Investgating PAL 7800 picture issues (was 7800 RGB Musings)

[juansolo]

6 hours ago, MrZarniwoop said:

I realize I’m near at the end of this French SCART saga, but I wanted to note that all this has been worth it for @juansolo’s posts.

Yeah it's back together now. Just a bit of a rattle test to make sure it's all still solid then I think Cleggy will be in touch regarding sending it off.

[juansolo]

It mildly annoyed me that my own 7800 had a pretty bad 2600 picture, so given all we've learned recently I decided to re-mod it with the new relay board going to the UAV. Mainly because I can, but also because that machine has been so hacked around with (it and Rusty were used to develop things...) it's a random mess of ideas inside rather than having a coherent mod on it. I didn't take pics because it ain't pretty that one.

 

We're now convinced that the luma out on the UAV when used on a PAL 7800 is too high. We really do need to measure it at some point to find out for sure, but it was resulting in some fairly epic bloom and even smearing on my big PVM. Putting a 30R resistor between the luma out of the UAV and the S-Video jack tames it to about where it should be and things look way better. Also re-doing everything with the latest mod has cleaned up my 2600 picture. It's still not the best, but it's massively better than it was. I'm actually quite pleased with it, which I never have been to this point. Maybe the excess luma was screwing with that too...?

 

Mildly updated the mod also. When I finally get around to updating the all-in-one board with the relay switching I'll do a new revision of the doc, but for now, this is the lastest AJM-UAV board.

 

[juansolo]

2 hours ago, juansolo said:

We're now convinced that the luma out on the UAV when used on a PAL 7800 is too high. We really do need to measure it at some point to find out for sure, but it was resulting in some fairly epic bloom and even smearing on my big PVM. Putting a 30R resistor between the luma out of the UAV and the S-Video jack tames it to about where it should be and things look way better.

I actually mean a 33R, ...that 3R makes all the difference...

[juansolo]

I'm fairly sure this is the last version of of the All-In-One AJM. My poor 7800's have once again been guineapigs for buggering around with this and it will result in a version 2 of the doc when I get around to updating it. Very long story short, 7800 is immensely variable and the best solution to getting a great picture out of it varies just as much. From a *LOT* of experimentage we've decided the following:

 

- If you're running an NTSC machine the UAV mod is the easiest mod out there with the best results for the effort required. You can give it a cleaner signal with the chroma switching from the AJM, but that means removing parts from the 7800 and losing RF out.

- If you're running a French RGB machine, you're pretty much boned. The RGB encoder massacres any signal you put into it. Any improvement you can get by feeding the encoder a better signal is improved upon vastly by binning it and just runing the AJM-AIO as composite video.

- If you're running a PAL machine the AJM-UAV or AJM-AIO will give you the best picture currently available. But it's still subject to the 7800's inherent variability, but is a vast improvement on anything else out there.

 

The v2 of the mod we've gone to relay switching to simplify it and refined it some more. We've found some interesting things out of late that have gained us some improvements and some others that have led to acceptance of the 7800's flaws (the MARIA's colour rolling thing). Again, I'll put it all in the doc. As it is, this is as compressed as I can make the vero for the AJM-AIO, it's not nice to make as it's very tight and even has a link underneath (the green wire) to keep it as small as is possible. I think we might have reached the point where we turn this into a PCB... We shall see on that as the mod is kinda tough because analogue, so isn't something really I think many people will take on. Indeed having a scope and a proper PVM/BVM helps a great deal in setup. A Dragonfly/Concerto also helps for test patterns. Otherwise you're going on experience, which we can do now and be pretty accurate, but anyone else I feel would be stabbing a bit in the dark. You can't get away with thinking the trimmers are in the right place because they won't be...

 

Anyhow, here is Rusty, my original resurrected 7800 now running the AJM-AIOv2 as composite video.

 

 

We've taken the audio and the colourburst off board to make the video board as small as possible. For audio now we pull the two audio caps (C54 & C55), replace them with 8K2 resistors. We now pull the two resistors above C55 and connect the right hand two pads together (they're the other end of the two resistors) and send the output through a 1uf cap.

 

The colourburst we've just put on it's own little board for now. I'll find a nicer place to pull power for that if we do another mod...

 

 

Below is the current vero. With how the audio and colourburst is wired as seperate boards for reference.

