Individision AVA MK3 4000D/CD32 no signal at all

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.


Also users that are not logged in can not create new threads. This is a, unfortunately needed, counter measure against spam. Please create an account and log in to start new threads.

Don't Panic. Please wash hands.
  • Hi! I ordered and received an AGA MK3 4000D/CD32 earlier this summer and am finally installing it. The instructions seemed reasonably clear and installation proceeded smoothly. Bending the pins out of the way was easy enough and it was clearly needed, so I know I have the board in the proper orientation. Also, everything matches the photos.


    However, on power up, I have no video signal from the board at all, either on VGA or HDMI. The computer is booting and behaving normally, with correct output on the built-in output, but I have no signal from the MK3 whatever as far as my monitor is concerned. The monitor works fine on both HDMI and VGA with other systems, a PS4 (HDMI) and PC (VGA).


    I also installed the optional CIA chip adaptor and I have no response to left shift-control-tilde; I press it, and if I have my cursor in a command line, I see an inverted character on the CRT, but no behaviour changes happen - perhaps of course because there is no video signal.


    Since nothing was working, I tried removing the CIA chip adaptor and that did not make any change in the lack of video signal, and I also removed, re-cleaned, and re-set the main MK3 board, and also reset the ribbon cable (making sure not to change how it was connected) and again, no response.


    I suspect that I have a DOA board but I am happy to try suggestions, I'd much rather find I've made some correctable mistake!

    Thank you!

  • DOA is extremely unlikely with our products, as every single unit has passed a test, including cables and adapter boards. What you unpack is the exact setup that has been tested here.


    First guess is that the main unit isn't pushed all the way down on the chip. If the unit isn't powered, it will of course not produce any picture.


    If you don't get anything on the HDMI monitor at all, you might want to check if it's set to show a picture from the HDMI input. On most monitors, the button to select the input is labelled "source". While most PCs also control the HDMI device through a command channel, Indivision AGA MK3 does NOT do that, so you will have to choose the input manually.


    As far as I can see from the pictures you've provided, cables are connected correctly.

  • I pushed it down as far as it seemed able to go each time. It makes a definite clear click and it at least appears to be parallel with the motherboard. The second time I tried pushing down further after it clicked, but the motherboard flexed enough that I got nervous about it. But I will try re-seating the main board again, and report back.

    (I don't have a camera small enough to look under the board when it's in position, really. But I'll see what I can do with my phone ... and maybe a mirror. I may have a small mirror somewhere I can use for this.)


    I'm certain it's not a wrong-input-on-the-monitor issue. I tested every possible combination of monitor input and board outputs with both VGA and HDMI (the monitor accepts both and both are known working, as are the cables I used), but all monitor inputs report no signal regardless of how they're connected.


    It is good to know that the MK3 doesn't issue a device select through. Since the monitor has only one VGA input (vs. 2 HDMI inputs) I will just set it to that and test on VGA.

  • Okay, so I retested everything to no effect, reseated the board again, and retested everything again, still to no effect, and took some more photos.


    I also re-verified the cable as working and also tried using a second VGA-only single-input monitor that stays on even if it has no signal, and it reported no signal and displayed no image either. So there's definitely no signal.

    I managed to get a few decent photos showing what appears to me to be the board seated all the way down. When I push it onto the chip it makes a nice even chunk / brrrrrit! sound, kind of like if you remove heavy-duty velcro very quickly. I don't know whether that's the correct sound, but it's the sound it makes, and it goes on very straight (as far as I can tell) every time.


    I'm attaching photos of the board in place, the board removed, and also a couple of pictures of the primary monitor showing images from my first PC gaming computer which I recently got back up and running (because I still had it and I could xD ) as verification of cable and monitor.


    I am in no way an electrical engineer, but I do have a small amount of test kit and am capable of successfully building fairly complex electronic kits. I've even built my own microphones and simple hardware controller interfaces from bare components, using my own designs, successfully. If you were to give me some test points and measurements I could make with a multimeter, I could try to get readings from the board if that would be of help.







  • kind of like if you remove heavy-duty velcro very quickly.

    That is exactly what you want to hear - it's the sound of "making good contact".


    Please try to run the rescue disk - it will attempt to detect the hardware and re-flash it. The output of that disk is very interesting.


    If that doesn't help, I'd like to take a look at high-res pictures of Indivision, as it wouldn't be the first time that a component got sheared off somewhere between our QC and your computer.

  • Okay so I went searching and I can't find anything about a rescue disk - could I have a pointer to where I can download this? (And will it fit on a floppy? My A4000 doesn't have an optical drive.)

  • Okay. Real progress!

    Getting a floppy made was a bit of an adventure, but I did it, and then I had to make another one to make a bootable recovery disk, but that also worked. My housemates saw me hauling a floppy drive out of the server closet and think I am insane, but that's another matter.

