Jump to content
IGNORED

Investgating PAL 7800 picture issues (was 7800 RGB Musings)


juansolo

Recommended Posts

19 hours ago, juansolo said:

We've had @GDT2019's 7800 in to give it a service and swap out the comp mod for a AJM+UAV. There's been some interesting things to note:

 

- This one had lovely nice and easy to deal with solder, unlike the last few we've had.

- Despite the last 4 or so 7800's all being the same board revision, they all behave a little differently. In this case this one behaves like Rusty (my original 7800) when it comes to the Dragonfly cart. In that original carts seem ok in it, but some images on the DF crash. BallBlazer being the noteable one as the original cart runs just fine. Just noticed it was the same with Food Fight today. There's just something odd about this machine. Both the ones with these identical symptoms have a pair of bodge resistors underneath. They don't seem related, but mine (the one that works fine and plays most 2600 homebrew, but has a shitty 2600 picture) doesn't have them, and nor does Cleggy's super horrible solder one (the other one that works fine but doesn't like 2600 homebrew, though has a great 2600 picture).

- This was another one with a poorly formed cartridge guide that needed some love with the Dremmel to get the DF cart to sit nice in there. It was also uncomfortably tight with with some original carts.

 

Anyhow, a hard to track down fault aside (a fet connected to the TIA had a burned up pad on it that was causing the TIA output to be very low and therefore the 2600 picture was super shitty, solved with another bodge wire...), this one now has a revised version of the AJM+UAV mod on it. We've just simplified it, so I thought it warranted another post. Much easier to make as rather than the IC, it's now running a 5v non-latching relay for the chroma switching.

AJMv2-UAV.thumb.jpg.cbdbc35f3005e2ee8f113362f6b8cda5.jpg

The audio I've not put on the board to keep it as small as possible. It's super easy to add in as you see fit regardless. Otherwise, the usual work done here: re-cap, replace buttons, swap reg, replace power jack, tidy up and what have you.

 

GDT-7800-1.thumb.jpg.d7c96d676392aa52171e82166f550279.jpg

 

Pics of the screen, they actually look better in person. You still get the colour running on certain dark colours, that just seems inherent on the PAL 7800. It's still massively better than stock.

 

GDT-7800-2.thumb.jpg.5a4c67e78a080a72777d1d68a22628e1.jpg

 

GDT-7800-3.thumb.jpg.d135e3cf0b4c723225544b0f17493977.jpg

I just wanted to chip in... and for note here on the forumn.  Thank you so much for all the hard work yourself and Cleddy have put in on my console (and indeed the 2600 and 5200 you are working on).  I can't wait to see the results in person.  I really can not say enough about the whole process and how the pair of you have kept me in the pitcure (even when it's been stuff that has gone right over my head, I have still found it utterly fascinating knowing the process behnd getting the machine to where it is). 

 

Can not recommend you guys enough to anyone else considering having work done on their own machines.  Cheers!

  • Like 2
Link to comment
Share on other sites

21 hours ago, GDT2019 said:

Can not recommend you guys enough to anyone else considering having work done on their own machines.  Cheers!

Agreed, I sent them my NTSC 7800 for them to play with their mod. In the process they fixed a dodgy power connector and replaced some of the switches that were on their way out.

I'm saving up to have them fit UAV and their mod to another couple of 7800s I have 

  • Like 1
Link to comment
Share on other sites

I don't think I've posted this here so I thought I'd add a summary and I think we've now tracked down all the issues with the PAL 7800. Essentially it's a basket case with three main problems. The biggest by far is the shared clock. The NTSC machine doesn't have this problem as the NTSC MARIA sends the clock to the TIA. On the PAL machine however the single clock signal is split between both ICs and it causes massive amounts of interference. That we solved by disconnecting the MARIA clock and adding another just for that.

 

Problem two is the blended chroma signals causing all sorts of issues with the colour. This one effects both PAL and NTSC machines but it's not bad and mainly just results in either the 7800 side having great colour at the expense of the 2600 side or vice/versa. Our latest solution to that is the relay switching of the chroma lines from the MARIA and TIA depending on which mode the console is in which also grounds the one not in use. The downside of this fix is that keeping the RF modulator is no longer on the table. You're gonna be doing that regardless on a PAL machine, but on an NTSC machine you might want to keep it and just use a UAV on it's own, it does the job well.