 

[ckrtech]

On 4/15/2021 at 1:24 AM, juansolo said:

- If you're running an NTSC machine the UAV mod is the easiest mod out there with the best results for the effort required. You can give it a cleaner signal with the chroma switching from the AJM, but that means removing parts from the 7800 and losing RF out.

 

I had an extra UAV sitting around that I put in an NTSC Atari 7800 the other day. The 7800 games that I have tested so far look pretty good, but of course I tried the 2600 side with Pitfall and had jailbars everywhere.

The RF modulator was removed prior to the install, and now I am considering pulling/lifting R16 and R17 to separate them from the leftover combination circuit and using the chroma switcher with the CD4053 from the AJM 2.2 PDF.

Were there any other tweaks you thought about trying to clean up the signal (I saw the photo of the LCD with the jailbars a few pages back)?

I am also curious if anyone else has done this mod on an NTSC 7800.

[juansolo]

5 hours ago, ckrtech said:

I had an extra UAV sitting around that I put in an NTSC Atari 7800 the other day. The 7800 games that I have tested so far look pretty good, but of course I tried the 2600 side with Pitfall and had jailbars everywhere.

The RF modulator was removed prior to the install, and now I am considering pulling/lifting R16 and R17 to separate them from the leftover combination circuit and using the chroma switcher with the CD4053 from the AJM 2.2 PDF.

Were there any other tweaks you thought about trying to clean up the signal (I saw the photo of the LCD with the jailbars a few pages back)?

I am also curious if anyone else has done this mod on an NTSC 7800.

The jailbars can just be a thing. Depends on the 7800 and the display how bad the show up. Interestingly the one NTSC 7800 we've done also had fairly prominent jailbars on the 2600 side even with the chroma switching. We didn't really have it long enough to go deeper and see if it's a trait of that machine (I still really need to buy one...), but the switcher will make things better by virtue of separating the two (you'll need to re-balance the two colour pots if you do that). But the jump isn't massive. The NTSC machine on the whole works nicely enough with the UAV on it's own to be fair and had you not already pulled the modulator, I'd have been tempted to stay there. With it gone though, there's no reason not to put switching in there if you're happy with an iron.

[ckrtech]

6 hours ago, juansolo said:

The jailbars can just be a thing. Depends on the 7800 and the display how bad the show up. Interestingly the one NTSC 7800 we've done also had fairly prominent jailbars on the 2600 side even with the chroma switching. We didn't really have it long enough to go deeper and see if it's a trait of that machine (I still really need to buy one...), but the switcher will make things better by virtue of separating the two (you'll need to re-balance the two colour pots if you do that). But the jump isn't massive. The NTSC machine on the whole works nicely enough with the UAV on it's own to be fair and had you not already pulled the modulator, I'd have been tempted to stay there. With it gone though, there's no reason not to put switching in there if you're happy with an iron.

I would have definitely left the RF modulator in there, but it had stopped functioning properly. I created a breadboard-based transistor mod just to verify the console was working and then elected to put the UAV mod in for something a bit more in-spec.

 

I certainly don't mind building the switching board for the chroma circuit and even poking the chroma lines with my oscilloscope just a little bit. Unfortunately, I do not have a flash cart or a test cart at the moment.

[+-^CrossBow^-]

12 hours ago, ckrtech said:

I had an extra UAV sitting around that I put in an NTSC Atari 7800 the other day. The 7800 games that I have tested so far look pretty good, but of course I tried the 2600 side with Pitfall and had jailbars everywhere.

This is something I've seen on a few 7800s I've done. Where do you have everything routed and where do you have the UAV located? Also the type of wire etc?

 

Something else I've NEVER been able to quite figure out that happens with my personal 7800 sometimes. I will fire it up and I have jail bars?! Both on the 7800 and the 2600 side of things. But then I could power it on the next day and they aren't there. I've not been able to figure out if the issue is something off in my 7800 or perhaps something goofy with the OSSC and Extron combo that is used that might trigger it? But they aren't present on CRTs and 80% of the time, they aren't displayed.

 

Also, try plugging in composite at the same time to see if anything changes. If you already have composite plugged in, then try unplugging it.

 

[ckrtech]

28 minutes ago, -^CrossBow^- said:

This is something I've seen on a few 7800s I've done. Where do you have everything routed and where do you have the UAV located? Also the type of wire etc?

