amaount of time, so even if they not selling the projected amount of units, the parts are still arriving. That's the only way you can get the component prices down to the level you want/need. The questions remain: Where were they being delivered to? How much stock piled up after the production was ramped down? And what happened to the parts afterwards?
Having worked in this industry, I could imagine, that (if they weren't subsequently dumped), there could easily be 10's or even 100's of thousands ASICs stored somewhere.

Bryce.

[beaker]

I would recommend spending a few hundred euro to ensure you have a good branded soldering station such as Weller / Ersa, because you need to be sure that the tips which (should) get changed regularly are still available in a few years
Bryce.

So time to throw out the Antex soldering iron that I borrowed from work 7 odd years ago and forgot to return if I want to do the ACID inside mod?

[dragon]

How much stock piled up after the production was ramped down? And what happened to the parts afterwards?Having worked in this industry, I could imagine, that (if they weren't subsequently dumped), there could easily be 10's or even 100's of thousands ASICs stored somewhere.

1,is in usa.(Is the domberg sell).


These are the component survivors of amstrad:


http://www.netcomponents.com/es/results.htm?r=1&t=f&src=&so=&lgc=begins&d=1&pn1=amstrad&slgc1=begins[size=78%] majoritary,older gate arrays.+400 acids chips.[/size]

From,uk,france,netherlands,depend the source.

[Bryce]

@beaker: If it's one of those small yellow non-regulated ones, then I would only use it to solder wires, not components. If you seriously intend soldering an ACID, then I would only recommend a regulated iron, where you know exactly how hot it is.

I've gone around in circles with these types of companies in the past trying to get rare parts. - Supplier A says he has a source in the US, but that's also a re-supplier who thinks he can get the part in Korea, but that's also a re-supplier who thinks he can get one from the original Supplier A. Also, there are several re-suppliers who scan lists (which are way out of date) from other suppliers or that automatically generate lists of all known parts and claim they have them in stock. So I gave up using these "services" years ago.

Bryce.

[Badstarr]

I went round in circles trying to order an FDC with these sorts of companies. Just as Bryce said they tend to cross link themselves to the point where each "supplier" is referencing the other as a stockist. I was trying to get an exact replacement, right down to the last id digit. In the end I took a chance on a couple of FDC's on ebay that matched up when I looked at the data sheets and I confirmed they worked ok by installing them in my Plus machine. Even the SED data separator was labelled as an 'A' rather than 'AC' and that too works fine. Like I've said before, I'm no expert on IC code numbers so I suppose that as long as the main ID is correct anything that follows may not be that relevant in terms of spec.


Actually, a quick question, without the intention to hijack the thread, as you can see in the picture posted by Spirantho, there is a variable resistor near the RGB connector, does anyone know what this should be set to? I just want to check that mine has not been disturbed at any point during its lifetime. Spirantho, if you could check for me while you have the bonnet up, so to speak, I would be very grateful!   

[Bryce]

If you ever need parts like the FDC or whatever, then post a request here, usually someone (like me) will have a few spares. The second suffix on an IC number rarely signifies a functional change, so Type A can be considered equal to Type AC.

The Mini-Potentiometer beside the Monitor output is VR101 which is used to set the peak value of the Luminance. If you use a green monitor, then you can tweak the brightness of the display slightly by turning this. If you use a colour monitor, RGB, SCART, S-Video or whatever, then it really doesn't matter where this is set to, because you're not using the signal anyway.

Bryce.

[Badstarr]

Briliant! Thanks Bryce! 

[Spirantho]

A little upate...


I received the broken GX4000s today and sure enough, they're broken. In fact, both of them had two dead RAM chips. I mean, seriously, what is it with Plus machines and their RAM chips?!?
Anyway, having replaced them, I now get two almost working machines. One of them shows no RGB/Audio, but if you tune it in you can see the image and there's sound there too - Burnin' Rubber is working. The other one also has no RGB/Audio, but again there's a TV picture - quite a bit stronger, this time - but missing colours and still rather weak - but again, it runs BR.
I lifted the ASIC from the stronger of the two machines, and I've put it into my 6128. At the moment, though, there's still some pins not quite touching the pads, so I probably need to touch them up slightly. There's no signs of life on the 6128, but the RAM is now getting a CAS signal.
My question is this: Could the weak TV signal be a fault of the ASIC? I assumed not as the game was running visibly, including a (weak) sync to go with the (weak) graphics. What should I look at to correct the picture, if I need to? On the GX4000 there's a TV modulator, but there isn't on the 6128.....

[Bryce]

The GX4000 produces the RGB picture exactly like the 6128 does ie: ASIC produces 3x 4bit colour signals and sends them to the 3x 4bit DAC to make analogue RGB and these can be measured as mentioned earlier. However, the GX4000 also sends these (and the Audio Signal) to a CXA1145 Encoder to produce the SCART and TV signal. If the CXA is blown it may be causing problems there. But the CXA also drives the UM1234 Modulator, so if this was dead, you wouldn't get a UHF signal either.

Eitherway, No the ASIC can't be to blame for a weak picture signal.

The variance in picture strength might be due to some idiot having tried playing with the modulator settings. As you are accustomed with Speccy repairs you'll know how sensitive the Speccys UM1233 can be, well the GX4000 uses the UM1234 which was no better

Does the SCART work? When you say RGB not working, are you referring to the monitor Socket RGB or the SCART RGB (They are two totally different signals).

