Stella 6.4 released

[Schvenn]

 

3 hours ago, Thomas Jentzsch said:

That sounds very weird, maybe a connection problem with the console port? What do you mean with "Stella joystick"?

 

You can try in the advanced settings (Input) to suppress jitter more than by default.

 

Another weirdness. The game should start when you release the button after a short time. A long button press brings up the "Power-On options".

 

Hard to tell what's going wrong on your side. Anyone else experiencing similar problems?

 

I hope @DirtyHairy can help with your USB problems.

By "Stella joystick" I was referencing the Hyperkin Trooper, Retron joystick.
It works flawlessly.

I find it very hard to believe it's a hardware/pin/wiring problem since a CX-10, Thrustmaster and Genesis controller all suffer from the same problem with "up" only on the Retron and yet the Retron joystick works perfectly.

That seems more likely a software issue.

 

I will try to adjust the jitter settings to see how that works.

I've never noticed that setting before; newb mistake.

 

As for the paddle button press, I've tried holding it down for about a second, but no menu opens.

I wasn't aware of that feature and will test it out.

Either way, I wasn't referring to the game not starting. 

I'm referencing in game usage.

I need to press and hold the button for probably 0.25 to 0.5 seconds in order to get it to act like a proper fire button.

It's simply not sensitive enough to recognize proper rapid fire, like you would use in real world gaming.

Otherwise, the Stella doesn't seem to recognize it.

Is there a setting for sensitivity in the interface?

I haven't found one.

 

As for testing the USB, I did use a freshly installed image file, in order to eliminate any artifacts, with the same results.

I also deleted all the settings files, not just the SHM and I got the same results, testing various combinations of files.

Thus the reason that I specified the SHM, because I narrowed it down to that file, rather than the original settings file.

If the original setting file exists on boot-up, the USB works, but as soon as I restart the Retron, wherein the SHM file now exists, the USB no longer works.

[+stephena]

Can you try all this stuff on Stella and an actual desktop computer (Windows, Linux, Mac, it doesn't matter)?  At least we can narrow down if it's a Stella issue everywhere, or only on the R77.

[Rodney Hester]

Going the other way, I'm planning a fresh setup of 6.4 today on my R77 with both Hyperkin and third party sticks (including a 2600-daptor) and can provide a second testbed.

Edited November 5, 2020 by Rodney Hester

[Rodney Hester]

OK, limited controller testing looks OK so far, but I don't seem to be able to use a USB keyboard with the OTG spider cable.  Any particular magic required here?  (Switch is in OTG position, have tried multiple keyboards)

 

Kept screwing with it until it worked, but whether I use the joystick or keyboard, I can't seem to turn off (or even select) "Correct aspect ratio", which makes the scanlines look really crappy.

 

Still rounding up controllers to test

Edited November 5, 2020 by Rodney Hester

[Rodney Hester]

Controllers - tried the following, all successfully:

 

- Retron 77 stick (two different ones)

- Atari Flashblack 9-pin sticks (two different ones)

- Atari Flashback 9-pin stick connected to 2600-daptor II and OTG cable

 

I fear Schvinn really does have a hardware problem of some variety ?

[Rodney Hester]

On 11/2/2020 at 4:38 PM, stephena said:

The R77 port gets a few new features, in addition to the bugfixes from the general release.  Specifically, it now supports an AtariVox plugged into a USB adaptor, just like the mainline Stella port.  As well, any ROM from a cart you insert can now be saved to the SD card.  So the R77 essentially becomes a ROM dumper!

What is the procedure for saving the cart ROM to SD card?

[+ZeroPage Homebrew]

13 minutes ago, Rodney Hester said:

What is the procedure for saving the cart ROM to SD card?

I did a quick demonstration of dumping a ROM to SD card on the ZeroPage Homebrew Twitch stream last night. Here's the direct link to it if you'd like to watch (38m 16s)

 

Alternatively, here's the text instructions on how to do it (scroll down to 'Dumping ROMs to SD'):

https://github.com/DirtyHairy/r77-firmware-ng/blob/master/README.md

 

- James

Edited November 5, 2020 by ZeroPage Homebrew

[Rodney Hester]

Thank you!