 

The final problem @marauder666 finally tracked down the other day. There is an issue with colour running on most PAL 7800's on darker colours. It's most apparent with the blue mazes on Pac-Man Collection and the front screen of Commando. The degree to which this happens varies massively from console to console. Mine is not bad at all, it's mitigated somewhat by the Canon RGB encoder mine runs into which seems to handle it very well also, but it also does appear to almost go away after a while of being powered up. Sadly the problem appears to be inherent in the PAL MARIA and it's heat related. On Rusty (my other 7800) if you point a hair dryer at the MARIA for a few seconds the running slows then stops. Turn off the hairdryer and it'll be stable until it starts to cool, then it'll slowly start again. Keep the hairdryer on for a little longer and the running goes the other way. This is a bummer as given the variability of PAL 7800's, this can actually be pretty bad. Other than swapping the MARIA and hoping to get one that's more tolerant, there's nothing you can really do about it.

 

In a nutshell that's it, there is one other odd thing we've noticed. The UAV luma output when used on a PAL 7800 is a little bit high. This kinda makes sense as it is different NTSC vs PAL. Adding a 30R resistor on the luma out line when running S-Video seems to bring it down to a more sensible level. You get the same problem with the composite out, but adding the resistor will bring down the whole signal, it is however better than without, so I'd recommend a suck it and see approach to that one.

 

@MrZarniwoop's French RGB machine has got here and now we're pretty much done with @GDT2019's machines, we'll be looking into that today. Initial signs are not great as the picture is terrible. I'm gonna try splitting the clock first as it's simple to graft in and out, but I'm not expecting that to make a big enough difference. What it has on there is an RGB transcoder where the RF modulator used to be, now this is a much, much simpler device than the external Canon I use so I'm not expecting miracles. But we can likely do something with what's going into it at least. Hopefully it's chewing on separated luma and chroma and not composite video. Either way, if we can tap the signal before the transcoder and see what state it's in, that's going to be a start to see if we can get improvements or if the transcoder itself is just stepping all over the signal. We shall see, this one might take a while...

  • Like 2
Link to comment
Share on other sites

7 hours ago, juansolo said:

What it has on there is an RGB transcoder where the RF modulator used to be, now this is a much, much simpler device than the external Canon I use so I'm not expecting miracles. But we can likely do something with what's going into it at least. Hopefully it's chewing on separated luma and chroma and not composite video.

I believe the RGB transcoder is based on this PAL decoder, in case this helps: https://archive.org/details/SonyV7021/

Sony V7021 NTSC PAL Decoder specifications 1991.08.pdf

Link to comment
Share on other sites

Ok, after going down a bit of a rabithole trying to track down the weird intermittent incompatibility issues with a couple of 7800's we cracked open the French RGB one. First thing of note, given the sheer amount of bodge resistors and wires, this wasn't anything super developed! The next thing is that the picture is BAAAAD.

 

Before diving too deep in we thought we'd see if anything simple would bring benefits. Adding the clock (which oddly beforehand now goes through a voltage divider bodge to bring the signal down from the internal clock, not sure why), anyhow the new one gets rid of the diagonal interference. But honestly that's the least of the picture's issues. Spent a while trying to balance the on-board chroma trimmer and the two on the RGB board and got a better balanced pic, but it was still awful. Next stage was just a mini service just incase anything was not behaving. Higher rated reg, cap swap (the vast majority were ok, a couple were out of spec by a reasonable amount. One on the RGB board and one on the main board). That didn't do a great deal but we want to remove any random variables and it's a quick and simple job to do.

 

So with the RGB board out we get a look at what's going on... The RGB board is fed with 9v, 5v, GND, Luma & Chroma. This is a best case scenario for us because we know just how bad a stock 7800 is so the signal going to the RGB board is likely rubbish. So with crap in / crap out currently playing a significant part in the bad picture our next course of action is to graft in our S-Video mod and get that all dialed in bypassing the RGB circuit. If we can get that looking good, and I don't see a reason why we shouldn't, we can then pipe it into the RGB board and see where we're at. That's our job for when I'm next over on Wednesday.

 

Anyhow, some pics of the machine before we got going. The caps aren't leaking that's glue (and it's been cleaned off now).

