A3000D - NTSC Screeen Issue

[Castellen]

Any ideas why resistance would be less than 10k - I measure its less than half.

Are you measuring this in-circuit?  In which case you're measuring a complex impedance network, or in other words, you're measuring across other components in parallel with that resistor.  Therefore you'd need to unsolder one end of the component to test it correctly.

It's unusual for a resistor to become low in value, so that wouldn't be my first guess.  But it's just a pullup resistor, the exact value in that location will make little difference to the circuit.

Bypassing R484 and/or R454 removes the fuzziness. I'll try replacing these two resistors and see what happens.

I didn't see if you mentioned if the video was OK on the 23 pin RGB port or not, my skim read may have been too hasty.  But if the 15kHz video is OK, and the 31kHz video on the VGA port has the issues as described, that sounds like an issue with the 28MHz video clock.  That is created by a phase locked loop (PLL, U481).  This circuit is fairly sensitive, and it can be affected by corrosion damage from a previously leaked RTC battery.  You'd usually remove components that are visibly corroded, clean corrosion from the PCB surface, and replace the components with new ones.

If there's not been previous corrosion damage, then the solution might involve a simple adjustment of the PLL clock phase.  With the video artefacts visible, turn VR470 (beside the flicker fixer enable/disable switch) 1-2 turns in one direction to see it if improves.  If not, turn it back to the original position, then 1-2 turns in the other direction.

Failing that, the PLL centre frequency (28.375MHz) may need adjustment, which involves making the PLL run open loop (J482 on pins 2-3), measuring either the 28.3755MHz output (J483 pin 1) or the output /2 (14.18775MHz at J483 pin 2) while adjusting VC470 for a free-running centre frequency as close as possible.  Obviously you'll need a frequency counter for this, or a scope with a frequency counter.  Then replace jumper J482 pins 1-2 to run closed loop.

A quick Google search just found this well written PLL adjustment guide:
https://www.tsb.space/knowledge-base/a3000-flicker-fixer-repair-and-adjustment/

[Castellen]

Pushing down on VC470 seems to have sorted it. Switched off. Fitted my Picasso II and the pass through is working.

That suggests that the trimmer capacitor VC470 may be intermittent.  That would fit the symptoms, since it would affect the PLL centre frequency, which will impact the performance of the scandoubler.

Personally I'd replace VC470 to prevent the problem resurfacing, but that will require recalibrating the PLL centre frequency as described above.

What sort of scope would do the job? These can be pricey and Ive looked at obtaining one previously but the price has always put me off.

You can do a lot with a simple monochrome 60MHz dual channel digital storage scope, which are easily available on the second hand market and not expensive.  Since I rely on my test equipment for a living, I stick to professional brands such as Tektronix, HP, Agilent, Fluke, etc.  I've seen plenty of people have good success with the many Chinese brands out there, though your mileage may vary, and the user interface can be hit and miss.  Keep in mind that you'll then need to learn how to use it, though there are probably countless internet videos on the topic, if you're into that kind of thing.

Another suggestion would be to buy a simple used frequency counter.  Something that counts up to 100MHz would be fine.  You can sometimes pick them up for $20 since they're considered old-tech and no-one wants them anymore.  And in this case it won't matter if it's out of calibration.

[Castellen]

I'll replace VC470 and also look at replacing the other compoents U481 and the OKI buffers as per the link in your other post. Second hand scopes maybe not as expensive as I thought.

I'd recommend only replacing the parts likely to be defective, just VC470 at the moment.  U481 is clearly working OK, and it's unlikely to be intermittent.  If the de-interlacer is working correctly, there's no need to replace the Oki field memories.

To help with testing, I've copied the Commodore test images for testing/calibrating the A3000 scan doubler and de-interlacer to my webserver:
http://amiga.serveftp.net/temp/A3000_video_adjust.lha

That HP 54600B scope is a good instrument at that price, it's ideal for what you'd need.  My only concern is they don't guarantee it being functional, it states "Item powers up unable to test as we do not have the knowledge to".  WTF, that seller deals in a huge range of advanced test equipment, yet they don't know how to conduct a basic functional test of a simple oscilloscope?!  That would normally ring alarm bells, though that appears to be the standard disclaimer they put on every Ebay listing.  I'd check first if they will refund/exchange if it turns out to be defective.  Or ask for a photo of the display showing a sine wave or whatever on both channels, they've got dozens of other listings for signal generators which can do exactly that.

[Castellen]

The low prices look suspiciously low vs prices from other sellers which are at least double. I'll may be seer clear. I had a l ook at some low cost Chinese ones on Amzon.