[+Nathan Strum]

Long AtariVox phrases are being cut off early (even earlier than they were when being garbled). Once they're cut off, the AtariVox stops working in that game until the game is reloaded.

 

I've tried this on the Gorf voice test ROM that I sent to Stephen, as well as the final binary for Wizard of Wor Arcade (both have long phrases that get cut off). Both work fine on real hardware, with the same AtariVox. I'm using one of Richard's USB interfaces with Stella (same one I've always used).

 

I've also tested this on 6.3 (which I hadn't been using with the AtariVox yet), and it does it there as well. 6.2.1 garbles the voices instead of cutting them off. I'm running MacOS 10.14.6.

[DirtyHairy]

@Schvenn I am afraid I cannot reproduce  any of your issues here --- the device runs stably and predictably as it should. As there are no variations between different R77 (apart from possible hardware defects), I think the issues that you are experiencing are somehow related to your setup. The SHM file does not contain any persistent data, and deleting it is of no consequence to the contents of the DB, so I do not understand how this could be related to your issues.

 

There is no delay whatsoever when a button is pressed --- the press gets processed immediately. Your input issues with the builtin ports make me suspect a hardware defect on the input ports or the circuitry that drivers them.

 

A few suggestions for things to check:

• Are you sure that your SD card is actually saving written data properly? Going into a "silent read-only" mode is a known failure mode for SD cards (happened to me only a few days ago). In order to test for that, clean out the contents of your SD card, eject it, reinsert it into your computer, check that it is still empty and then rewrite the SD image.
• Try a different USB power supply; maybe there are voltage fluctuations or noise on the USB bus.
• If there is latency, make sure that your TV is set to game mode (if available) and doesn't run any image post processing; this can cause a very noticeable lag between gameplay and video / audio.
• If you have a different USB Y cable, try that one, maybe the cable that you are using is defective.

In addition, the firmware can dump information on the system setup into a log file. For that, create an empty file called "developer" at the root of the SD card in order to enter developer mode (the next boot will take about 30 seconds as a key file is generated, subsequents boot in dev mode will only be slightly slower than "ordinary" boots). After the device has booted, you will find a file called "sys/diagnostics.log" on the card. Post that file for both the cases of working and non-working USB controllers.

Edited November 5, 2020 by DirtyHairy

[Keatah]

On 11/3/2020 at 11:40 AM, Thomas Jentzsch said:

That might be feasible, but probably not first priority.

 

First there are multiple things which change during warming up, e.g. some TIA glitches only happen on warm consoles. And in between they fluctuate. Also, how about the RGB phase shifts, shouldn't they be affected by temperature too?

And are the phase changes linear (I doubt that) and how do we sync these? We could only find out by extensive experimenting how things change in time. And then we need heaps of parameters to simulate one console exactly.

 

Yes, we could skip all this, but then this is more like a nice effect instead of correct emulation.

Emulating these effects (as I see it today) are nice amusements and certainly not necessary. And I don't know of any games that rely on a warmed-up or warming-up VCS to affect gameplay.

 

These anomalies seem to be annoyances the original engineers didn't want or just considered not important. These anomalies may also have changed or become more or less severe over a console's lifespan.

These anomalies are different for each console.

 

IMHO I'd rather see development efforts focused on other features and big fixes..

 

[DirtyHairy]

4 hours ago, Rodney Hester said:

Kept screwing with it until it worked, but whether I use the joystick or keyboard, I can't seem to turn off (or even select) "Correct aspect ratio", which makes the scanlines look really crappy.

You have found a Stella bug there, I can confirm that the option cannot be focused via keyboard or game controllers. I'll create an issue.

 

Unfortunately, there is not much that can be done about the scanlines (apart from disabling them). The emulated TV displays around 250 lines. With scanline emulation enabled, this is doubled to around 500 lines. The R77 runs at 720p, so those 500 lines have to be scaled vertically to 720 lines, and that causes a slight moirée pattern. Disabling aspect ratio correction will give you integer scaling only, but the price are black bars and an aspect ratio that does not match a real TV. I have done some experiments with 1080p on the R77 (which produces much nicer scaling), but the GPU cannot sustain 60FPS and causes frame drops.