P1050847.jpeg

P1050848.jpeg

P1050849.jpeg

P1050857.jpeg

P1050858.jpeg

P1050859.jpeg

P1050861.jpeg

P1050863.jpeg

Edited by juansolo
  • Like 3
Link to comment
Share on other sites

7 hours ago, juansolo said:

our next course of action is to graft in our S-Video mod and get that all dialed in bypassing the RGB circuit. If we can get that looking good, and I don't see a reason why we shouldn't, we can then pipe it into the RGB board and see where we're at. That's our job for when I'm next over on Wednesday.

Episode 1 of this miniseries for me was watching the USPS tracking status of the shipment from Chicago to the UK. Way more drama in early 2021 compared to the before times.

 

Episode 2 for me was this initial diagnosis. Chroma! And luma! (Not just composite!) And charming narrative and shockingly good lighting for retro console board pics.

 

I can't wait for Episode 3.

Link to comment
Share on other sites

16 hours ago, ApolloBoy said:

Didn't see it mentioned here but @SainT is working on an RGB board for the 7800: 

 

Interesting as that's kinda/sorta doing what the French machine does, but taking a different route. Curious as to where he's getting that composite signal from as the MARIA and TIA output analogue chroma. I'm super looking forward to what he comes up with and have offered to be a guineapig if he needs someone out there for testing.

 

Cleggy's ( @marauder666 ) been busy. I suspect he got bored waiting for me to get over. Blame Covid, I can only really get over there twice a week and it's where we do all the work.

 

P1050869.thumb.jpeg.e689c63c5da0b16d8f5d28bb812d9d98.jpeg

 

He's made and grafted in our S-Video board and with the results we hoped for. Basically it's as good as any we've done in terms of S-Video out. Haven't tweaked the 2600 side as it's pretty close (it's got a little too much chroma), and the trimmer is under the memory board which makes adjusting it a bit of a problem. There's a bloody tiny hole in it, but it's really too small to get anything useful through!

 

P1050872.thumb.jpeg.db40879158a0ad7418a254c6d26ff841.jpeg

 

Anyhow, picture is just fine, so next to pipe that into the RGB board and see how it deals with it...

Edited by juansolo
  • Like 3
Link to comment
Share on other sites

1 hour ago, juansolo said:

Haven't tweaked the 2600 side as it's pretty close (it's got a little too much chroma), and the trimmer is under the memory board which makes adjusting it a bit of a problem. There's a bloody tiny hole in it, but it's really too small to get anything useful through!

Amazing work so far! It'll be great to see 2600 mode improvement when you get to it, as Kaboom! is likely to a primary use case for this particular European 7800. :) 

  • Like 2
Link to comment
Share on other sites

Ok for reference Cleggy and I run 7800's into a Canon RGB-100 RGB encoder. We're essentially doing what the French machine does, I even think it uses the same Sony chip, however there's about a vast amount more circuitry going on and active filtering, which the on-board one doesn't do. My 7800 is running a UAV-AJM and Cleggy's is running the All-In-One AJM the same as above. The results we get are really surprisingly good. The UAV-AJM solution is a little better out of the two options.


So with that in mind, Cleggy has continued his testing...

 

P1050880.thumb.jpeg.f912747964ce1c6fb36af06b3c777ad6.jpeg

 

The chroma signal was way to high so there's just a pot grafted in there to bring it down. It's nowhere near as good as the Canon, but then we didn't expect it to be. But it's a lot better than it was! The sad fact is that it doesn't look as good as the S-Video out, but again, that was kinda expected sadly.

 

P1050878.thumb.jpeg.729da7af6b2fdcb01680deb51f0090f7.jpeg

 

Where we go from here depends on Mr Z. I'll be sending you a PM momentarily...

Edited by juansolo
  • Like 2
Link to comment
Share on other sites

Long story short, the RGB encoder on the French 7800 is merde.

 

We can get a great 7800 picture out of this machine, and quite frankly one of the best 2600 pictures from it via our S-Video mod. It really looks amazingly good. Then you fire it through the encoder and it mutilates it. We've ended up adding trimmers so we can manipulate just about all the levels coming out of our mod to give the encoder something it can deal with. The 7800 pic is a great deal better than it was in fact. However it's not a patch on the S-Video output that it's chewing on and we just don't think it can handle the 2600's picture bizarreness at all. Given just how clean and solid what's going in is, it's all coming down to the encoder's signal handling, and it's rubbish.

 

We've tried. We've spent most of today trying to get something I'd consider tolerable out of the 2600 side and it just isn't. The 7800 side, if you've only got an RGB input, is usable, and not half bad. But it's worse on a panel. We can get a better composite video signal out of the console than the RGB one it's outputting, and S-Video annihilates it.