'Bryce.

[Spirantho]

Interesting. I was talking about the monitor port. Not tried the SCART one..... need to do that....

[Bryce]

Strange, the RGB and sync signals to the monitor connector are the same signals used to create both the TV UHF and SCART pictures, so if the TV works, then the RGB and sync must be present at the monitor socket. Are you sure your monitor works?

Bryce.

[Spirantho]

Yes, I'm sure.  It works with my working GX4000 anyway...

[Bryce]

Then check the RGB signals on pins 15,16 and 17 of the DAC (AMS48464). They must be there, otherwise you wouldn't be getting a TV signal and confirm that the same signal is getting to the monitor socket. Then check that your ASIC is sending an NSYNC signal and that this is also making it to the socket (it goes through a 220R resistor, so expect a voltage drop).

One reason for the dodgy DRAM in the CPC+ might be the fact that the Plus actually uses two different Ground plains. The DRAM had it's own ground plain that (extremely strange decision) was also used by the video circuitry. A spike on this plain could be the reason for the deceased DRAM and might also explain why something on the video circuitry is also not doing what it should.

Btw, If you examine the Plus schematics closely you'll find several other forehead-slapping decisions that no sane designer could ever justify.

Bryce.

[CanonMan]

Btw, If you examine the Plus schematics closely you'll find several other forehead-slapping decisions that no sane designer could ever justify.

Bryce.

Go on, Bryce. Spill the beans!

[MaV]

Go on, Bryce. Spill the beans!

Seconded! I'd like to hear about them as well!

[Cholo]

Btw, If you examine the Plus schematics closely you'll find several other forehead-slapping decisions that no sane designer could ever justify.

Bryce.

Sanity has very little to do with it indeed .. but i bet some of it is cost-effective 

Check this video at 04:00-ish:


"Why is it there? Why?", lol.

Later he also tells about his great idea of squeezing everything into one.

[Badstarr]

I've noticed on one of my GX4000s that the monitor connection puts out a week signal. Its washed out and rather grey, I assumed that this could have been perhaps due to the monitor socket being active when you connect the 5v psu. It was just a hunch really, if the monitor output is live all the time then its certainly a bit weird that this happens. 


As for getting the consoles going, well done! I think I may have some more as yet non investigated GX4000 boards knocking around that I might have try through the modulator, however, the majority of the ones I have got I have pulled the RAM ICs and they were working fine so I guess that wont fix them.


What I would say though, is that every time I connect my project GX4000 to its PSU, the TV senses it switches automatically to SCART. On the dead ones I've tried this does not happen. I began to wonder if the SCART pin 16 (blanking signal) might have lost its feed so the tv receives no sync? So no on screen activity makes the console appear dead? However, I would assume that the audio would still come through?


I really hope you manage to resurrect the 6128! I think it would be the first time its ever happened, but feel free to correct me if I'm wrong!

[Bryce]

What I would say though, is that every time I connect my project GX4000 to its PSU, the TV senses it switches automatically to SCART. On the dead ones I've tried this does not happen.

Sadly (and this is one of the very questionable decisions on the GX4000), the 10V signal is responsible for the Automatic switching of the TV, so that must be dis-connected or broken on the broken GXs. It would have been so easy to solve this without an extra supply / connector etc, but that was the Amstrad solution??

I'm not going to go through all the strange decisions, but... Unnecessary pull-up resistors and capacitors, non-optimised boolean solutions and the CPC could also be modified to require quite a bit less current without changing a single IC. And if they really wanted to make a low-cost computer, then they PCB could have been half the size (ok, admittedly they did do this on later versions).

Bryce.

[Badstarr]

Perhaps the weird design decisions are the result of having a team of radio engineers produce the CPC range. I read somewhere recently that this was the case. I do remember when I found my first, and none working GX4000, thinking that it was a crazy idea to have a twin PSU solution. I figured that the previous owner must have attached the 11v PSU to the 5v line. After opening the machine up and seeing that ASIC, I decided that a repair was probably beyond my knowledge and eventually threw the machine away. This was back in days when we had no CPCWiki and very little info on the WWW about the GX4000s. I really wish I hadn't gotten rid of it now, as I left all the ICs in place which is tragic really as I would have found some of them useful these days! I would think that between all the broken GX4000s I have found over the last year or so I could cobble together a working machine. Thats a project for another day I guess.

[CanonMan]

Perhaps the weird design decisions are the result of having a team of radio engineers produce the CPC range.

Where did you get that from?

The original CPC certainly wasn't designed by radio engineers, Mark Eric Jones and Roger Hurrey had both worked for a company that designed very expensive word-processor computers.

I'm fairly certain that MEJ and his team designed the plus range as well.

It's all in "The Amstrad Story" by David Thomas. A very good book, IMO.

[Bryce]

I think that was exactly one of the problems. The designers were accustomed to building expensive highend hardware and had no idea how to build low cost hardware.

As far as radio engineers are concerned, I think this was the case with Spectrum, but not with Amstrad.

Bryce.

[CanonMan]

I think they did a pretty good job, given that they only had about eight months to do it in!