Harmony Encore issues with >= 64K ROMs

[+Karl G]

FYI Penult is now 128K DFSC. 

[+ZeroPage Homebrew]

24 minutes ago, Karl G said:

FYI Penult is now 128K DFSC. 

Ah, thanks! Sorry about that, I was copying my own info from earlier in the thread, probably from back when it was 64K. I just checked the 2020 Games List for 2600 and it's listed correctly there.

 

- James

[+groundtrooper]

@ZeroPage Homebrew

 

OK when I responded I must have just had a mini stroke. LOL

 

I do have a I light sixer it's my childhood one. Not modded. 

 

I have a Vader - AV Modded. - Not sure if that would reproduce the same issues. 

 

I have a 7800 AV - S-Video Modded - Same as above. Not sure if this helps. 

 

I must have been combining these into some sort of Super Console that I thought would be the same as your request. Sorry. 

I will still drop the games on my Encore and have a go on the AV Modded ones and see what I come up with. Sorry to shoot my mouth of so quick. I wasn't thinking straight. 

[+ZeroPage Homebrew]

2 hours ago, groundtrooper said:

I must have been combining these into some sort of Super Console that I thought would be the same as your request. Sorry. 

I will still drop the games on my Encore and have a go on the AV Modded ones and see what I come up with. Sorry to shoot my mouth of so quick. I wasn't thinking straight. 

No problem! It'll still be good to get more test cases, probably the most interesting one would be the 7800 S-Video modded one.

 

- James

[+batari]

A console that exploits this incompatibility is on its way to me so I can sort this out. I will let you all know how it goes.

[+McCallister]

Don't have a modded light sixer, but did just try it out on a modded wood grain 4 switch, all black, and a heavy sixer.  All worked as they should, with no hiccups as far as I could tell.  Zippy, Robot Zed, and Penult all seemed to function without going dark.

[+ZeroPage Homebrew]

33 minutes ago, McCallister said:

Don't have a modded light sixer, but did just try it out on a modded wood grain 4 switch, all black, and a heavy sixer.  All worked as they should, with no hiccups as far as I could tell.  Zippy, Robot Zed, and Penult all seemed to function without going dark.

Thanks so much for trying the games out on your consoles. What kind of mods do you have on your systems: composite, svideo, component or RGB?

 

- James

[+McCallister]

8 hours ago, ZeroPage Homebrew said:

Thanks so much for trying the games out on your consoles. What kind of mods do you have on your systems: composite, svideo, component or RGB?

 

- James

Knew I was forgeting something important...svideo/composite on all three. 

[+ZeroPage Homebrew]

 I have another EF+ game to add to the incompatibility list!

 

On 8/2/2020 at 1:18 PM, ZeroPage Homebrew said:

1) Penult (DFSC* 128K) <- Updated from EFSC* 64K

2) Robot Zed (EF* 64K)

3) Dystopia (EFSC* 64K)

4) Abyss of Chernobyl (EFSC* 64K)

5) Zippy the Porcupine (EF* 64K)

6) 2600 Words (EF* 64K)   NOTE: Manifests itself with randomly changing colours and flashing the answer onscreen.

 

On 8/3/2020 at 5:48 PM, batari said:

A console that exploits this incompatibility is on its way to me so I can sort this out. I will let you all know how it goes.

That's awesome Fred, thank you so much for looking into this! ?

 

- James

[+Karl G]

I want to help nail down what is going on with EF ROMs on Encores if I can. The effects end up looking like memory corruption to me, so I have made an EF ROM that tries to detect memory corruption. What this ROM does is display all 128 bytes of Atari RAM as colors on the screen, and switches to a new bank every frame, all without using any RAM or stack space itself (so its own operation should not be affected by memory corruption).

 

Here's what the display looks like if RAM is not cleared out at the start (each byte containing a random value):

 

 

However, it does clear all bytes of RAM to 0, so the display actually starts out as blank:

 

 

If no memory corruption occurs, then the display should stay blank.  If any colored rectangles appear on the screen, then that is evidence of memory corruption.

