TMS9918 works but no color?

[Tom Plano]

Hi all,

I recently built up a TMS9918 pcb and I thought it was working great. I only had a black and white display so I didnt even notice the problem until I picked up a commodore 1902 last weekend at VCF midwest. Now that I have a color display, I realized that I still only have a black and white image coming out of the card. Its very odd to me that H/V sync should work fine, but no color would be decided.The monitor works fine, its been tested with a c64(comp and luma/chroma) and a Nabo(comp) and does show color fine in all cases. Ive also checked the signal out of the card and it seems that there is a color burst being generated (see included image of an intermediary field). I'm a little at a loss on what the look at next to debug this issue. Thoughts would be greatly appreciated. 

-Tom P

[S P Dixon]

What's the crystal you're using - the exact part?

I've been building these boards for a while and have experienced the b/w display. I didn't notice the issue at first because modules with the issue do display in colour with some displays or when going via a converter like a retrotink. In the situations where I saw the mono display, the coloured areas appear as vertical black and white strips - is that your symptom?

I tracked this down to the crystal. 

Boards with the exact part that J B Langston specifies (LFXTAL029962REEL) + 16pf load caps work great. I thought that a crystal of the correct frequency (with appropriate caps according to its data sheet) would be fine. That's not always the case, it seems pretty fussy and I've had to do a bit of trial and error with different parts.

Beware of crystals of the right frequency that give the word 'serial' for load capacitance. I've had no luck with those (I think they're designed for a different type of circuit). I've tried others, including one that was almost an identical match for all specifications, just a slightly higher frequency, but no luck with that one (the TMS9918A data sheet does say "10.7 + MHz" in one place, but it also says 10.738635 (± 0.005) MHz  in another. 

I've finally found a part that is available and correctly display colour on all fo the displays and converters I've tried: MP107-E. it's the correct frequency and the other specifications are a reasonable match for the one that JBL specifies. It needs 32pf load caps rather than 16 though.  (And it's taller, which might risk touching the back of the module in the next slot, but I think it's advisable ot use a heatsink on the TMS chip which rules out using the next slot anyway.)

Shiela

--
You received this message because you are subscribed to the Google Groups "RC2014-Z80" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rc2014-z80+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/rc2014-z80/b57ef147-f1dc-4c74-b5a0-c2b7f3996397n%40googlegroups.com.
<SDS00001.png>

[Mark T]

Also remember the TMS9918A is NTSC composite output and not PAL. Most recent TVs in the UK will work with either PAL or NTSC composite, but I’m not sure if a monitor will be so flexible.

TMS9928A or TMS9929A would also give a monochrome display, as they do not output a composite signal.

[Tom Plano]

Shiela I don't see the black bars but its definitely only black and white, color and tint controls have no effect.  The crystal is ECS-107.3-S-4X on digikey, so I suppose its possible it's off. It specs min10pf to typ20pf load caps, so I just used the 16pf I had. 

Mark, Im in the US so everything I own is NTSC. However, its good to know there is a similar monochrome chip out there. I got my chip from UTsource, which Ive had good luck with in the past, but fake labeling is always possible. Maybe Ill swap with the chip out of the Nabu as they share the tms9918. should reveal if mines a fake.

-Tom

[S P Dixon]

That particular part is one that I tried and also found that it produced mono output. (on Mouser's product page for that part, load capacitance is given as 'series'.)

I recommend MP107-E with 32pF caps.  It's readily available from the big sites and did the trick for me.

To view this discussion on the web, visit https://groups.google.com/d/msgid/rc2014-z80/32161c92-bef0-47eb-88fa-6823af020e75n%40googlegroups.com.

[Tom Plano]

Awesome, thanks for the tip!!

[Mark T]

I just saw a note in the comments on hackaday that someone built a 74ls04 oscillator to use a serial cut crystal, so that might be an option.

https://hackaday.io/project/159057-game-boards-for-rc2014

I was also wondering if it might be possible to run a colour burst crystal at third overtone using an external oscillator.

[Samuel Falvo II]

On Thursday, September 14, 2023 at 10:11:32 AM UTC-7 Mark T wrote:

TMS9928A or TMS9929A would also give a monochrome display, as they do not output a composite signal.

To be more clear about this: these chips (9928/9929) output split luma and chroma (think S-Video, similar to how the C64's VIC-II generates video); they absolutely do have color capability.

 

[J.B. Langston]

As Shiela said, the frequency of the crystal must be correct. The NTSC colorburst frequency (315/88 = 3.57954MHz) is derived from the crystal and if it is not close enough to the expected frequency then the TV won't be able to detect it and will fall back to black and white.  You can try using a frequency counter or oscilloscope to check the frequency of the output clock signal (should be 3.57954MHz on pin 38).  Don't try to test the input clock frequency because the probe will load down the crystal and alter it's frequency.