News:

Printed Amstrad Addict magazine announced, check it out here!

Main Menu
avatar_JonB

CPC6128 white screen on startup

Started by JonB, 12:04, 18 June 13

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

gerald

The CPC only need the 1st bank (IC127-134) populated to run.
The 1st bank is the right one when you have the CPC in front of you.
The second bank is not used by FW, and you do not need it right now.

You can also check the PAL (IC118), which is already on socket. It is dealing with the memory access.
You can even try without it by closing LK5/LK6/LK8.

Next on memory/cpu data path are the bi-dir buffer made by IC115 and IC114. If IC115 is dead, write to memory and GA will not work. if IC114 is dead, read from memory will not work.

JonB

#51
Alas, swapping out the gate array or the other socketed chip (IC118 - 40031) hasn't fixed it, neither has swapping out the Z80.

:(

I suppose I will have to swap out those buffer chips next. I tried piggybacking an LS244 over the one on the board but it had no effect. Don't have an LS373 right now.. any way to test these in situ?

Munchausen

Hmmm. I guess you could test them in situ, but it will be a bit of a pain.


Here's the pin out:





The output control pin enables/disables the outputs (Q pins), I think it is active low, so to enable the outputs this should be low. Then you should see that whatever you put on any of the D pins (a high or low level) appears at the corresponding Q pin whenever you transition the enable pin from low to high.


So you could test it in place by connecting the power lines and control line up, putting inputs on some of the pins and toggling G, and checking the outputs with a multimeter but it seems a bit tricky to me (you'd probably need a couple more pair of hands to get it done without soldering!)


I'm quite surprised you haven't managed to coax some kind of life out of it yet, TBH... I have problems I haven't figured out yet on both my 6128s, but they worked to varying extents without this much trouble. You'll get there if you keep at it though!

Munchausen

Quote from: JonB on 22:04, 25 June 13
Alas, swapping out the gate array or the other socketed chip (IC118 - 40031) hasn't fixed it, neither has swapping out the Z80.


Also, just in case: tried both at the same time, and the PAL (40031)?

gerald

Have you tried the other way : testing IC from defective CPC in a working one, one by one ?
I would also re-check the RAM path with a multimeter for shorted or open track

JonB

@gerald: It's a good idea. I'll see to it. As to the ram path, yes, it's possible there is a problem there of course. I noted the 6128 PCB has particularly thin tracks in places, which makes it fragile. A close examination under a magnifying glass reveals nothing obvious, but you're right, the meter is the only way to be certain...

endangermice

Checking continuity with the schematics is definitely the way to go. I spent ages trying to get my 6128 board to work but in the end I did manage to track a single broken trace on the data bus. It's time consuming but if there is a break somewhere this will find it!


I think if you have a border the gate array is probably working. Are you able to check the address bus for pulses, assuming you haven't already done so. This will tell you whether the CPU is addressing anything. It's possible that it starts then crashes which is what was happening to me with the broken data bus.
For all the latest Starquake remake news check out my website - www.endangermice.co.uk

Bryce

You should also keep an eye out for dry joints. These will usually look dull compared to the other joint, or sometimes you will see a tiny crack around the pin in the solder.

Bryce.

JonB

@endangermice: Yes, pulsing. I checked with a logic probe.. :)

@Bryce: Good advice, thanks. As I said, I inspected the board very carefully using a magnifying anglepoise type lamp (the sort with a ring shaped flourescent bulb with a huge magnifying glass in the middle) and could not see anything in particular... except that all joints look a little dull, which is probably down to the age of the thing.

Bryce

Quote from: JonB on 15:04, 26 June 13
except that all joints look a little dull, which is probably down to the age of the thing.

That may be your problem. It can't hurt to slightly freshen up those joints. You don't need to remove the old solder, just reheat each one with a tiny bit of new solder and they'll be like new. It's no guarantee, but it may solve your problem.