 

@ZeroPage Homebrew - Could I ask you run the ROM below on your Harmony Encore for a while to see what appears on the screen, if anything? Thanks for all of your help with this so far!

 

efram.bin

[+ZeroPage Homebrew]

3 hours ago, Karl G said:

I want to help nail down what is going on with EF ROMs on Encores if I can. The effects end up looking like memory corruption to me, so I have made an EF ROM that tries to detect memory corruption. What this ROM does is display all 128 bytes of Atari RAM as colors on the screen, and switches to a new bank every frame, all without using any RAM or stack space itself (so its own operation should not be affected by memory corruption).

This makes a LOT of sense if it turns out to be the issue! This is definitely a great visual way to test it out too.

 

Quote

 

@ZeroPage Homebrew - Could I ask you run the ROM below on your Harmony Encore for a while to see what appears on the screen, if anything? Thanks for all of your help with this so far!

Running it right now from a cold console and recording the results. I'll update this thread afterward with the video. Thank you so much Karl for trying to find a lead on the cause of this strange behaviour! ?

 

- James

 

============

 

UPDATE:  @Karl G: I ran (and recorded) the binary running for over two hours beside me while I worked. I didn't see anything but a black screen the whole time unfortunately so I won't be uploading a 3GB file of black screen. I *DO* think you're on the right track and it just needs to be narrowed down to the thing that invokes the video errors and program crashes. Is there anything else that special about EF that isn't used in any other 2600 bankswitching technique?

 

- James

 

 

[+Karl G]

3 hours ago, ZeroPage Homebrew said:

I ran (and recorded) the binary running for over two hours beside me while I worked. I didn't see anything but a black screen the whole time unfortunately

Thanks for trying! It was worth a shot at least. 

 

3 hours ago, ZeroPage Homebrew said:

Is there anything else that special about EF that isn't used in any other 2600 bankswitching technique?

There really isn't. It's the logical extrapolation of the standard F8/F6/F4 schemes. Nothing about why this one is an issue makes any sense to me, though I'm admittedly not a hardware person. 

[+ZeroPage Homebrew]

4 hours ago, Karl G said:

There really isn't. It's the logical extrapolation of the standard F8/F6/F4 schemes. Nothing about why this one is an issue makes any sense to me, though I'm admittedly not a hardware person. 

Maybe there's a way pare down what what is happening during the 2600 Words letter selection process to help point in the right direction to what might be inducing the screen issues?

 

- James

[+batari]

On 3/9/2022 at 2:46 PM, ZeroPage Homebrew said:

That's awesome Fred, thank you so much for looking into this! ?

 

- James

The console in question turned out to have a damaged cart connector that was making poor contact and preventing full insertion of the Harmony cart. After fixing that, the console/cart worked perfectly together.

 

So if others are having problems, it is sounding like 64k+ games are more sensitive to bad connections, so I would make sure everything is clean and intact.

 

Unfortunately, this means that I no longer have a problem console, as I found the problem with the console I had, but this doesn't mean there isn't something else going on here.

 

So, those who are having problems, try removing the SD card after the game loads. If this fixes the problems for some, that could provide some insight into the issue.

[+Karl G]

Would the removal of the SD card fixing the problem mean that potentially this is a power-related issue?

 

Also, I'm not sure if it's all 64K+ ROMs that are an issue. @ZeroPage Homebrew - I don't remember if you continued to have issues with Penult on the Encore after I switched to 128K DFSC? Do you do your playthroughs on the Encore or on Uno/PlusCart?

[+Karl G]

Also - I've modified my memory corruption detection tool to be much more irritating than the previous version. This version writes to a bunch of TIA registers every frame (including audio - so you will probably want to test with the sound off!), and then flips the bits in every byte of RAM every 16 frames. That way memory is accessed/modified, but if there's any corruption, that will still be preserved, and part of the screen will still be a different color. I don't know if I'm barking up the wrong tree with this, but maybe it's worth another try?