 

Also the audio as stock was unbalanced. I don't think that a lot of time was spent at Atari on this mod...

 

The way forward from here is up to Mr Z. As I see it he has 3 options:

 

1) Have it as it is now and just accept that the 2600 side is pants and the 7800 side is usable.

2) Revert it to stock and then at least from a collector perspective it's original.

3) Pull the RGB board and run it as an S-Video or Composite Video machine. But that means having something that can handle PAL to plug it into.

 

Personally if he has something that can chew on the signal, I'd go with option 3.

 

I've attached some pics, just bear in mind the 2600 pic is way worse than it looks in the photo...

 

P1050891.thumb.jpeg.427af9b8605b9e6c5584376243a231d0.jpeg

 

P1050884.thumb.jpeg.882b2d3364b9c9f58226acfa6b8b5e1b.jpeg

 

P1050885.thumb.jpeg.21e3f6346e861d32049019946bda6ae2.jpeg

 

P1050887.thumb.jpeg.33a97599dfb78e95080cb8ea73e4b7e8.jpeg

 

P1050888.thumb.jpeg.af11cabf5241f3e3904ec0c58dcee76e.jpeg

 

P1050883.thumb.jpeg.8a8b74041696ce2c21b9edf4f48a2262.jpeg

  • Like 3
Link to comment
Share on other sites

8 hours ago, juansolo said:

we haven't finished with the RGB one yet. Still working on it and there's been some improvements. Will update soon...

I know I’m swimming against the tide here on AA, but I did ask @juansolo and @marauder666 to keep the SCART RGB rather than ditch it in favor of S-Video.

 

This was because I ended up valuing the French 7800’s DIN13 (Atari-standard!) SCART RGB connection. I know it’s not better than their S-Video upgrade, as I now have proof beyond doubt from them that S-Video can look much cleaner than feeding video through the stock Atari RGB converter. But, I intend to use a RetroTINK 2X-SCART to HDMI upscale the converted RGB rather than change that part of what is a fairly unique piece of hardware. Other 7800 consoles are somewhat common, but the French PAL Atari 7800 with SCART RGB is in a class by itself!

 

For that reason, I’m still hoping they can still clean up the SCART RGB a bit, and help it with the best video possible, realizing it won’t ever match their PAL S-Video upgrade work.

 

Hey, but maybe that’s an excuse to buy another (non-French) PAL 7800 in the future? ?

  • Like 1
Link to comment
Share on other sites

1 hour ago, -^CrossBow^- said:

You do have a 3rd NTSC 7800 ready to come home anytime you ask don't forget that?

I believe at this point I have enough NTSC 7800s to last a lifetime. ☺️

 

I need to get a few of them into some friends’ hands to share in the retro joy.

  • Like 1
Link to comment
Share on other sites

We've thrown in the towel. We spent another day and ended up leaving in the mods that make the most difference and backing out the ones we couldn't make work right. In the scheme of things, it's not the worst picture we've seen... But it's nowhere near as good as we'd like. The RGB encoder is horrible. We can feed it with one of the best signals we've ever had out of a 7800, and it just mangles it, no matter what we do.

 

To the bodges, the 3 resistor bodges on the TIA luma were actually in the wrong place! We were putting them back today and it was just baffling us looking at the photo we took of where they had been installed because it made no sense whatsoever. We found another pic of a different French machine and they were in the place we'd expect them to be... So yeah that was just wrong. Fixed that. Also the audio from the TIA was unattenuated but the cart audio was, causing a mis-match. Fixed that.

 

What we've ended up doing is backing out our nice luma DAC we use for the S-Video mod, we just couldn't get it to balance right with the RGB board. We thought we'd had a breakthrough, which was what we were trying today, but that turned out to just make matters worse sadly. In the end we just decided to feed it with the luma signal it was expecting by reverting that side of the circuit back to 'stock'. We've left the 2nd clock for the Maria however, which eliminates the rolling diagonal interferance and the chroma switching that's allowed us to seperate the two and adjust them individually.

 

Other than that, swapped the power jack for a 2.1mm DC jobbie, and I think we can call this one done.

 

P1050915.thumb.jpeg.a1f0b84950c7dc6036ba897c807a943e.jpeg

Edited by juansolo
  • Like 3
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...