Mk3 horizontal glitches only at startup

Caution: Non registered users only see threads and messages in the currently selected language, which is determined by their browser Please create an account and log in to see all content by default. This is a limitation of the forum software.
Don't Panic. Please wash hands.
  • Hello,


    I have a problem when I power up my A1200 with the Indivision AGA Mk3. For some seconds (In some cases 2-5 minutes) the screen presents some horizontal glitches like you can see in the photo. This happens to me with different televisions and different HDMI cables. I tried some options like CCKLine Pull-Up or CCKLine Capacitance but the problem appears again when I power off the Amiga for some hours and power up again. The problem is strange because after the first few minutes, the Mk3 works like a charm without any problem.


    I don't know, but I think this problem appeared when I updated to 1.5 or 1.4 because when I received the product, I don't remember having this problem. Yesterday I installed the latest firmware 1.6 and the issue continue though it seems to disappear faster.






    Another question. Is there somehow to know the firmware version installed in my Mk3?


    Regards.

  • Hi,

    I am also getting same artifacts when my Amiga is first powered from cold. I've been waiting to see if a firmware update might improve things before reporting this but since others are having a similar issue, I thought I'd better report it at the same time.


    The glitching on my Amiga starts a few seconds after booting and looks exactley the same as it does in the pictures provided by wolfa. If I activate the CCKLine pull-up they disappear completely but once the Amiga has been running for an hour or so the glitching starts again, and then I have to deactivate the CCKLine pull-up to stop them - after that the Amiga behaves fine all day. Activating the CCKLine capacitance does not seem to improve the glitching, and causes occasional pixel mis-read on my Amiga if active, so I leave that switched off. Changing the PCLK setting doesn't seem to make any difference to the glitching at all.


    I know this isn't the Mk3's fault, it's my Amiga that's causing the issue, and it mostly only happens in DPLPAL mode or higher on my Amiga. I've tested all firmwares and indivision tool versions up to and lncluding the v1.6. and so far whilst none have made things any worse, they haven't completely removed the issue either.


    My Amiga is a 1d4 A1200 with a CA-PSU, and a I believe its got a CBM Lisa in it, in case that helps.


    Just out of interest may I ask what you folks at Individual suspect is the likely cause of these glitches? Is it likely the phase or duty cycle of the CCK shifting over time and getting out step compared to another clock line, or is likely just the Amiga's CCK signal is a noisy mess, and the Mk3 has trouble locking onto it?

    Cheers!

  • What's the power supply you are using with this setup?

    I use an original A500 PSU (5V-4.5A, 12V-1.0A, -12V0.1A).

    If I activate the CCKLine pull-up they disappear completely but once the Amiga has been running for an hour or so the glitching starts again, and then I have to deactivate the CCKLine pull-up to stop them

    I have the same problem as you.

  • Sounds like a PSU-heat related voltage drift. The 4.5A PSU is a good one in terms of regulation, but will require a re-cap after all these years. Please have that checked.

  • Sounds like a PSU-heat related voltage drift. The 4.5A PSU is a good one in terms of regulation, but will require a re-cap after all these years. Please have that checked.

    I tried with another PSU and same problem. Anyway, I'll recap both because they are old A500 PSUs.


    I tested putting my A1200 with the new SuperPlus resolution and for now, I have zero problems. The issue appears only using HighGFX.

  • I tried with another PSU and same problem. Anyway, I'll recap both because they are old A500 PSUs.

    It's not excluded that both PSUs have the same problem. With the latest Firmware update, dependence on clock drift due to temperature has been removed, so what remains is effects that happen on the Amiga side, not inside the flicker fixer.


    If an electrolytic cap gets old, the equivalent series resistance (ESR) rises, and the usable capacity goes down. With rising temperature, the ESR goes down, and the effectiveness of the capacitor returns to a certain degree with rising temperature. I therefore believe that the added load of Indivision AGA MK3 takes the PSU to a limit that is "rather low" in relation to it's rated amperage, caused by the output cap leaking or drying out. Your description is just so fitting to the technical description that I'm pretty confident that you'll have this fixed after you've replaced the output cap on the 5V rail.


