VBXE problem - overheating or something else?

[morelenmir]

I have a rather annoying problem with the VBXE I fitted to my 800XL a couple of months ago. It is something that manifested immediately - even before I had the output fully wired - but I have spent the time since, off and on trying to sort it out.

 

Specifically; if I run the machine for more than half an hour - and shorter if I have not let it fully cool between tries - the display from the VBXE will become corrupt or fail entirely. The failure states are either a completely black screen or a black screen with running, flickering white lines. For diagnostic purposes I left the 'Monitor' output in place after installing the VBXE. If I switch to this composite video I do now get a screen, but the colours are badly shifted and there is heavy cyclic noise running up the screen in the background. Occasionally it will enter one of these states and then return to a normal display after a couple of minutes. Also, when doing this the rest of the machine will function correctly. I often leave it running the self-test cycles to provoke the failure and can still hear the 'Promenade from Picture at an Exhibition' snippet playing in turn behind the failed screen. I can also here key presses and so on. The forum will not let me upload the wobly-cam I took of the failure states but can supply via some other route it if anyone is interested.

 

On opening the case I find the majority of the components are cool or cold - including the "Ultimate1MB" and the "SimpleStereo"(Lotharek version). However the GTIA IC is quite warm to the touch, the ANTIC mounted to the VBXE pass-through socket is hot and the chips on the VBXE board proper are very hot. Obviously these are subjective measurements, but GTIA<ANTIC<<VBXE.

 

At first I hoped heat itself may not be the problem. Accordingly I spent much of my free time last week removing all the native sockets on the "REV C" motherboard and replaced each and every one with a precision-engineered socket. Once in place I then reflowed all the joints, first with the iron and then with hot air. I ran a final visual inspection to ensure all looked good. I then cleaned every trace of oxidation from every leg of every IC before attaching them to their new sockets. I have also replaced the stock "F 4050BPC" IC with a brand new "CD4050BE" one. You can imagine how extremely disappointed I felt when I discovered that despite all this labour the problem was still there.

 

So... Any suggestions?

 

As I say I suspect this is a heat problem - maybe playing in to a bad joint on the VBXE itself somewhere. The motherboard itself is as well soldered as I can possibly make it. I really do not want to do any work on the VBXE - it is too fine by far for my skills and only three months old anyway so well within the guarantee. I would far rather it be a failing electrolytic capacitor or other discrete component on the motherboard that I can replace. For instance I have recently read about 'C50' causing display problems. I have already counted out the "F4050BPC" and for the purposes of problem-solving restored every other IC and piece of logic with the originals that came stock. Absolutely no dice.

[fujidude]

Run it opened up with a stiff fan blowing across it and see what happens.

[flashjazzcat]

Sounds like a dry joint on the FPGA. Seen and fixed the same issue several times, if that's what it is.

[Joey Z]

Sounds like a dry joint on the FPGA. Seen and fixed the same issue several times, if that's what it is.

I'm starting to really think the only way to do these is with by hand with a microscope...

Edited August 19, 2015 by Joey Z

[flashjazzcat]

I took hi-res digital photos. Inspected close ups, reflowed, wash and dry board, re-inspect and test. Laborious but did the job.

[morelenmir]

Yeah... I was scared of that. This SMT stuff is at least part of the reason I lost interest in electronics after university. The legs on the big IC are tiny - even with a lense. I don't have a super-fine end for my - cheap, unregulated - iron, nor do I have the right sized nozzle for my vintage hot-air kit or a TV-microscope... How good is the mask?

 

At the end of the day I suppose I am going to have to have a try at fixing this myself - but it binds a bit as the thing is only three months old! Nonetheless the postage back to old Lotharek in Poland for a repair/replace would be a good chunk of just buying a new board. Why is the Royal Mail so expensive???

 

I have attached a photo which is the highest res I can get off my telephone and still be in focus. There are a few places where the work looks uneven to me so I suppose I will target those and keep retrying, taking photos to guide me as FJC suggests. What is it? A 140 leg chip? Bloody hell!!!

 

Update:

Ah, well... Sad to say my VBXE is no more. Clearly I either cooked the big chip or made a solder bridge which the strongest lense I have cannot find. Bit of a shame! All I get at switch on now is the cursor in the top left corner where it hangs permanently - the green LED lights on the VBXE daughterboard but that is it. Something of a bitter pill to swallow. Still - I am trying to commiserate with myself given the fact it wasn't working beforehand anyway!

[flashjazzcat]

Well, the cheap option would be to send it to me and if I can't fix it you haven't spent more than the cost of postage. Regarding photos: I can't deduce much from that photo owing to reflections around the legs. What I did was take a separate flash photo of each side of the large chip from a slight angle. I took lots of photos from each side, in fact: scrolling through them, it was easy to pick out the most revealing and informative ones. Sometimes I got lucky and slopping flux over the pins and then running the iron over them and then washing the board immediately fixed the problem, either by destroying some tiny bridge, removing some tiny wire strand, or consolidating a marginal solder connection.

[morelenmir]

Well, the cheap option would be to send it to me and if I can't fix it you haven't spent more than the cost of postage. Regarding photos: I can't deduce much from that photo owing to reflections around the legs. What I did was take a separate flash photo of each side of the large chip from a slight angle. I took lots of photos from each side, in fact: scrolling through them, it was easy to pick out the most revealing and informative ones. Sometimes I got lucky and slopping flux over the pins and then running the iron over them and then washing the board immediately fixed the problem, either by destroying some tiny bridge, removing some tiny wire strand, or consolidating a marginal solder connection.

Solid advice FJC!

 

Right now I'm at that throwing my hands up in the air stage. But... We'll see - give it a couple of days. Between the SimpleStereo, that PHI2 business which screwed with the "SIDELoader', compatibility issues with "The!Cart" and now this I am starting to feel a bit... cheesed off with retro modding. I've enjoyed what I've done to an extent but... Well.

 

'Calmer heads' and all that right?

[flashjazzcat]

Definitely sleep on it. Frustration is the absolute enemy of successful problem solving. Here, I have four different machines set up for VBXE, so providing a board isn't hard-soldered into a computer, the host machine can be completely ruled out as the source of problems in a matter of minutes.