Issue 2 colours not right

[jfharden]

I got an issue2 48k ZX Spectrum a few weeks ago, initial tests suggested it needed a new voltage regulator, so initially I recapped it, replaced the voltage regulator with a MuRata Switch Mode 5v Regulator. I took the opportunity to also recap it completely, do the DC-DC mod, and a composite video mod (using the capacitor method), including replacing TR1 and TR2.

After all that it boots "ok", except the colour was yellow, so I proceeded to try and tune VR1 and VR2, I used the multimeter method and tuned the IC14 pin4 to pin3 voltage to +50mv and the pin2 to pin3 difference to be -50mv. I also used a frequency counter to get TC2 to be exactly 4.433619MHz using the method described in the service manual. During this it was clear VR1 and VR2 were not in good condition, so I also replaced those.

The overall output was still yellow (I tried more manual tuning of VR1 and VR2 using the oscilloscope method but nothing gets an ok picture.

I tried measuring the voltages of the 2 colour difference output pins on the ULA (pins 15 and 16) using the method where you set the paper, border, colour to be the same so the whole screen is the same colour. This showed the colour difference voltage on pin 16 to hardly vary at all (pin 16 only varied between 1.71and 1.75 volts for all colours):

I then replaced IC14 (with another LC1889) and the ULA (to another of the exact same type) to no improvement.

The video in voltage on IC14 was only 11.2 volts, so I replaced TR4 and TR5 which brought it up to 11.5v (the 12v rail for the ram is 12.02v) and possibly improved the video output slightly, I feel I've got closer to a decent output since I replaced these, but it might be in my head.

If the majority of colours look good the background is yellow instead of gray, if I get a gray background the colours get really flickery and sometimes phase into being the wrong colour.

I recorded a very small video which shows the closest I've been able to get https://photos.app.goo.gl/ZKEQwGbdPaUJ1bqg6

This is the very very best I've managed to get the copyright screen to look:

Both the video and image above are from the composite mod output through a Retrotink 2X-Pro

This is a photo of the board as it is now:

If anyone has any suggestions of what I might be able to do to improve it I'd love to hear them?

There's a lot more photos, including more closeups of the board in this google photos album, all in much higher resolution: https://photos.app.goo.gl/JM9MA1yd2jBCE9SSA

Thanks,

Jonathan

[1024MAK]

Hello and welcome Jonathan

Obviously there is something else wrong. I presume you only changed the electrolytic capacitors?

You’re already done most of what I would suggest…

BTW 11.5V on the supply pins for IC14 (LM1889) is normal.

Out of curiosity, do you have another Spectrum? If yes, do the ULA and the LM1889 from this sick machine work okay in the other Spectrum?

Have you checked to make sure that there is no problem with the ULA socket pin 16?

Did you check continuity between the leg of any replaced components and the various PCB tracks that (are supposed to) connect to said pin/terminal/leg. I mention this, because the issue two board does have PCB tracks and through hole plating that can be damaged.

With the ULA removed, power up, now what voltages do you get on the ULA socket pins 15 and 16? Does removing IC14 make any difference?

Power down. Now test the resistance between ULA socket pin 15 and the +5V, and +12VA (the +12V to IC14) supplies, then test it to 0V/GND. Repeat for pin 16.

You may also want to measure R38, R39, R56 and R63. Also test the operation (resistance) of the preset resistors VR1 and VR2.

Mark

[jfharden]

Thanks so much for your response.

Sadly I don't have another spectrum to swap chips around with.

I went through all the things you suggested, all the replaced components connect to everything it should according to the issue 2 schematic.

Pin 16 on the ULA socket correct connects to the board and goes to the expected places.

Voltages are as follows:

With the ULA removed and IC14 in place:
Pin 15 0.52v
Pin 16 0.50v

Pin the ULA removed and IC15 removed:
Pin 15 0.54v
Pin 16 0.52v

Resistances are as follows:
ULA Pin 15 to +5v: 1.506 kΩ
ULA Pin 16 to +5v: 1.506 kΩ

ULA Pin 15 to +12VA: 1.887 kΩ
ULA Pin 16 to +12VA: 1.888 kΩ

ULA Pin 15 to 0v: 214 Ω
ULA Pin 16 to 0v: 214 Ω

The resistors I measured in-place in the board and got:
R38: 1.42 kΩ
R39: 1.46 kΩ
R56: 216 Ω
R63: 214 Ω
VR1 has a range from 1.427 kΩ to 2.1 kΩ
VR2 has a range from 1.427 kΩ to 2.1 kΩ

I notice C51 isn't populated, but from a quick google that seems normal on the issue 2 (and explains why I couldn't find it on the schematic! spent way too long looking before I googled!)

I have only replaced the electrolytic caps, apart from c48 which I replaced because on my thermal cam it was getting a lot hotter than everything else nearby (other than the ULA). I haven't seen these wire wound axial caps before, I replaced C48 with a ceramic cap of the same value so hopefully that's ok (it made exactly no difference to the picture too, and the old cap measured fine in my component tester after it was out of the board. I had initially tested it using an ESR tester while in the board, which said open circuit, it was only when replacing it I realised the ESR tester can't test low enough values to test a 47nf cap! turns out my component tester can though)).

