Can't get NTSC colour working on TDA4555

[133MHz_]

I've been trying to get this 1988 Philips 27CE4599 that you guys helped me restore to display NTSC composite video in colour by modifying the original PAL-only chrominance decoder card - seemed pretty straightforward but I haven't been able to get it to work and I've ran out of things to try.

The chroma decoder card comes with a TDA4510 but the board has the facilities to take the multi-standard TDA4555 as a drop-in replacement, so my plan was to install a TDA4555 and the missing NTSC crystal + switching circuit & hue control - in fact the application example on the Philips datasheet is pretty much exactly what I needed (automatic PAL/NTSC decoder using transistors to switch between colour crystals) so I pretty much copied the circuit verbatim:

Pins 25-28 can be driven high with >9V to force the chip to decode a particular colour standard so I've added a 3 position switch to be able to do that in addition to the automatic mode. I've built the circuit on the smallest piece of perfboard I could and placed it as close to the IC as possible with very short leads, but I can't get it to work correctly.

• In auto mode both NTSC and PAL display in b&w, with occasional and erratic color flashing on PAL, like it's trying to lock on but it can't.
• In forced PAL mode PAL displays 100% correct colour and NTSC displays in b&w.
• In forced NTSC mode NTSC is b&w and PAL is also in b&w.

Hue potentiometer has been set to provide 3V at pin 17 for a 0º phase shift as per the datasheet.

Things that I have tried without avail:

• Twiddling the trimmer cap for the NTSC crystal
• Replacing the trimmer cap with a fixed 15pF and a short
• Tried another 7.15909 crystal (got 20 of them online - they oscillate but I don't know their accuracy)
• Tried another TDA4555

At one point I removed my circuit entirely and installed the NTSC crystal directly on the card to rule out any wiring errors on my part - I managed to get pulsating but correct color that way in auto mode (at that point I wasn't aware of the forced standard mode so I didn't try it).

I'm stumped. What can I be doing wrong?

[Freya]

I am no expert on this topic, but are you using a reputable supplier for the IC, and is it a Philips version. The reason I ask is there are now so many fakes on the market, personal experience has delivered me two fake IC`s on separate occasions.

[133MHz_]

Sourced them locally and they do say Philips, although I've never gotten any fake ICs from the place I bought them from, in this day and age and with older parts you never know for sure - and they weren't cheap either. Damn.

I bought two of them because I also want to do the same modification to a Commodore-Philips monitor with the same overall setup. I made a 600 dpi scan of the chips in case it helps to determine if they're real or fake.

[Maarten]

They look real enough.

[133MHz_]

Well I got it working... in an odd way. Removed the crystal switching circuit, put in a single 2xNTSC crystal + trimmer cap and the hue control, forced NTSC decoding by pulling pin 26 high and set out to follow the chroma oscillator alignment procedure outlined in the Philips service manual: feed some colour bars, short pin 17 to ground, adjust the trimmer cap until the colour pattern on the screen is almost stationary then remove the short - I was able to get a slowly moving pattern without any flickering or banding, but as soon as I removed the pin 17 short... back to black & white, with maaaaybe a brief flash of a few frames of (correct) colour here and there. I had a closer look at the hue control pin on the TDA4555, and according to the datasheet it goes like this:

• 0~0.5V: Service position 1 - burst OFF; colour ON (for oscillator adjustment)
• 2V:NTSC Hue control -30º phase shift
• 3V:NTSC Hue control - 0º phase shift
• 4V:NTSC Hue control - +30º phase shift
• 6V: Service position 2 - Hue control OFF; colour ON (for forced colour ON)

The voltage divider network on the application example gives an effective output of roughly 2 to 4V for the range of the 10k potentiometer, which I had set to 3V for a neutral tint, but I got curious about what it would do outside those voltages so I removed the resistors and just threw the 10k pot in there as a crude voltage divider from 0 to 12V with the wiper going directly to pin 17, then started slowly raising the voltage while watching an NTSC encoded picture, everything went as usual until I passed the 5V mark, that got me perfect NTSC colour! No odd phase shifts either! If left disconnected pin 17 floats to about ~2.6V which gets me back to b&w.

So it seems that the only way to get NTSC colour to work is to force disable the hue control by pulling pin 17 to 6V, which to me is odd to say the least since a hue control only makes sense for NTSC signals - so no way to adjust tint at all, but I can live with that (IMO contemporary NTSC doesn't need a tint control anyway). Another odd thing is that automatic mode doesn't work in neither PAL nor NTSC, it has to be forced to the appropriate standard otherwise I just get very weak and erratic bursts of pink-ish garbage instead of colour. I just left a 3 position toggle switch hanging out the back while I figure out a good place to mount it. I'd rather avoid having to drill holes in the cabinet and stuff but I'd take a manual NTSC/PAL switch over no colour at all any day of the week.

