QL freezing after an hour

[markdryan]

Hi All,

I'm having some issues with my QL (Issue 5). The machine boots up fine and works without issue usually for about an hour before freezing. When the machine freezes there's no screen corruption. All that happens is that the cursor stops flashing and the computer stops accepting input. If I press the reset button the machine resets without issue, passes the Minerva memory tests, and boots normally. It will work fine for about 10 minutes or so before freezing again. I've recently fitted a Hermes chip and have noticed that I still hear the key click noises when pressing a key after the machine has frozen. The keys continue to make the clicking noise even after lots of presses.

Here's what I've tried.

1. Removed, cleaned and replaced all socketed chips (as recommended by the service manual).
2. Removed all the expansions (memory upgrade and QIMSI) apart from the Minerva ROM and the vmap plugged into microdrive 1.
3. Replaced the CPU and the 8049.
4. Replaced the voltage regulator with a modern replacement (Murata I think).
5. Reflowed the solder on about half of the board. The solder looked quite degraded in places under the microscope (particular around the sockets). I've reflowed all the sockets and the connectors.

None of the above has made any long term difference. I say long term as sometimes after I've been fiddling around inside the computer it works fine when put back together. Often after "fixing" the machine, I'm able to use it for over 3 hours without experiencing any freezes. I congratulate myself that I've finally fixed the issue, but after a few weeks, it's back to its old tricks.

I suspect the issue is heat related. On Monday (when it was really hot in the house) the machine hung after only about 20 minutes of use (as opposed to the normal 50-60 minutes).

Does anyone have any suggestions on what to try next? I think my next plan is to try replacing the sockets on the motherboard and perhaps placing heatsinks on the main chips. As this will be quite a lot of work (and not without risk) I thought it best to check to see whether there were any less drastic things I should try first.

Thanks in advance for any advice

Mark

[tofro]

Hi All,

I'm having some issues with my QL (Issue 5). The machine boots up fine and works without issue usually for about an hour before freezing. When the machine freezes there's no screen corruption. All that happens is that the cursor stops flashing and the computer stops accepting input. If I press the reset button the machine resets without issue, passes the Minerva memory tests, and boots normally. It will work fine for about 10 minutes or so before freezing again. I've recently fitted a Hermes chip and have noticed that I still hear the key click noises when pressing a key after the machine has frozen. The keys continue to make the clicking noise even after lots of presses.

Here's what I've tried.

1. Removed, cleaned and replaced all socketed chips (as recommended by the service manual).
2. Removed all the expansions (memory upgrade and QIMSI) apart from the Minerva ROM and the vmap plugged into microdrive 1.
3. Replaced the CPU and the 8049.
4. Replaced the voltage regulator with a modern replacement (Murata I think).
5. Reflowed the solder on about half of the board. The solder looked quite degraded in places under the microscope (particular around the sockets). I've reflowed all the sockets and the connectors.

None of the above has made any long term difference. I say long term as sometimes after I've been fiddling around inside the computer it works fine when put back together. Often after "fixing" the machine, I'm able to use it for over 3 hours without experiencing any freezes. I congratulate myself that I've finally fixed the issue, but after a few weeks, it's back to its old tricks.

I suspect the issue is heat related. On Monday (when it was really hot in the house) the machine hung after only about 20 minutes of use (as opposed to the normal 50-60 minutes).

Does anyone have any suggestions on what to try next? I think my next plan is to try replacing the sockets on the motherboard and perhaps placing heatsinks on the main chips. As this will be quite a lot of work (and not without risk) I thought it best to check to see whether there were any less drastic things I should try first.

Thanks in advance for any advice

Mark

Such apparently heat-related issues are hard to resolve. The classic way of detecting what's overheating is using freezer spray (to cool down chips that apparently fail through heat and see if that improves anything) or a heat gun (to heat up selected chips to see whether that speeds up the issue.). Hermes is apparently not the issue as the keyclick still works (and than needs only a working IPC). It might even be no chip at all but rather the PCB (µ-cracks in traces) or sockets/bad solder joints that lose contact due to thermal expansion.

[Popopo]

Hi!
Is your CPU ceramic or plastic version?