    For Nutties "I don't think so": Please provide ripple measurements to back up your conjecture.

  • My response about my PSU was because it is a CA-PSU from yourselves at Individual. I have two of them, and they've fixed every other issue I was having with both my machines which might have been power related. I was just saying I didn't think my issue could be related to the PSU as it's one of yours, and not a flaky old one, or a wretched MeanWell hack. I'm not daft enough to challenge you on whether Amiga problems are likely power related, I'm pretty certain you know what you're talking about more than I do! :)


    If I had to guess on the cause of my problem it'd be my flaky, nearly 30 years old, Lisa chip. I'm not complaining about it not working how I'd like, I'm actually just reporting it and hoping you guys can somehow work out a fix via the Indivision, to somehow compensate for Commodore's less-than-stellar chip processes and the age of my Amiga.


    To be fair to Commodore, I don't suppose they designed their things to last forever, or expected people to be running their graphics chips at 4x-8x the pixel count of a standard TV either but it's fun to push the envelope a bit.


    I think a lot of folks like me are pushing the resolution because we don't (yet) have the ability to easily get hold of an RTG card - I don't think I'd even be trying to use DBLPAL or higher if I had one of those. I'm just hoping you manage to release something before the oh-so-stupid-Brexit (please remember about half of us didn't vote for it) finally completes and I have to pay an additional 20% VAT on imports. Whatever the price is, it'll most likely be more than worth it to me though. :)

  • There's also electrolytic capacitors inside the A1200 that may have an influence. The Lisa chip being a digital component doesn't age too much, and I wouldn't be too worried about pushing the resolution. Most of the glitches happen not inside the Lisa, but on their way into the chip, which you can see clearly on the pictures of the original poster: It's all happening in blocks or whole lines, which can't happen inside the chip because it doesn't have that much memory. Lisa ist "mostly shift registers", and the maximum amount of pixels it's storing at a time is 64 pixels on double-CAS modes. It can't possibly be the source of a rectangular artefact that's extended over several lines.


    So if the pull-up/down resistors and CCK capacitor don't work, you need to dig deeper - my first guess would be the VBB capacitor C2X under Lisa (which I'd always top up with another 10µF), but I'd also measure if there's some noise on the 5V line - maybe from the spectrum of the noise, you can deduct what the source is (I'd look for mechanical harddrives, floppy or CD drives first).

  • PC PSUs (especially ATX) are known-bad for the Amiga. Please read our PSU FAQ. In short: Most PC PSUs have 12V as their main regulation rail, while the Amiga requires 5V to be the main regulation rail.

  • PC PSUs (especially ATX) are known-bad for the Amiga. Please read our PSU FAQ. In short: Most PC PSUs have 12V as their main regulation rail, while the Amiga requires 5V to be the main regulation rail.

    Thanks Jens. I am familiar with that. However, my Indivision worked perfectly between original firmware (one it has been delivered with) and latest one. Original one was bad and I have minor issues with 1.6 (artefacts on cold boot and EDID issue reported in separate thread). In general I find Indivision awesome device.

  • Hi Jens,


    Thank you for the information and advice, I've been too busy with work to try the extra capacitor but I should have time over the weekend, so I'll report back once I've fitted it and tested things.

    So if the pull-up/down resistors and CCK capacitor don't work, you need to dig deeper - my first guess would be the VBB capacitor C2X under Lisa (which I'd always top up with another 10µF), but I'd also measure if there's some noise on the 5V line - maybe from the spectrum of the noise, you can deduct what the source is (I'd look for mechanical harddrives, floppy or CD drives first).

    When I checked the pcb layout it shows the C2X capacitor to be on the reverse of the motherboard underneath where Alice is located, is that correct one (I'm a little worried I'll put the new one in the wrong place if I don't double check)? Is it better to replace the 330nF at C2X with the 10uF one, or to piggy-back the 10uf on top or to the side of the original? Also, do you think it might it be worth adding an extra capacitor at the C2B location as well? It looks from the schematic that this is the main decoupling cap for the VCC connection to Alice.