 

Something else I've NEVER been able to quite figure out that happens with my personal 7800 sometimes. I will fire it up and I have jail bars?! Both on the 7800 and the 2600 side of things. But then I could power it on the next day and they aren't there. I've not been able to figure out if the issue is something off in my 7800 or perhaps something goofy with the OSSC and Extron combo that is used that might trigger it? But they aren't present on CRTs and 80% of the time, they aren't displayed.

 

Also, try plugging in composite at the same time to see if anything changes. If you already have composite plugged in, then try unplugging it.

 

22 awg stranded. UAV is where RF modulator used to be. Grounding wire is a bit long as it runs to the center-right rail (i will most likely change that...) All other wires are short. I do plan on replacing the 7805 - in fact, I have some spares around here somewhere.

 

I only composite + composite ground hooked up to the output pads at the moment. No s-video.

 

The jailbars are definitely present on the 2600 side.

 

I see the benefit of switching between the two chroma circuits, but I also wonder if the chroma<->luma relationship is a bit out of spec even with the UAV in place. Specifically that the chroma amplitude is too high relative to the luma's signal.

[+-^CrossBow^-]

Well when I install the UAV in that place, I just attach ground with a larger gauge wire right off one of the holes that anchored the RF modulator in place since they are right there. I would be curious if you see the same with s-video should you connect it. What I also do, although it shouldn't make any difference but it seems to. I tie the two ground pins off my s-video connector together and then run a single ground from there to the ground next to the chroma output. The composite and audio grounds I tie together and run that back to the other ground on the opposite side near the artifact trimmer off the UAV. I don't use any of the actual main board grounds for any of my outputs on the 7800 except for occasionally during testing.

 

[juansolo]

An interesting test for you. Disconnect the chroma line on the S-Video out and see if the jailbars are coming from the chroma signal or the luma. If they're still there, you can try pulling the MARIA chroma in also to the UAV and see if that makes a difference. Just out of curiosity trying to find where the jailbars are being introduced.

 

As I say though I'm not 100% adding the chroma switching is worth the effort on the NTSC machine as you can get really decent results on it just with the UAV.

Edited April 17, 2021 by juansolo

[ckrtech]

juansolo, I don't want to hijack your thread as I am diverging from your thread title as this is an NTSC system. I can always split this off and start a new thread if you prefer.

 

I did move the ground wire as @-^CrossBow^- suggested. It is super short and connects to one of the holes where the RF modulator was. No change in the jailbars on the 2600 side.

 

I disconnected the MARIA col input from the UAV and lifted resistor R17 to disconnect the TIA color from the chroma tying circuit. No change.

Of course, pressing the "Pause" button to disable color results in an image without jailbars.

 

I put the wire back, put R17 back in place and scoped the UAV's composite output (yellow) as well as the stock 7800 composite feed that normally goes to the RF modulator (cyan) - with the probe clipped on the RF modulator side of C8. I managed to single-shot a line of video with the trunks of the trees in Pitfall (I took a photo of the CRT image and lined the trees up in the image below for kicks).

 

 

 

What is interesting about the two signals (both lack 75 ohm termination as I disconnected the AV cable from the CRT so they would be the same) is that the UAV output of the 7800's 2600 signal (yellow) is kinda "lazy." Compare it to the stock 7800 composite that is fed to the RF modulator (cyan signal), and you can see how much sharper the pulses are - how they ramp up/down versus the UAV output.

The amplitude's peak-to-peak level from the UAV (yellow) between the trees shouldn't vary like that. It should be even peak-to-peak - swinging around the desired luma level all the way across just like the stock 7800 signal (cyan). It should be the same, solid color, the same luma level, the same Hue, and the same saturation.

 

 

[juansolo]

8 hours ago, ckrtech said:

juansolo, I don't want to hijack your thread as I am diverging from your thread title as this is an NTSC system. I can always split this off and start a new thread if you prefer.

This is very interesting for me and I think this has just become more of a '7800's have bizarre video' thread. I'm totally cool with it. Hell it started out at me just bemoaning the lack of an RGB mod, and just went from there. I'm loving the directions it's taking

[juansolo]

On 4/17/2021 at 9:43 AM, ckrtech said:

What is interesting about the two signals (both lack 75 ohm termination as I disconnected the AV cable from the CRT so they would be the same) is that the UAV output of the 7800's 2600 signal (yellow) is kinda "lazy." Compare it to the stock 7800 composite that is fed to the RF modulator (cyan signal), and you can see how much sharper the pulses are - how they ramp up/down versus the UAV output.