You could post a picture of your PCB here in advance.

The heat is one of the most common problems that could freeze the computer after a while. But not the only one.

Caps and Transistors are pretty bounded to the working temperature too.

It is not an easy task to know which component (or group of) is responsible of it without thermal tools.
But since you have done a lot of tests and described which have been done, I think we could manage to reduce the number of possible sources.

The picture or some pictures of your board may help (very much).

[Pr0f]

The keyclick is actually produced by the Hermes chip - as it drives the sound and reads the keyboard - so provided that has not crashed - it will continue to 'click' when you press keys. So unfortunately - that's not a good indicator that the machine has 'locked up'.

Re socketing chips - and if you can, cleaning their legs with some deoxidizer can help - the stuff guitar players use is good - but it's not cheap. Try and avoid abrading pins - as they just oxidize much faster after that.

If the screen is still working - you could try putting the QL in the fridge overnight (seriously - put it in a plastic bag, with some uncooked rice and let it cool down for a good long time) - then see if you can get longer use period out of it - the idea of this test is to determine if it's a general contact type issue, solder joints which can be affected by heat buildup, or actually a failing component - in the last case, the cooling won't buy you much additional time, as the localized heating if a component is on the way out - that will still happen pretty quickly.

Also - running a program that writes updates to the screen frequently - that may help in determining when exactly it goes unresponsive.

[markdryan]

Here are some pictures of the top side of the board. Pictures of the bottom of the board, will take a bit more time as I'll have to take the whole thing apart. Let me know if you need to see the bottom side as well.

I can post close ups of some areas if required. I realize I still need to find a better way of attaching the new voltage regulator to the board.

[Pr0f]

Interesting choice of regulator - does it lock up if the original QL regulator with it's mahoosive heatsink are used?

[markdryan]

Interesting choice of regulator - does it lock up if the original QL regulator with it's mahoosive heatsink are used?

Yes. I replaced the regulator in the hope that the machine would run cooler, which it does, but it didn't stop the freezes.

Looking back at my own pictures it doesn't look like I did a great job at cleaning the legs of the ZX802. Here's a picture of the other side of that chip.

This is the socket which had the most degraded solder and which I reflowed first.

[Pr0f]

This is the contact cleaner I've used in the past:

https://amzn.eu/d/bg8tnaa

Like I say - not cheap - but it's effective - Search for Deoxit. I am sure there are other capable brands too - I found that one first after a recommenation from a guitar playing friend.

[Popopo]

Plastic CPU version I see.
That doesn't get too hot. So I would not consider the CPU responsible by now.

I don't see the 2 resistor mod that should lay on the ZX8302, perhaps you removed for cleaning purpose. They were there?

Capacitors doesn't look so bad.

Do you have oscilloscope? It would be interesting to check the clocks signals in case the searching for a hot IC fails.

The DCDC that you use for +5v is not good enough. I have used it for testing purpose and it may works barely.
That is a Chinese one. Please to use a linear one or something better (maybe TracoPower).

[markdryan]

I don't see the 2 resistor mod that should lay on the ZX8302, perhaps you removed for cleaning purpose. They were there?

No. This is how I got the machine. I think the mod may have been there in the past but was removed by the previous owner. What does the mod do? It doesn't prevent freezing by any chance does it?

Capacitors doesn't look so bad.

Is it common practice to replace the capacitors on a QL?

Do you have oscilloscope? It would be interesting to check the clocks signals in case the searching for a hot IC fails.

I do although I must admit I don't really know how to use it properly. I can probably measure a clock signal though. Is the idea to measure it when the freeze occurs?

The DCDC that you use for +5v is not good enough. I have used it for testing purpose and it may works barely.
That is a Chinese one. Please to use a linear one or something better (maybe TracoPower).

IIRC the replacement voltage regulator is an MuRata OKI Switch Mode 5v rated up to 1.5amps. I bought it originally for a 128k Toastrack but never got around to fitting it to the Spectrum in the end. The freeze was happening before I replaced the voltage regulator but if this regulator isn't recommended, I'll swap it out. Thanks for the heads up.