    Initially just running the configuration software, it could not find the card hardware at all, so it was definitely not responding to anything.


    After making the second (actual rescue) floppy and booting of of it, it was able to re-flash the card, and it has come up! It came up strange, insisting on showing only part of the screen and on the left side of the monitor, but it came up. I'm including a photo of that.

    After a lot of playing around in settings, I've got a stable image! It looks the same on VGA as it does on HDMI, which is good.

    BUT...

    The only thing at this point is that I am getting new intermittent horizontal noise. It's not happening right now but it happens regularly. I'm attaching a couple of photos. They are now showing up on both the CRT on the native monitor output and on the HDMI and VGA output and they appear to be the same noise at all times.

    Could this be a power load issue? I am on my original PSU. I have purchased a Corsair PSU and cable set to adapt it to use in the Amiga, and hopefully? that will solve the problem? But I don't know.

    Hopefully the photos will show it clearly for you.

  • Those lines are not FLickerfixer-related, but they originate in the chipset. If you are operating the Amiga chipset with a higher resolution (or at least higher pixel clocks; not sure what I am seeing, but it seems like your standard installation is on a "double" mode, which uses SHires-pixels and double-CAS memory access). These modes are inherently instable on some AGA chip sets, and we've included software-settings to apply hardware patches (yes, as weird as this may sound).


    Go to the "advanced" settings and play with the CCK capacitance, CCK pull-up and pixel clock pull-up features until the artefacts go away.


    On another note, seeing "BBS" and "UUCP" directories on your "WideOpenSpaces" partition gave me a flashback of my own ever-growing scripts to connect my BBS via UUCP/ISDN to the internet and get usenet groups into the BBS. Noisy A4k running 24/7 in my one-room apartement...

  • sumax!ukma!lorbit - later uunet!microso!lorbit - that was me. AmigaUUCP on... well, what started as an Amiga 500, anyway. By the time it was doing all that it had a 14mhz 68010 and two Zorro slots and a case I made of lumber acquired from the scrap bin at school. xD The BBS you see there was single-user dial-in BBS software I wrote to let people use my Usenet feed. It wasn't amazing, but it worked. I never ran it on the A4000 but I did kind of archive it there.

    (My biggest BBS was a multi-user system with... well, in theory, I think the limit was 100 simultaneous users? We rarely broke 20 at once, and I think our peak was 28. _That_ was written in FORTRAN IV and ran on the school's IBM mainframe. I even had my own hypertext language before HTML called DISPLA, but the links were all down at the bottom of the page and HTML was better in other ways anyway.)

    Anyway, that's incredibly off-topic, sorry, but, well, there y'go. xD

    I'm running in 680x480 non-interlaced productivity mode on the desktop. I've never had these flickers appear before today, which is why I was wondering about additional power load of the flickerfixer. It also looks like bus noise I saw on my Commodore 64 when I was a little kid, too. (At least, that's what I was told it was.)


    I will play around with those settings - I see they're just checkboxes so it shouldn't take too long to go through all the possible options - and hopefully that will fix it. I'll report back either way.

    Thanks!

  • Whelp

    I turned my Amiga on this afternoon (for the first time since getting things mostly working as reported above) to try the various hardware patch settings and...


    ...we're back to the beginning with NO SIGNAL again.

    I try a power cycles and a couple of reboots reboots, still no signal.


    Running the rescue disk brings signal back but broken again, everything squished and off to the left. See photo.


    I pulled the (prone to rotting) CMOS battery a long time ago and have not wired in a replacement. The working configuration survived much shorter power cycles while trying to get things to work, but overnight is apparently too much for it.

    Having to run rescue and reconfigure the board from scratch every time is obviously not a very good solution. Will a battery fix this or is something else wrong?

    BTW: selecting CCKLine Pull-Up seems to get rid of the bus-noise-like horizontal line flicker, at least so far, so that's good!

  • Please check the PSU for ripple and startup-behaviour. If the 5V rail goes high,. but then drops again (swing-in behaviour), this can cause the observed behaviour.

  • Okay so, here's the latest.


    First, voltages:


    1. I did a bunch of power cycles. Most of the time it starts up at 5.06-5.07V. I only saw higher than that once, at 5.13V. It then settles down to 5.02-5.04V depending upon what the machine is doing. This is measured using the 5V pins on the hard drive connectors. If the markings on the power supply are correct, I think that should be the same power rail as 5V for the motherboard. I did it this way because it was a lot easier to get a short ground loop. If that's bad, let me know what is better and I will do that.


    2. My little digital oscilloscope isn't very good and is probably not up to this task, I'll be up front about that. But I'm not able to detect much ripple with it. So I tried using my (true RMS) multimeter, testing for AC voltage on the DC 5V lead. Doing that, I see 1.2mV of AC, which seems about like what you'd expect. But I don't know that for sure.