 

efram-irritating.bin

[+batari]

3 hours ago, Karl G said:

Would the removal of the SD card fixing the problem mean that potentially this is a power-related issue?

 

Also, I'm not sure if it's all 64K+ ROMs that are an issue. @ZeroPage Homebrew - I don't remember if you continued to have issues with Penult on the Encore after I switched to 128K DFSC? Do you do your playthroughs on the Encore or on Uno/PlusCart?

On Concerto when some reported issues with the menu, the problem was that some SD cards were not releasing the bus as expected. Removal if the card could help determine if the same issue is happening here, or if it may be something else.

[+ZeroPage Homebrew]

2 hours ago, Karl G said:

Also - I've modified my memory corruption detection tool to be much more irritating than the previous version. This version writes to a bunch of TIA registers every frame (including audio - so you will probably want to test with the sound off!), and then flips the bits in every byte of RAM every 16 frames. That way memory is accessed/modified, but if there's any corruption, that will still be preserved, and part of the screen will still be a different color. I don't know if I'm barking up the wrong tree with this, but maybe it's worth another try?

 

efram-irritating.bin 64 kB · 2 downloads

 

WILD RIDE COMING... HOLD ON TO YOUR HATS....

 

Wow, this new program did it! The RAM corruption started occurring within 5 seconds and slowly devolved into 2600 madness! I almost expected the AtariVox to start saying "My mind is going. There is no question about it. I can feel it. I'm...afraid."

 

After a while vertical lines started appearing on the screen and it briefly lost sync for a frame or two at points. I also noticed both the lines and sync issues on EF games while I play them on the Encore and also when they crash. Check out the video below...

 

WARNING, LOUD AND FLASHING!

 

 

I then went for lunch and turned off the 2600. After I came back I cleaned the cartridge, which had a very small amount of black come off of on the Qtips. I inserted the cart in and out to clean the console end and kept cleaning the cart until there was no more black coming off of the pins.

Video Timestamps:

[00:00] I turned on the 2600 from cold and ran the EF tester. It ran PERFECTLY for 20 mins without corruption!

[21:29] I ran through all the EF games. Sadly they all had the normal issues of corruption, crashing and resetting, except for Zippy, which strangely worked perfect.
[29:40] I ran the EF tester again and the corruption was back within 30 secs.

[37:00] I tried the EF RAM Testing program in the UnoCart, just to see. As expected, there were no issues as I've never have issues running EF games on the UnoCart.

[42:32] Here comes the really crazy part. I then ran the EF RAM Testing program on the Encore again and the screen was IMMEDIATELY filled with increasing amounts of corruption, and shortly thereafter, vertical lines!! Even the sound was starting to get a little weird here and there. WTH?? It's as if the RAM was already full of random information!! Stepping through your binary I can see that the first thing your program does is zero out the RAM. SO STRANGE!!

 

WARNING, LOUD AND FLASHING!

 

- James

 

 

[+batari]

16 minutes ago, ZeroPage Homebrew said:

 

WILD RIDE COMING... HOLD ON TO YOUR HATS....

 

Wow, this new program did it! The RAM corruption started occurring within 5 seconds and slowly devolved into 2600 madness! I almost expected the AtariVox to start saying "My mind is going. There is no question about it. I can feel it. I'm...afraid."

 

After a while vertical lines started appearing on the screen and it briefly lost sync for a frame or two at points. I also noticed both the lines and sync issues on EF games while I play them on the Encore and also when they crash. Check out the video below...

 

WARNING, LOUD AND FLASHING!

 

 

I then went for lunch and turned off the 2600. After I came back I cleaned the cartridge, which had a very small amount of black come off of on the Qtips. I inserted the cart in and out to clean the console end and kept cleaning the cart until there was no more black coming off of the pins.

Video Timestamps:

[00:00] I turned on the 2600 from cold and ran the EF tester. It ran PERFECTLY for 20 mins without corruption!

[21:29] I ran through all the EF games. Sadly they all had the normal issues of corruption, crashing and resetting, except for Zippy, which strangely worked perfect.
[29:40] I ran the EF tester again and the corruption was back within 30 secs.