I'm really at a loss at this point. I notice someone has drawn a red N on the CPU, no idea if that was from factory, or when a company upgraded this from a 16k to 48k model (I'm convinced this happened, it had a warranty sticker on from a repair shop, 2 of the lower ram chips are in sockets, and all the upper ram is in sockets), I also don't know what it meant, it could be a Z I suppose to looking at it a different way round ). The whole range of RAM tests fine using the retroleum DiagROM though.

[1024MAK]

C48 is only a decoupling capacitor for the +12V rail, so it’s value is not critical. Oh, and yes, axle leaded ceramic capacitors or poly type have been available for, well, a long, long time. And you can still get them (well, some types/values).

The ESR value is only really relevant to electrolytic capacitors.

I wouldn’t worry about the marking on the Z80A, it appears to be working. And it can’t cause the yellow tinge that you are having problems with.

I’ll have to have a think about your results.

Mark

[1024MAK]

For me, the big unknown is how the Retrotink 2X-Pro handles the non-standard video output of the Spectrum. Whereabouts in the world are you? If possible, it would be good if you were able to try using a TV that has a composite video input.

The above does not mean that I don’t think there is fault with the Spectrum board.

Mark

[jfharden]

Good question about the RetroTink. I'm in the UK and my main TV has a real component in, and my Denon AV receiver also has component in. I can at least try it in both of those and see if the results are any different.

[jfharden]

O...M...G

So I retuned the IC14 voltage differences on pin 4 to pin 3, and pin 2 to pin3 to +50mv and -50mv.

Then tried it on my TV (which is a Sony Bravia android TV from 2015 which has a real composite in) and....

You have perfectly hit the nail on the head!

The RetroTINK is clearly not understanding the video out...and it's clearly been working all this time!

This means I bought a second ULA, and a second LC1889, and replaced TR4 and TR5, probably all for nothing!

I'll try firmware updating the RetroTINK (there have been multiple firmwares since I first got it) and see if that makes a difference, but I highly doubt it.

I just need to figure out what I'm going to do to get this working on the monitor on my desk since thats where I can actually use the Spectrum. I don't want to take up the user port since I'll be using it with a DivMMC Future.

Thank you so much for all your suggestions and help, and sorry for leading us on a merry meander down a crazy road!

[jfharden]

I updated the RetroTINK 2X Pro to the latest firmware (1.7) and it's still the same. Clearly this just isn't going to work.

Again, thanks for all the help.

I just realised, add to the list of stuff I didn't need....I didn't need to buy and build a frequency counter kit! At least that can be added to my testers and tools

It's been quite a journey but I learned an absolute ton about the spectrum which is great! I also started getting used to the keyboard typing that basic colour test program over and over, while it's a strange beast I can see how it could make writing BASIC programs quick if you were proficient!

[1024MAK]

Glad I could help with ending the wild goose chase

The so called single keyword entry system is a love it/hate it thing. I found it very useful on a rubber key back in the eighties, but in contrast, was also using a Acorn BBC B keyboard at school.

It worked better on the ZX80/ZX81 with less keywords. On the Speccy, even now, for some of the less often keywords, I still may spend ¼ of a minute looking for the right combination of keys to press…

The 128K editor did away with the single keyword entry system, which would make it easier for those who know the QWERTY keyboard layout, but unlike the BBC Micros and Electrons, you can’t abbreviate

Back to the video, with the 16K/48K/+ machines, without using the expansion port, your options are limited to composite video or S-Video (it’s possible to modify a board to output S-Video).

Converting the colour difference signals to RGB has been done as well, there are articles on the web about it.

The 128K machines have a RGB output available on their monitor ports, but apart from the toastrack, they don’t have the Sinclair look and feel.

The SPECTRA interface produces a RGB output on a SCART socket and has a full expansion port on it’s rear. Details here. I have one and it’s really good. But not cheap.

With any of these options, some upconverters/frame buffers may still have problems, as the underlying non-standard video signal is still there with its non-interlaced fields and close but not quite right timing. But with the RGB outputs at least there is no longer the problem of the colour carrier not being locked to the rest of the video signal…

If you do use a upconverters/frame buffer or TV/monitor that does has a SCART input, you may need to feed a 12V DC supply in to perform the function switching needed for devices that have automatic input selection switching.

Mark

[jfharden]

Another mystery solved.....I think that mark on the z80 is to denote there's something wrong with it.....

....my ZX-HD and DivMMC future arrived. The ZX-HD works and is AMAZING! I spent a while trying to work out why the DivMMC wouldn't load ESXDOS...and I'm sure this won't be a surprise to anyone knowledgeable but the M1 signal is non-existant from my Z80. Even tested on the CPU pin it's just noise in the 50-100mv range.

So I've ordered a replacement Z80A and at the weekend I'll remove the current chip and add a socket. Annoyingly it's an issue 2 with the tr6 transistor sat on top of the z80 so I'll have to work out where to put it....maybe I'll just put it back on top of the new CPU