48k issue 2 crash/reset at idle

[FelisSapien]

Hi all, a little brain picking if I may.

Recently picked up a rubber keyboard Speccy up from eBay. Purchased in an unknown condition, apprarently it showed the Sinclair message on boot. When connected at home I found the image to be incredibly bad. Unusable so. The composite mod had already been performed (without a capacitor) and tweaking VR1/VR2 made very little difference.

After reading up a bit I purchased new caps and replacements for TR2, TR2, VR1 and VR2. I’ve worked through and replaced components and now the image seems more stable and hopefully I can finish the job of tuning the output. First though I’ve encountered another issue...

If I boot the machine I get to the Sinclair message. Leave for a few mins and it crashes. By leave I mean no interaction with the computer at all. The keyboard isn’t even plugged in. It kind of feels like a heat issue in that if left off for a while I can return and reproduce the issue. Removing the upper RAM chips makes no difference (not that I think it necessarily would).

Could this be a lower RAM issue? Readings to the lower memory chips are good (5v, -5v and 12v) and consistent.

I’m not sure if this was an issue before the recapping as I’d no reason to leave it on for long in as the image was so poor.

The only component that seems to get hot is the ULA.

I’m awaiting a logic probe (and some free time) before diving into the detailed problem solving guide(s). Just wondering if anyone here has encountered a similar problem?

Thanks,
Carl

[Ast A. Moore]

Needs investigating, obviously, but it could be the Z80 itself. If you have a known good one, try replacing it.

[1024MAK]

In the majority of cases, it’s a RAM fault. However, if any of the following ICs are socketed, it’s worthwhile eliminating them by swapping with known working chips: Z80, ULA, ROM.

The memory test that the machine performs at power up, is only designed to work out how much RAM is fitted/connected. It does not perform any meaningful reliability test. So it is possible for the machine to get to the copyright screen, but not respond to keyboard input, and then after a while show the classic crash screen.

You can simulate a key press by using a piece of suitable solid core wire and briefly touching it on one contact in the 8 way keyboard connector while also touching one contact in the five way keyboard connector.

If it does it with the ‘upper’ (32K expansion) DRAM chips removed, then if it is a RAM fault, it’s going to be one of the ‘lower’ 4116 (or equivalent) DRAM chips.

Using a ROM based RAM test program is the quickest way to fault find this problem.

Mark

[FelisSapien]

This morning I added a socket on the issue 2 and switched the CPUs between this board and one of my +2s. Both machines seem to be in pretty much the same state following the brain swap.

At the moment things aren't stable enough for me to get any tests running. I've previously attempted to load diagnostics via DivMMC but things get funky even as the menu is displayed.

I'm wondering whether the ULA is in some way to blame as I can't get a stable picture from this board. Unfortunately, I don't have a suitable replacement chip to hand so if things are leaning in this direction this machine might have to sit for a while.

I'm tempted to switch in some alternative lower ram chips and see if that changes anything at all. If it does then great. If not then I'll have to wait until I can find a ULA replacement.

Cheers!

[1024MAK]

Because the ULA is the core of the machine, a failure in the ULA can cause all kinds of strange symptoms.
However, ULAs are generally more reliable than the 4116 (or equivalent) DRAM chips.
Mind, that does not mean that the ULAs don’t fail.

What exactly is the problem with the video output? And do you have another TV/monitor to try it on?

Mark

[FelisSapien]

The video issue goes a little like this...

https://youtu.be/hMdqWx9PMS8

The video doesn't really show the screen 'noise' that pulses along with the flicker. Towards the end you'll also see the crash I describe.

This output is pretty consistent on the LCDs screens I have within easy reach. I could also try with a CRT set but that'll require a bit more effort.

Still awaiting on the memory replacements so no hardware change since my last update.

[Ast A. Moore]

Interesting. We’d have a better idea on a “dumb” CRT that doesn’t blank the screen when if loses the signal, but it’s curious nevertheless.

It does seem like it might be multiple issues. I decided to compared the frequency of the screen flickers with the frequency of the ULA’s FLASH attribute. They’re close, but not identical.

Ideally, I’d remove the ROM and all the RAM and test the computer without it.

Incidentally, what happens if you halt the CPU in reset (short pins 26 and 29 for a few seconds)? Does that stabilize the picture?

[1024MAK]

Was the bleeping coming from the ZX Spectrum?

