Spectrum lost colour on composite but gained stripes on YUV

[willinliv]

Hi,

I've got a Spectrum 48K which has got a black and white composite picture, and a stripey picture via YUV out (RGBTOHDMI converter).

Both were working fine previously but I am worried I might have zapped something through poor edge connector hygiene. The RGBTOHDMI is supplied YUV video from an edge connector cable/adapter (which works fine with another machine), and then samples the picture to create a HDMI signal to an LCD screen. It's difficult to make out but it looks like it also has lost some colours, especially from odd/even lines, white seems to be there, but I think the TV screen is showing something different. Any ideas where to start?


Many thanks as always, Will

[willinliv]

Pin out on RGBTOHDMI to edge connector is:
Pin1 (GND) 14B
Pin 2 16B
Pin3 18B
Pin4 16B
Pin 5 17B
Pin 6 NC

Also testing with a Retroleum test cart reports IC25/IC26 as being suspect

Traced between the 3 edge connector contacts to the 3 relevant pins on the ULA ok, no shorts between them, also between same ULA pins to capacitors . I have an oscillosope so will take a look at the output of those pins. Also swapped the voltage regulator for a switch-mode OKI one. Testing now via the composite and TV and still getting b/w no colour. Bit fearfull I have fried the ULA, but I'm still wondering about that upper RAM error.

[mikeh_nz]

Can you compare

• hue (U, edge connector 18B) and
• saturation (V, edge connector 17b)

signals with that coming from the machine that works?

I think you implied you have a working machine - what sort is it?

What issue of motherboard are the two machines?

After checking that the ula chip version is compatible in the other working machine, you could take the ula and put it in the working machine and see if that machine now stops displaying color (as far as color is concerned)

[willinliv]

Thanks Mike!

Yes I do have a second machine that is fully operational - issue 3B with ULA 6C0013E-7 (the one that's faulty is issue 3).

I will compare those signals between both machines as suggested.

And then onto the ULA - just need to doublecheck compatability. I have ordered a few replacementIC including an LM1889N but whilst I imagine this has bearing on the composite (mod), if there is an issue via the raw yuv on the edge connector this will also effect I guess the LM1889N rather than it being the cause.

Thanks again for your suggestions I would not have thought of switching the possibly fault ULA to the good machine

[mikeh_nz]

Cool

In case you haven’t found this site - there’s some ULA compatibility details here

I believe its safer practice to put a suspected broken chip into a known working machine, than to put a known working chip into a busted machine with unknown faults.

However in this case, I’d personally just put your known good 6C001E-7 into the faulty machine and see if that fixes the problem. Your busted machine runs most of the diagnostics ok (just complaining about upper ram addressing chips. There’s a separate IC25/IC26 test - does it fail on that too? Retroleum DiagROM details)

Good idea about the LM1889 - I could believe it’s that too.

[willinliv]

Thanks again Mike for your help - a bit of an update and some success!

Transplanting later ULA into faulty machine fixed the horizontal bars seen over the yuv edge output (into a RGBTOHDMI device), so assume faulty ULA. But had no effect on composite out suddenly becoming black and white (losing colour).

Got out my scope (this might have been before the above) and the signals were pretty similar to my working unit, some of these measurements are a bit sketchy, but I thought they looked comparable.
Broken
Y = 15.58kHz, 2.56Vp-p
V = 18.87kHz, 1.60V
U = 16.78kHz, 980mVV

Working
Y = 15.53kHz, 2.56V
V = ?, 1.72V
U = 15kHz, 1V

I tried out further RAM tests with my test cart with Retroleum test suite, and even though on first check IC25.IC26 always reported as suspect, in the further tests they passes, also all upper and lower ram compound passed.

Next up replaced LM1889 with another - no change.

Then I did some internet searches, maybe I was spelling colour wrong but I never saw Retroleum's advice about a black & white image, but I had already changed LM1889, so was thinking about the XTAL X2 crystal, but the signal seemed as expected going into LM1889. Then I came across this post https://www.sinclairzxworld.com/viewtop ... 7&start=10 and I was thinking about TR2, and was about to change it out but did some continuity tests and found that it was very poor/non-existant between base of TR2 and C65, so a little via where it joined the capacitor solder point underneath must have been slightly damaged suring a recap.

A bit more solder to the joint to tin the connection and all good, back to colour on the composite out. Also the IC25/26 reported error has gone so must be connected. Tried out the original ULA to see if this too miraculously fixed but sadly no (so still with stripes), but at least now have a working unit again!

So thanks again for your help in getting me thinking through the issues! Will

Notes to self: Check continuity in the most likeliest fault locations following the schematic before suspecting component failure; try to limit system changes one at a time with testing beforehand and inbetween; be extra careful about that edge connector

[1024MAK]

Problems with “very poor/non-existant between base of TR2 and C65” have been found before…

It may be that this part of the PCB is more vulnerable to being damaged.

Mark