I suspect the HP 54600B seller is genuine given their extensive inventory and sales history.  It would help a lot if they put slightly more effort into the sales detail.  They've gone to the trouble of powering it on and taking photos, it only takes seconds more to connect a signal generator to show it's fundamentally working.

You could potentially get a good deal with that, but definitely get something in writing that they will refund/exchange if it's not working, or ask for a photo of the display with some kind of waveform on both channels; even the built-in square wave probe calibration signal is fine.


But the Chinese scope looks OK for home/hobbyist use as well, and you'll probably get a set of probes with it, which I doubt they'll include with the HP 54600B.

[Castellen]

Forgot to put the jumper (J482) back!. Once I did that - all looks to be ok.

The clock waveform doesn't look great, are you sure you have the probe ground connected correctly?  The easiest place is usually on the top lead of the flicker fixer enable switch, which happens to be a convenient ground connection point.

Also make sure your probe is set to x10 (for 10M Ohms impedance).  With J482 set to the open loop position, you should be able to get the PLL free-running centre frequency very close to to what it should be; doesn't need to be exact.  The range of the white trimmer capacitor is likely to be close enough if that table is correct.  Otherwise you can put the original trimmer back in to see how that performs by comparison.

No idea where you were seeing ~10MHz, you'd need to describe how and where things are connected, and what's displayed on the scope.

[Castellen]

Does anyone know where I can source schematics for the Picasso II?

Could the crystal be bad? or some other fault?

I've never seen schematics available for the Picasso II.

Doesn't look like an issue with the crystal, it's running at the correct frequency.  You'll also get different voltages depending if you're measuring on the input or output side of the crystal oscillator, i.e. one side of the crystal will have a larger AC waveform than the other.  You should be using the scope probe in x10 mode to avoid loading the oscillator too much.

Picasso II repairs are usually a lot of guesswork.  The last few of these I've repaired have been faulty GD5428, which you normally need to find on old PC graphics cards, look on Ebay.  But that's not to say that's the problem or not in this case.

You could exchange some parts such as the DRAMs between the working and non-working boards to narrow down potential problems.  I've no idea exactly what the differences are between the rev 1.4 and 1.6 boards.

[Castellen]

1. Replace the crystal
2. Replace the caps - there's not many of them - and my desoldering/soldering is good enough to remove / replace.
3. I can swap out socketed chips. Was nervous about doing so in case there is a fault on the board that damages them and then I end up with two broken PIIs
4. Replace the relays
5. Replace resistors
6. Replace the GD5428 (think I have one in my box of PC boards) - replacing this will push my soldering skills (I'm not very good with drag soldering)

1. I don't expect it's an issue with the crystal since it's oscillating at the correct frequency, the voltage levels on the GD5428 xtal-in pin only need to be meeting the high/low detection threshold of whatever the GD5428 needs, the datasheet will have that detail.  But if you want to test the crystal theory, you'd be better to exchange the crystal from the working board.  You won't know the characteristics of the existing crystal (i.e. load capacitance, series or parallel load), and if you get these wrong in whatever you select as the replacement crystal, you'll have new problems such as the oscillator not starting reliably, or at all.

2. The capacitors appear to be just general AC decoupling on the DC supply, I don't see how they'd affect the issue.  i.e. I expect you'll find that the board would run without the decoupling capacitors.

3. Exchanging socketed ICs is quick and easy with low risk if you're careful about removing/inserting them in sockets.  If you can't reliably remove or fit the DIP devices without bending the pins and putting in them in backwards, then this is probably best avoided.

4. You're getting some degree of video signal, so unless it's missing a sync signal from the VGA connector, then I doubt that's causing the issue.  They look like fairly generic double pole relays, you can soon test them in circuit.  When the coil is energised, both pairs of normally open contacts should be closed.  They might also have normally closed contacts which connects to the 'common' pin when the coil isn't energised.  Find a datasheet from a similar relay and you'll soon understand the pinout.  Besides, the contacts typically start going intermittent before they fail completely, and I'm picking the video issue is permanent.

5. It's even more unlikely you'll have an open circuit resistor unless something is physically damaged.  Either way, you should be able to sanity check them in-circuit using an Ohmmeter.

6. I've no idea what drag soldering is, presumably something involving wearing clothes of the opposite gender.  Not into that myself.  The correct way is to wick off the old solder from the QFP pads, clean with isopropyl, apply no-clean flux, run a single line of solder paste over the pads, position the QFP device accurately on the pads, then reflow with hot air.  Only takes a few minutes and you can even wear your normal clothes.  If you have a spare GD5428 on hand, this is the first thing I'd try.  You might need to find someone to do the soldering for you.