The weird thing is, it looks like the whole image is changing in brightness, but the CPU can’t make the border area bright (unlike the text and graphics area of the screen). The border is only controlled by three bits, giving normal brightness black, blue, red, magenta, green, cyan, yellow and white. Of course the changes in brightness could be due to the telly auto setting the picture contrast and brightness based on sampling the video signal during the “back porch” part of the video signal. But once the ULA has been powered for more than 20ms, it’s timing and video circuits should be working and stable regardless of what the CPU is (or is not) doing.

If you can connect up to a CRT TV, that would definitely help. One trouble with LCD TVs, is that they sample the video signal, then try to make sense of it. Before displaying the results. If they fail to make sense of it, they can do strange things...

One other thing, are the +5V and the +12V supplies stable and without significant ripple? By stable, I mean the DC voltage does not fluctuate by more than 50mV and the AC ripple is also less than 50mV. A oscilloscope is the best instrument, but if you only have a multimeter, test the DC voltages for at least five “cycles” of the flickering video image. Then switch to AC voltage (a 200mV range if you meter has it, otherwise the 2V or 20V range). Post up your results.

The ULA will happily produce a picture (of garbage) without a Z80 CPU present. So see if the same flickering/flashing happens with the Z80 removed from the socket. Note that the “garbage video data” will result in random attributes - random colours and some attribute squares flashing.

Mark

[FelisSapien]

Thanks for the thoughts and advice. Very much appreciated!

Doubt I'll get a chance today but hopefully over the weekend I'll whip out the chips (lower RAM & ROM) ready for socketing. Will test more then... and with a CRT monitor.

Cheers!

[FelisSapien]

Ok, ok...

ROM out and memory removed. Things seem a little more chaotic (as predicted)

Here we have the machine connected to the LCD as previous
https://youtu.be/OS4Eh4dy5ac

Now on a CRT monitor - excuse the flicker
https://youtu.be/Icx4MKvTMDg

When I first connected to the CRT it too struggled to keep the picture 'locked'. The OSD switched from NTSC to PAL every few seconds. This took me back to VR1 & VR2 for a twiddle. I wasn't able to get a 'good' image but I was at least able to remove the confusion at the monitor. Is there a quick way to get 'in the ballpark' setting these variable resistors on an issue 2? I don't have a 'scope which seems to be the official method.

Power readings from the IC6 show:
-04.55
+04.93
+11.96

It seems that without the RAM/ROM chips it's still trying to crash when things warm up (trying not to say overheating). The scrambled screen clears.

It looks as though the +12v creeps up. Over a few mins it got to +12.01. Don't know yet how that maps to the 'crash'. will report back on this as my son's just woken up, so playtime over for now...

[FelisSapien]

Would've edited the above post but it's not letting me

Got a little more time with this. Not sure what I was seeing re the 12V. Had things set up for a while now and all voltages seem stable.

On the AC measurements I see nothing on pin 1. Pin 8 = 0.045V. Pin 9 = 0.001V. The values show the upper readings when observed for say 10 seconds. I repeated the measurements over a couple of mins.

[Ast A. Moore]

In my book, your voltages seem okay. An oscilloscope would give us a better idea, however.

Here’s a pretty good guide for adjusting the video on Issue 2 boards.

[1024MAK]

The voltages (as measured with a multimeter) are good.

When I try to watch the You Tube Video’s you posted today, I get a message saying that they are unavailable

Do any of your TVs have a manual (in a menu or via a remote control button) control setting of PAL 50Hz?
As with most 1980s home computers, the ZX Spectrum range does not generate a fully compliant video picture. This can confuse some TVs.

If the TV is struggling to try to work out if the signal is PAL 50Hz or NTSC 60Hz, the displayed image may jump or change in brightness as the TV switches mode.

Mark

[FelisSapien]

I've changed the permissions on the YT videos... silly me

I'm pretty sure my CRT can be set PAL or AUTO. Will retest with PAL.

I've received the DIL sockets I was waiting for so will pop those in place for the removed chips. I can then start to play a bit more. Yesterday I also acquired a rev. 4A Spectrum (with goodies) so I now have something to compare with/to. I've already run the 4A through the JoulesperCoulomb initial tests and currently have a bit of a soak test going on. So far all is looking good expect for the TV image. In this case the issue is hopefully just the RF signal...