The amplitude's peak-to-peak level from the UAV (yellow) between the trees shouldn't vary like that. It should be even peak-to-peak - swinging around the desired luma level all the way across just like the stock 7800 signal (cyan). It should be the same, solid color, the same luma level, the same Hue, and the same saturation.

We have noticed that the UAV when used with the PAL 7800 isn't a perfect solution also. Which to be fair to it, it's meant to work on all Atari's so it's ability to handle the two-in-one 7800 in itself is nice. But it's luma output on the PAL 7800 is a little high, causing blown out white on PVM/BVM (panels hide it better). I'd once again be really interested to get my hands on maybe another NTSC 7800 and fit our mod to it just to see how that does. In some respects the UAV does a better job to be fair, but in others ours does. With the NTSC machine, because the UAV works well with it (and is massively easier), it's what we've recommended... With the PAL machine the UAV on it's own just doesn't because it has way more issues than the UAV adresses.

 

Also, so I don't have to make another post, we've refined our AIO proto board again finally removing the luma trimmer. After modding our 3 machines (again...) recently with the new mod and found this is now settable, so we've set it The vero is still not pretty but it's small now at least. For composite out you can now just slam the luma and chroma outs together, which is nice.

 

This should be right. Ours are a bit more bodged than this... I'm sure I'll make one and verify it at some point.

Edited April 20, 2021 by juansolo

[ckrtech]

39 minutes ago, juansolo said:

For composite out you can now just slam the luma and chroma outs together, which is nice.

Is that typical for combining chroma and luma for PAL? No other components needed?

 

As far as the UAV, I feel that it is the current go-to board for the 7800, but yeah - it is far from perfect. I tried messing around with the TIA signal I fed the UAV (using a few other passive components) just for a bit. Nothing of note really changed. I was also unable to duplicate the scope signal I shared earlier, but I may have also had MARIA chroma disconnected from the UAV this most recent time. I moved TIA chroma from the tcol input to the main color input just to see if a circuit change would make a difference, but it didn't.

Great work on your latest AJM revision, btw. I hope you get a chance to mess around with an NTSC console as well.

[juansolo]

28 minutes ago, ckrtech said:

Is that typical for combining chroma and luma for PAL? No other components needed?

Needs the cap we have on the chroma, but otherwise, it goes together well and the picture can be shockingly good. Rusty (my original 7800) is running composite out this way and my main 7800 is S-Video running the AJM to UAV with a 33R resistor on the luma out line to bring it down a bit. @marauder666's 7800 is running S-Video with the above AIO AJM. So we've got one of each essentially.

Edited April 20, 2021 by juansolo

[CatPix]

On 4/3/2021 at 8:48 PM, juansolo said:

Also the audio from the TIA was unattenuated but the cart audio was, causing a mis-match. Fixed that.

Hello!

I own a French 7800 here and I would like to fix that audio issue, that is apparently a common issue with PAL 7800 as well?

The issue is : apparently, nobody cares. All the very rare mods I found for PAL 7800 just never mention that issue or just glance it aside as "Only a few games have POKEY so it's not important".

Word I heard around is that the R5 resistor is a 6.8K ohm and  should be replaced by a 18 k ohm one.

Does that applies to French models too, and, where is R5? Or, what is the right solution?

 

 

 

FYI, pictures taken on a LCD; despite my best efforts my camera refuse to take a picture of a CRT without bleaching the colors and/or focusing.

Pictures are jsut there for your potential curiosity; I am satisfied enough with it - and I don't have enough 7800 games sadly.

 

 

Thanks in advance.

 

 

 

 

[+-^CrossBow^-]

2 hours ago, CatPix said:

Hello!

I own a French 7800 here and I would like to fix that audio issue, that is apparently a common issue with PAL 7800 as well?

The issue is : apparently, nobody cares. All the very rare mods I found for PAL 7800 just never mention that issue or just glance it aside as "Only a few games have POKEY so it's not important".

Word I heard around is that the R5 resistor is a 6.8K ohm and  should be replaced by a 18 k ohm one.

Does that applies to French models too, and, where is R5? Or, what is the right solution?

 

Thanks in advance.