    Second:


    3. I've discovered that I do not actually have to re-flash the device to get an image restored, and in fact, not doing that is actually better. I don't understand this at all. I boot to the rescue floppy, control-C before it starts the firmware flash, soft reboot, and the machine comes up with proper display. It's not even narrow, it's just correct.

    I am so confused.

  • (I mean, off the rescue floppy, it comes up in some lower-clock-rate graphics mode. There are chunky pixels, as opposed to productivity. Since this is an NTSC Amiga, I'm presuming it's 640x200 or... wait. I can check!)

    Pre 800x436 NTSC

    Pre 800x600 72Hz VESA

    H: 15.79kHz 1820 20210108 22%

    V: 60Hz 263 DVI


    I hit control-C before the flash happens, eject the floppy, soft reboot from the keyboard, and it comes up correctly and reports this:


    Pre 640x480 MultiscanProd

    Pre 1280x1024 60Hz VESA

    H: 31.52kHz 912 20210108 34%

    V: 60Hz 526 DVI SHRES

  • But I'm not able to detect much ripple with it. So I tried using my (true RMS) multimeter, testing for AC voltage on the DC 5V lead. Doing that, I see 1.2mV of AC, which seems about like what you'd expect. But I don't know that for sure.

    That won't capture ripple - you truly need an oscilloscope for that, as ripple is measured in the few-kHz to 20MHz range (actually defined with 20MHz bandwidth limiter). Anything below 20mV is top-notch, and max. 50mV is acceptable. Your 1.2mV is impossible to achieve with a switchmode PSU.


    3. I've discovered that I do not actually have to re-flash the device to get an image restored, and in fact, not doing that is actually better. I don't understand this at all. I boot to the rescue floppy, control-C before it starts the firmware flash, soft reboot, and the machine comes up with proper display. It's not even narrow, it's just correct.

    That's a clear hint at the ramp-up behaviour of the PSU causing trouble - we've had this before with other customers who were operating unsuitable or very old PSUs. What happens is that power comes up, Indivision starts configuring the FPGA, and in the middle of the process that transfer from flash into the FPGA is interrupted due to a power-drop (swing-in behaviour). When the rescue disk boots, it soft-triggers a re-configuration of the FPGA, and since the PSU is now in a stable state, it works.


    You may fix this already with re-capping the PSU.

  • What happens is that power comes up, Indivision starts configuring the FPGA, and in the middle of the process that transfer from flash into the FPGA is interrupted due to a power-drop (swing-in behaviour). When the rescue disk boots, it soft-triggers a re-configuration of the FPGA, and since the PSU is now in a stable state, it works.

    So... there's something specific about the floppy reboot that does this soft triggering?

    I ask since a soft (control-amiga-amiga) reboot off of the hard drive never makes the FPGA come online. Stays at "no signal" if I just do that. I have to use the rescue floppy. (I haven't tried any other floppies.)

    Regardless, I have a replacement power supply I'd already bought. So I will hook that up and see if that fixes the problem. I will also see if I can find a recap kit for the original Amiga PSU so I don't have to order the components individually from digikey.


    (Do you recommend any particular recapping kit?)


    But until I figure that out, I already have something that should at least electrically work - they've been used successfully by others around here in Amiga 4000s - so that should at least answer the question.

    I do kinda want a better/proper oscilloscope but that's a lot of money for something I don't use all that much. xD

    Thanks!

  • So... there's something specific about the floppy reboot that does this soft triggering?

    Yes, the flash program triggers a re-config.

    I ask since a soft (control-amiga-amiga) reboot off of the hard drive never makes the FPGA come online.

    That's expected. We've had that on the very first Indivisio AGA (the black one from 2008), and it annoyed people to lose sync on every reset, so this was eliminated on the MK3 (I believe the MK2 still does it).


    (Do you recommend any particular recapping kit?)

    I'd always do my own research, but that's just me. Haven't looked at kits, and I don't even know if they exist for PSUs.

  • Okay, thank you!

    One thing I might suggest is that releasing a tiny application to do the reset and nothing else - the firmware resetter without firmware reset or delays, basically - could be very helpful, particularly since you've seen this before!


    I will hopefully solve this problem with a new PSU / recapped PSU, but it seems to me that I could also solve it by downloading such an application and throwing it into startup-sequence.


    Hopefully I wouldn't need that! It is, I agree, kind of a kludgey way to go about it. But it would also get the machine working apparently reliably much more quickly.


    Thanks again and I will hopefully have something to report next week! ^_^

  • I decided to quickly wire up the Corsair replacement PSU, an SF450, rated 80+ gold efficient so that's nice, just to see if things would actually work with it. Then, once it did, I could make it nice and clean later.


    The behaviour is exactly the same. No signal until I boot off the rescue floppy, then everything is fine.


    :(