The 4A came with the following:
Alphacom 32 printer
4 DATA (Channel Four's Teletext/Telesoftware Adaptor)
Joystick interface (unbranded)
Some kind of expansion port extender
Power supplies, manuals and a few 'guide to the ZX spectrum' type books.

[1024MAK]

We, the senior members of the right honourable Lords of Spectrum, hereby request and demand that photograph evidence shall be produced of the following aforementioned items of interest:-

1. 4 DATA (Channel Four's Teletext/Telesoftware Adaptor)
2. Joystick interface (unbranded)
3. Some kind of expansion port extender

The photograph evidence shall be produced within three working days.

By order!

[1024MAK]

Well, your LCD TV certainly does not like that ZX Spectrum

Despite all the garbage video data, the green coloured border area did not jump or flicker on the CRT TV.

Now using the CRT TV, plug the ROM chip back in (don’t worry about fitting any RAM chips for now) and try again (this assumes that you also have the Z80 CPU fitted).

Mark

[Ast A. Moore]

Well, your LCD TV certainly does not like that ZX Spectrum

Off with its head!

[FelisSapien]

The photograph evidence shall be produced within three working days.

By order!

Here's the hardware...


Over on the AtariAge forum there's a long running 'stuff I've recently acquired' thread. Is there something similar around these parts?

Now using the CRT TV, plug the ROM chip back in (don’t worry about fitting any RAM chips for now) and try again (this assumes that you also have the Z80 CPU fitted).

Well, the colour's off but we'll deal with that later. In general things look more stable. The image location remains constant and there's no flicker/shimmer as seen before. The patterns displayed on screen seem to cycle through a predictable routine. If I leave things running for a while though it still looks as through there's some kind of system reset happening. Will get more detail on that later... Maybe time to swap out the ULA? If I put the 6C001E-7 chip from the 4A into the rev. 2 do I need to worry about the existing CPU hack?

[1024MAK]

Here's the hardware...


Over on the AtariAge forum there's a long running 'stuff I've recently acquired' thread. Is there something similar around these parts?

There is now....

Well, the colour's off but we'll deal with that later. In general things look more stable. The image location remains constant and there's no flicker/shimmer as seen before. The patterns displayed on screen seem to cycle through a predictable routine. If I leave things running for a while though it still looks as through there's some kind of system reset happening. Will get more detail on that later... Maybe time to swap out the ULA? If I put the 6C001E-7 chip from the 4A into the rev. 2 do I need to worry about the existing CPU hack?

If by CPU hack, you mean a single transistor (known around these parts as a “spider”) mounted on top of the Z80 CPU (the most often encountered method that was used *), then you need to know that in later issue boards (issue 3 onwards) this transistor took root and was properly mounted on the circuit board as TR6. It was introduced to overcome a shortcoming in the ULA logic. If nothing is fitted to the expansion port/edge-connector then it does not matter if this transistor is missing. It also does not matter if you fit a Z80’s CPU with a spider into a issue 3 or later 16K/48K/plus board.

So, just to be clear, ALL ULAs for the 16K/48K/plus boards are affected. Sinclair never fixed the problem in the ULA.

* Some issue two boards have TR6, the spider transistor mounted elsewhere on the board, and NOT on top of the Z80 CPU.

Mark

[FelisSapien]

Quick update...

Changed the ULA and didn't see any obvious difference. Got bold and inserted the lower RAM. On power up I'm treated to a new pattern of dancing squares.

Tested continuity on the RAM sockets and found a couple of issues. Need to check out the ROM socket now. There's also the possibility that I cooked the ROM when removing it. Not sure how I'd determine that. I could burn a new chip - I've an EPROM programmer I bought for something else that I'm still waiting to play with...

[Ast A. Moore]

There's also the possibility that I cooked the ROM when removing it. Not sure how I'd determine that.

Plug it into the programmer and try to read it. Then compare the dump with a known-good Spectrum 48K ROM image.

[FelisSapien]

It sounds so reasonable when you put it that way!

I'll have a read up on how I might read a 23128. It doesn't look as though my programmer supports that directly. It's a TL866II+, btw

What I was considering was flashing a new (EP)ROM making further board mods to support it. The modifications don't seem too risky but on further problems I'll still be in a position where I don't know if the socket or the chip (or something else) is at fault.

This is all very new ground for me. Typically I've been able to revive old machines with little effort on the hardware front. I currently have this Spectrum and a BBC Master that are stretching my level of comprehension!