    I tried to read up on the VBB pin and why it is coupled to ground via a capacitor, so I hopefully could better understand why it helps to add extra capacitance but my web search skills weren't up to the task, does it maintain a reference voltage to part of the chip?


    Cheers and sorry for asking more questions, I'm just trying to learn! :)

  • When I checked the pcb layout it shows the C2X capacitor to be on the reverse of the motherboard underneath where Alice is located, is that correct one (I'm a little worried I'll put the new one in the wrong place if I don't double check)? Is it better to replace the 330nF at C2X with the 10uF one, or to piggy-back the 10uf on top or to the side of the original?

    Location is correct - "under Lisa" is of course wrong, I should have written "under Alice", with the circuit board in between. Since the 330nF cap is glued on, I prefer adding a second cap on top (or bottom for more confusion - it's the reverse side!).

    Also, do you think it might it be worth adding an extra capacitor at the C2B location as well? It looks from the schematic that this is the main decoupling cap for the VCC connection to Alice.

    Extra capacity always helps, sure. In case of the 5V rail, make sure that the voltage rating of a ceramic cap is at least 3x the loaded voltage - that's a rule of thumb to keep capacity degradation out of the equation. In this case, the cap should be rated 16V or more.

    I tried to read up on the VBB pin and why it is coupled to ground via a capacitor, so I hopefully could better understand why it helps to add extra capacitance but my web search skills weren't up to the task, does it maintain a reference voltage to part of the chip?

    Vbb is the negative bias voltage of the chip. In HMOS circuits, you require a small negative voltage "everywhere", and that's generated with a little charge pump on-chip. When the circuit gets more complex, you need some stabilizing, which is why Commodore has moved this voltage to the outside world on a pin. That was "cutting edge" back in the days - almost no other chip vendor did that. If you measure the voltage, it's about negative 3V. If you measure it while the computer is running, you'll notice that it drops considerably with heavy blitter operation (such as TGTest from AIBB). With the added 10µF cap, the drop is almost not measurable.

  • Hi,


    Just an update, I added extra 10uF 16V, 1206 10% Murata caps (they're quite big for smt and only just fit between the motherboard and the shield!) to both the C2X and C2B, piggy-backing them on top of (or is it underneath? :) of the existing ones. Unfortunately this doesn't seem to have changed the behaviour for me. I still get the flicker on cold boot if the CCK pull-up is not enabled and then flicker after an hour if the CCK pull-up is turned on. Not to worry, it was worth a try!

    Probably not related directly to this but while I was playing about with the overscan and resolutions to see if either made any difference to the flickering (it didn't seem to), I noticed that if I set the overscan to greater than 704 pixels in DblPAL progressive and laced, and then move the pointer sprite to the far right of the screen, the pointer doubles in height. The same seems to happen in DblNTSC prog too. I don't have an analogue multisync monitor to test if Amiga's normal video out to see if that does the same thing. If I find extra stuff like this, would you like me to put it into a different thread to stop the topic from wandering?


    Cheers! :)

  • Probably not related directly to this but while I was playing about with the overscan and resolutions to see if either made any difference to the flickering (it didn't seem to), I noticed that if I set the overscan to greater than 704 pixels in DblPAL progressive and laced, and then move the pointer sprite to the far right of the screen, the pointer doubles in height. The same seems to happen in DblNTSC prog too. I don't have an analogue multisync monitor to test if Amiga's normal video out to see if that does the same thing. If I find extra stuff like this, would you like me to put it into a different thread to stop the topic from wandering?

    Yes, please export that to a new thread. My guess is that it's an effect from the AGA chipset, as Indivision AGA MK3 does not make a difference between sprite and bitmap data, so changing the Y-height is nothing that the flicker fixer can do. In any case, it'll be interesting to investigate it, and especially find out what's happening on the chipset side, as I don't even know if this is properly emulated.