 

 

 

 

I'm not sure the specifics on the PAL systems. R5 and R6 are the resistors for the audio from pokey and TIA on the NTSC version of the 7800. And we don't mess or change those values. But I have heard that the PAL values are the same between both audio signals and so the balancing is off on the PAL units and always has been.

[juansolo]

Details to follow... I've got a memory like a sieve unless I document stuff and I forgot to do that for the French audio issue (it was definitely wrong!) thinking it was likely a one-off given the one we did had the resistor bodges on the TIA luma incorrectly applied. I've messeged @marauder666 He's more likely to remember what we did.

 

It will likely be more pronounced with Rikki and Vikki as I've noticed that it's audio is a little low (can be adjusted on cart to a degree). Best to use Commando to check as that uses both Pokey and TIA audio at the same time. Easy to tell when it's not correctly balanced with that.

Edited April 23, 2021 by juansolo

[juansolo]

 

This is how the one that we modded was wired before we did anything. Given some errors on it, double check yours is the same, it might not be. If it isn't post a pic and we'll have a look.

 

To simplify things I'm calling the signal path CA (cart audio) and TA (TIA audio).

 

The cart audio comes in on the right pad of capacitor of CA1, out on the left then into the right pad of the resistor CA2, out the left.

The TIA audio comes in on the left pad of capacitor TA1, out the right and into the right pad of the resistor TA2, out on the left.

Blue wire is the audio take off for the RGB encoder how it was on this machine. So it's taking the TIA audio before the resistor and the Cart audio has to travel through two resistors...

 

To sort this easily unsolder the blue wire, lift the left hand side of CA2 and TA2, twist them together and solder the blue wire to that. Then the signal from both sides travels through it's own cap and resistor before coming together and being output to the RGB encoder.

 

Hope that helps!

Edited April 23, 2021 by juansolo

[CatPix]

It helped greatly! I did just the change you mentionned, and the audio from cart is noticeably stronger. Also, the TIA is also slightly more loud (usually I set up the volume on my TV around 20/25, and here, 15/20 was satisfying enough).

I only own Rikki&Vikky for now as "audio-improved carts", but I had the cart audio crancked up to the max; From the tests I did, I still have a slight imbalance, but if R&V is still "low" even with the loud+ internal volume pot  to the max, carts with POKEY will certainly have louder POKEY audio. Either way it maks it much more enjoyable :

This is a video I make when I received my copy. And, yes, that is a video with the cart set up to the maximum sound output possible from the cart. I don't think having a POKEY chip would have been much more loud.

I may do a quick video to show the difference and share your solution for the benefits of Péritel 7800

Edited April 23, 2021 by CatPix

[juansolo]

You can mess with the values of those two resistors to better balance it if you want to experiment. Though I'd still get your hands on Commando to do that as it makes it really easy. R&V has always had slightly low volume in comparison for me.

[TailChao]

On 4/23/2021 at 7:15 AM, juansolo said:

Blue wire is the audio take off for the RGB encoder how it was on this machine. So it's taking the TIA audio before the resistor and the Cart audio has to travel through two resistors...

Ah, thanks for this! I thought the electronics' botch would be something similar from the symptoms, but it's excellent to get a confirmation (and solution) without having to import a Péritel 7800.

 

 

Also...

On 4/23/2021 at 2:31 AM, juansolo said:

It will likely be more pronounced with Rikki and Vikki as I've noticed that it's audio is a little low (can be adjusted on cart to a degree). Best to use Commando to check as that uses both Pokey and TIA audio at the same time. Easy to tell when it's not correctly balanced with that.

I've brought this up in previous A/V discussions, but I recommend adding a small audio amplifier to your design if this is going to be mass produced. This'll net you more consistent output and help protect the console + cartridge.

[juansolo]

3 hours ago, TailChao said:

I've brought this up in previous A/V discussions, but I recommend adding a small audio amplifier to your design if this is going to be mass produced. This'll net you more consistent output and help protect the console + cartridge.

Not likely to be mass produced, there's no real demand for it and it's a lot more intensive than something like the UAV. That said, we'll probably do a PCB for it now that I'm fairly comfortable it's as good as it's gonna get. Not sure it'll handle the audio, if it does it'll likely retain the simple solution. At least for now.

Edited April 27, 2021 by juansolo

[juansolo]

Noticed an error when it came to the last vero layout as I was making the schematic a little more legible for turning into a PCB. I'll post that shortly also if people are interested.