[+stephena]

15 minutes ago, Nathan Strum said:

Long AtariVox phrases are being cut off early (even earlier than they were when being garbled). Once they're cut off, the AtariVox stops working in that game until the game is reloaded.

 

I've tried this on the Gorf voice test ROM that I sent to Stephen, as well as the final binary for Wizard of Wor Arcade (both have long phrases that get cut off). Both work fine on real hardware, with the same AtariVox. I'm using one of Richard's USB interfaces with Stella (same one I've always used).

 

I've also tested this on 6.3 (which I hadn't been using with the AtariVox yet), and it does it there as well. 6.2.1 garbles the voices instead of cutting them off. I'm running MacOS 10.14.6.

Really?  I tested this with those ROMs, and on a Mac, just to make sure to duplicate your system as much as possible.  I'm running macOS 10.15, so that's the only difference.  And I tested with 3 different Avox-USB convertors; one from Richard, one from the second batch of Richards (with incorrectly inverted CTS signal), and a flashed 2600-daptor II.  All were working fine.  I guess I can test again.

 

Is there a ROM that you can provide that everyone can see and use?  So someone else can test this independently??

[+stephena]

9 minutes ago, DirtyHairy said:

You have found a Stella bug there, I can confirm that the option cannot be focused via keyboard or game controllers. I'll create an issue.

This is already fixed internally (about 11 hours ago).  It will be included in the next release.

[+Nathan Strum]

1 hour ago, stephena said:

Really?  I tested this with those ROMs, and on a Mac, just to make sure to duplicate your system as much as possible.  I'm running macOS 10.15, so that's the only difference.  And I tested with 3 different Avox-USB convertors; one from Richard, one from the second batch of Richards (with incorrectly inverted CTS signal), and a flashed 2600-daptor II.  All were working fine.  I guess I can test again.

I tested the same AtariVox and adapter on a Mac laptop running 10.15.6, and it's working fine there. But not on my Mac Mini (which is the one running 10.14.6). I plugged the adapter directly into a USB port (no hub), restarted the Mini, same results.

1 hour ago, stephena said:

Is there a ROM that you can provide that everyone can see and use?  So someone else can test this independently??

I'll ask John if we can provide the Gorf voice ROM, or part of it.

[+Nathan Strum]

John's letting me share the voice test binary for Gorf:

 

avox_gorf_test_rev3.bin

 

Left/right changes the lower number (the phrase), up/down changes the upper number (the "space rank", where applicable), fire plays the phrase. 00 is blank (a pause). 1E is one of the longest phrases.

 

[Rodney Hester]

3 hours ago, DirtyHairy said:

Unfortunately, there is not much that can be done about the scanlines (apart from disabling them). The emulated TV displays around 250 lines. With scanline emulation enabled, this is doubled to around 500 lines. The R77 runs at 720p, so those 500 lines have to be scaled vertically to 720 lines, and that causes a slight moirée pattern. Disabling aspect ratio correction will give you integer scaling only, but the price are black bars and an aspect ratio that does not match a real TV. I have done some experiments with 1080p on the R77 (which produces much nicer scaling), but the GPU cannot sustain 60FPS and causes frame drops.

I get it.  I have found that (even on a display that does not use overscan) simply setting overscan to 5% removes the moire effect and produces good-looking scanlines with all other settings at default.  It's a reasonable compromise.

[Schvenn]

4 hours ago, DirtyHairy said:

@Schvenn The SHM file does not contain any persistent data, and deleting it is of no consequence to the contents of the DB, so I do not understand how this could be related to your issues.

• Are you sure that your SD card is actually saving written data properly? Going into a "silent read-only" mode is a known failure mode for SD cards (happened to me only a few days ago). In order to test for that, clean out the contents of your SD card, eject it, reinsert it into your computer, check that it is still empty and then rewrite the SD image.
• Try a different USB power supply; maybe there are voltage fluctuations or noise on the USB bus.
• If there is latency, make sure that your TV is set to game mode (if available) and doesn't run any image post processing; this can cause a very noticeable lag between gameplay and video / audio.
• If you have a different USB Y cable, try that one, maybe the cable that you are using is defective.