[37:00] I tried the EF RAM Testing program in the UnoCart, just to see. As expected, there were no issues as I've never have issues running EF games on the UnoCart.

[42:32] Here comes the really crazy part. I then ran the EF RAM Testing program on the Encore again and the screen was IMMEDIATELY filled with increasing amounts of corruption, and shortly thereafter, vertical lines!! Even the sound was starting to get a little weird here and there. WTH?? It's as if the RAM was already full of random information!! Stepping through your binary I can see that the first thing your program does is zero out the RAM. SO STRANGE!!

 

WARNING, LOUD AND FLASHING!

 

- James

 

 

Thanks for the investigation. Can you try the test again with the SD card taken out once the game loads?

[+ZeroPage Homebrew]

5 hours ago, batari said:

On Concerto when some reported issues with the menu, the problem was that some SD cards were not releasing the bus as expected. Removal if the card could help determine if the same issue is happening here, or if it may be something else.

Thanks for the tip Fred! I just tried two things for a test:

 

1) Running the EF RAM test program on an older Concerto Encore I had

2) Removing the SD card immediately after loading the EF RAM test program

 

Unfortunately neither of them changed the outcome and RAM corruption still occurred while running the test program. ?

 

I also tried running the EF RAM test program from a cold boot (turning console off for 20 mins) and it ran perfectly for about 20 minutes again before the corruption started happening. After the corruption started happening on my older Concerto Encore, I tried it on my newer one and the screen went crazy with corruption. I then put in the UnoCart and it worked perfectly. I put back in the new Concerto Encore cart and the corruption was in full effect again.

 

- James

[+ZeroPage Homebrew]

9 minutes ago, batari said:

Thanks for the investigation. Can you try the test again with the SD card taken out once the game loads?

You bet! Just tried removing the SD card on Penult after it loads. It behaves the same as if the SD card was in the cart, after moving a couple of spaces the screen gets corrupt and then crashes.

 

- James

[+Karl G]

1 minute ago, ZeroPage Homebrew said:

You bet! Just tried removing the SD card on Penult after it loads. It behaves the same as if the SD card was in the cart, after moving a couple of spaces the screen gets corrupt and then crashes.

 

- James

So it happens on Penult with the Encore still? I couldn't remember if it was still an issue with Penult after I switched to 128K DFSC.

[+ZeroPage Homebrew]

5 minutes ago, Karl G said:

So it happens on Penult with the Encore still? I couldn't remember if it was still an issue with Penult after I switched to 128K DFSC.

Yup, the screen corruption and crashing still happens with 128K DFSC Penult on Encore, works perfectly on UnoCart.

 

- James

[+batari]

6 minutes ago, ZeroPage Homebrew said:

Thanks for the tip Fred! I just tried two things for a test:

 

1) Running the EF RAM test program on an older Concerto I had

2) Removing the SD card immediately after loading the EF RAM test program

 

Unfortunately neither of them changed the outcome and RAM corruption still occurred while running the test program. ?

 

I also tried running the EF RAM test program from a cold boot (turning console off for 20 mins) and it ran perfectly for about 20 minutes again before the corruption started happening. After the corruption started happening on my older Concerto, I tried it on my newer one and the screen went crazy with corruption. I then put in the UnoCart and it worked perfectly. I put back in the new Concerto cart and the corruption was in full effect again.

 

- James

Thanks,

 

Encore and Concerto have exactly the same loader and driver, so it sounds like it's related to that. I will take a look at the code and see if I notice anything strange. I personally do not experience the issue on any of my consoles or carts so I will have to rely on others to test things.

 

Are you only having issues with SC games? I know Zippy is a non-SC game and you said this works fine, but I am not sure whether the other ones use Superchip or not.

[+Karl G]

4 minutes ago, batari said:

Are you only having issues with SC games? I know Zippy is a non-SC game and you said this works fine, but I am not sure whether the other ones use Superchip or not.

My 2600 Words game is non-SC, and my little test program is also non-SC. I don't think that's a factor.