Bryce.

endangermice

#60
This is very good advice, at worst a dry joint will render the contact useless, it
may also make a part connection with a large amount of resistance which will play havok! I had a SID chip recently that I thought was knackered - it was playing very quietly and out of tune. In the end it turned out to be a bad joint on the VCC which caused the 5V to drop to 1.74. I was amazed it worked at all!

Have you checked the RAM address and data lines for pulses, at least the first 64KB not sure if the second bank is addressed until switching has taken place....
For all the latest Starquake remake news check out my website - www.endangermice.co.uk

Bryce

The most annoying dry joint is when they completely break through and then the surfaces of the break oxidise. It creates a perfect very low value capacitor that varies with temperature and creates the most random results possible. :D Especially on data or address lines.

Bryce.

endangermice

It's a shame that this "freak" capacitor can't restructure itself into something useful like say an extra bit of RAM ;).
For all the latest Starquake remake news check out my website - www.endangermice.co.uk

JonB

#63
Hmm, how many joints are there on that motherboard ?

I was thinking it least I might wash it with isopropyl achohol to remove any flux resue or other crap; then I could perhaps spot dry joints a bit easier. I am not sure I have the patience to resolder every joint, and I bet if I did it would make no difference.

As you can see I am losing faith. Don't want to spend much more money on it as it will be uneconomical.

endangermice

Totally understand, I know how infuriating it can be! How are you getting on with the other CPC, have you managed to get the keyboard up and running yet?

To be honest once you get going with the soldering iron you'll be amazed at how fast you can re-melt all the joints it really doesn't take that long at all.

One thing you could do (and I've done this on a C64 in the past) is assuming the other machine boots to the Basic screen (I think it does from what you've said) you could spend some time with your logic probe measuring the results on the pins of each chip and comparing to your broken board. Start with the major chips i.e. CPU, ROM, RAM. It does take some time and the easiest way is to note down you results which be be along the lines of low, high, pulse, pulse low, pulse high etc.


If you find a chip that's behaving differently to the working machine then you need to identify what that pin does, is it an input or output. If it's an input then maybe the chip supplying that input isn't doing the right thing, so check there - it could all boil down to a broken trace or bad joint but you might be able to identify it a bit more easily using this method.


It's not an exact science but with some knowledge of how the machine operates it should help you towards diagnosing the problem. It will obviously help if the two machines are the same model though by and large you should get similar readings on most chips anyway.


The only thing you need here is time since you already have the testing tool.
For all the latest Starquake remake news check out my website - www.endangermice.co.uk

JonB

OK, thanks; again it's good advice. I think my order of working will be:

1. Clean the board with alcohol (heh, I could use some of that myself!) and check for dry joints.
2. Resolder dry joints, or all if there are no obvious ones (  :( )
3. Check the data lines with a meter.
4. Check the IC readings with a meter.

Obviously retest between steps..

6128 #2 is up and running, with the keyboard and (re-belted) drive of 6128 #1. So at least I can start playing, indeed, I have done. I even ordered a copy of "CP/M The Software Bus"..  :laugh:

I also took delivery of a pair of Amigas (A500 and A1200) but there was a problem with transit damage and a slight misdescription and considering I paid a goodly sum (due to the way the lot was described) I am engaging with the seller and hoping for an amicable settlement whereby I can keep them. That said, I've been playing with them (primarily, to ensure they work, which they do) and I am not certain when I'll be back in 6128 land. A few days, for sure...

Bryce

Re-soldering really doesn't take that long. You only need to bring the solder to melt and jab it with some fresh solder at the same time (to make sure it gets some flux), which only takes 1 or 2 seconds per joint. I'd need about 30 minutes for a full 6128, but I've had a lot of practice, so you might need a little longer. But it's a well invested 30 Minutes. Make sure you clean the tip every four or five pins too. Set the iron to 260°C, then you're certain to be low enough for even the most sensitive parts on the board.

Bryce.

Powered by SMFPacks Menu Editor Mod