In addition, the firmware can dump information on the system setup into a log file. For that, create an empty file called "developer" at the root of the SD card in order to enter developer mode (the next boot will take about 30 seconds as a key file is generated, subsequents boot in dev mode will only be slightly slower than "ordinary" boots). After the device has booted, you will find a file called "sys/diagnostics.log" on the card. Post that file for both the cases of working and non-working USB controllers.

Testing round 3: Test alternate software solutions
Joysticks - "up" doesn't for for any of them, except the Retron DB9 or a USB controller. I swapped ports in the interface and got the same results.
Paddles - the button doesn't work well in any of the games, but controls are jittery. Adjusting the settings smoothed out the jitter tremendously. The buttons still don't work in game. They do however, work flawlessly in the GUI and joystick based games! That means this problem is software.
USB - reimaged the SD card and reset all the settings and now, USB works every boot!

Testing round 4: Try original software
I put the original Retron SD card in the device, with their antiquated version of Stella and tried all of the above.

Joysticks - Same result! UGH! Hardware? Really? Time to contact Hyperkin.
Paddles - the button and controls work flawlessly in games. This confirms that the problem is software.
USB - no issues whatsoever.

Results: Paddle controls work properly now in Stella 6.4, but paddle buttons do not.

 

[+Nathan Strum]

7 hours ago, Nathan Strum said:

I tested the same AtariVox and adapter on a Mac laptop running 10.15.6, and it's working fine there. But not on my Mac Mini (which is the one running 10.14.6). I plugged the adapter directly into a USB port (no hub), restarted the Mini, same results.

Okay, against my better judgement  I updated my Mini to OS 10.15.7, and five hours after I started*, the AtariVox is now playing the phrases properly.

 

*I had to run a bunch of updates, and back everything up first.

[+stephena]

19 hours ago, Schvenn said:

Results: Paddle controls work properly now in Stella 6.4, but paddle buttons do not.

We will try to address this over the next few weeks, and have a fix for the next release.  Thanks for the extended testing.

[+stephena]

17 hours ago, Nathan Strum said:

Okay, against my better judgement  I updated my Mini to OS 10.15.7, and five hours after I started*, the AtariVox is now playing the phrases properly.

 

*I had to run a bunch of updates, and back everything up first.

Hmm, very strange.  Perhaps the FTDI drivers that come with 10.14 are defective in some way?  It sounds like the 'Clear to Send' signal was not being processed, or even sent by the device.  Good to hear that it works in 10.15, and matches my testing.

 

I am unable to test older versions of macOS, since I have only one machine that is being used for work-from-home, and I can't risk messing with it in any way.

[+Nathan Strum]

I wasn't aware there were built-in FTDI drivers. I've been using the one from here: https://www.ftdichip.com/Drivers/VCP.htm

[+stephena]

6 minutes ago, Nathan Strum said:

I wasn't aware there were built-in FTDI drivers. I've been using the one from here: https://www.ftdichip.com/Drivers/VCP.htm

They're built-in in 10.15, so I assumed they were for 10.14 too.  Maybe this was the cause of the issue??

[+Nathan Strum]

31 minutes ago, stephena said:

They're built-in in 10.15, so I assumed they were for 10.14 too.  Maybe this was the cause of the issue??

No idea. But 10.14 wouldn't work without them being installed. I've installed them in 10.15, and it's working. But maybe MacOS is ignoring them now in favor of its own?

 

To be fair, I've had nothing but USB problems with that Mini (one of the reasons I went ahead with the update - hope springs eternal). I have a 10.14.x Mac at work though that's never had the FTDI drivers installed. I can bring my AVox and USB interface in and see what happens there.

[+stephena]

3 minutes ago, Nathan Strum said:

I have a 10.14.x Mac at work though that's never had the FTDI drivers installed. I can bring my AVox and USB interface in and see what happens there.

It would be interesting to see, whether it's a 10.14 issue in general, or something specific with your machine/hardware.