Well now that I finally have colour on NTSC I've got to do something about the chrominance trap - the chroma dots are way too strong on NTSC especially in the old video games I like to play with their large areas of pure saturated colours. My chrominance decoder card has an additional TDA4565 Colour Transient Improvement IC so I'm not sure which filter(s) would I need to switch out or modify for 3.58/4.43MHz operation. I've uploaded a good quality schematic of the card itself and some photos (parts side) (solder side) of it prior to my modding in case someone can help me out find an easy way to add chroma trap switching to this. Also what does Colour Transient Improvement actually do? Never heard of it before owning this set. Is it any good or is it like Scan Velocity Modulation which is considered to be best when disabled?

[Sideband]

Originally Posted by 133MHz_

Also what does Colour Transient Improvement actually do? Never heard of it before owning this set. Is it any good or is it like Scan Velocity Modulation which is considered to be best when disabled?

How it works is a mystery only the chip manufacturer/designer could tell you. What it does is actually clean up any noise between the different colours. For example if you display standard CVBS colour bars, you usually get noise between say Green and Magenta...the so-called Transient Noise. CTI effectively removes the noise and makes a standard composite video input look more like RGB. A series of digital filters within the IC achieve this.

In short, sharper colour resolution on standard composite video input.

It works well on PAL but no idea of the effect on NTSC.

[dominicbeesley]

How does the chrominance signal get into pin 15? If there is a filter before the TDA4555 (which there would be) this may be filtering the chorminance signal down on NTSC so far that the chip is struggling to lock (or ignoring the colour due to burst level being too low).

[dtvmcdonald]

One simple suggestion to try: perhaps there is a noise problem.

Add a small bypass cap from the center of the pot to ground.

[133MHz_]

I've been looking at the chroma filtering stage feeding the TDA4565:

I believe the L5350/C2350 parallel LC filter is the one I need to de-tune down to 3.58MHz to get rid of them chroma dots on NTSC, my first thought is that adding parallel capacitance would do but what would be the simplest way to do this electronically using the signaling from the crystal switching circuitry? Or am I missing something else?

Originally Posted by Sideband

How it works is a mystery only the chip manufacturer/designer could tell you. <...> In short, sharper colour resolution on standard composite video input.

Interesting. I'm using RGB most of the time so I haven't really had a good look at the composite video processing, but it roughly sounds like yet another high frequency boosting technique that can lead to unnatural looking pictures if done excessively. I tend to dislike such things butI won't mind it if it's done in moderation.

Originally Posted by dominicbeesley

How does the chrominance signal get into pin 15? If there is a filter before the TDA4555 (which there would be) this may be filtering the chorminance signal down on NTSC so far that the chip is struggling to lock (or ignoring the colour due to burst level being too low).

Here's the luma filtering stage feeding the TDA4555:

Your theory makes sense to me but when the hue control pin is pulled to 6V (so that NTSC actually works at all) it doesn't automatically lock on PAL colour either. Maybe it's a side effect of the forced service condition?

Originally Posted by dtvmcdonald

One simple suggestion to try: perhaps there is a noise problem. Add a small bypass cap from the center of the pot to ground.

Good point, sometimes I tend to overlook the simple stuff. I'll add one when I take the set apart again to mess with the chroma filter and see if it changes anything.

[Barry Lloyd]

I see that your circuit puts 12V on pin 26 (NTSC) or 28 (PAL) to drive them ON, but pins 25, 27 and the other one of 26 & 28 - which are supposed to be OFF - are essentially left open circuit and I am wondering whether they should be pulled down to zero volts (or at least, the 0.5 V quoted in the data sheet). I know that this is true for TTL inputs but as I don't know what kind of circuitry is inside the TDA4555 I can't say whether this would apply in this case (though the current of -3mA in the data sheet for I25 etc. does seem to suggest this). Just a thought .. Barry

[dominicbeesley]

Without the values for the inductors it's not clear what the chroma filter is filtering but it looks like two traps for (possibly) sound carrier intermodulation or something. might be worth bypassing from chroma in to the junctions of 3311 and 3313 with a 1n cap and see what happens. Or if you have a scope and NTSC colo(u)r bars see if they look to be around 250mV p-p at the chroma input.