Amiga Digital video not work in Amiga 500 Rev 3

[kipper2k]

Hi all,

Have tested this with 2 different Amiga 500 Rev3s and neither will work with the adapter. The Pi splash screen will appear but the Rev3 refuses to display (black screen only) through HDMI. I have swapped Denise and the Agnus with no joy. I can't find a schematic on the Rev3 for comparison so not sure what the issue is.

[culexus8]

Do you get the OSD menu to pop up if you hold the switch that is connected to the adapter board?

[IanSB]

If the board works in other revisions but the menu won't appear when you press the button, then the most likely cause is that the pixel clock is missing. This is generated from the CDAC line in "Denise" mode and the 7Mhz clock in "Super Denise" mode. Have you tried setting the link to Super Denise mode? Can you check if those signals are present on the rev3 with an oscilloscope?

[kipper2k]

@IanSB thanks for info, i will try this later today... edit,

i tried swapping the jumper and i ended up with the blinking... http://www.kipper2k.com/pics/rev3blinking.mp4

i tried changing pal/ntsc mode and reset the prefs back to default and no change

[IanSB]

@kipper2k

i tried swapping the jumper and i ended up with the blinking

Blinking means the sync timings don't match those in the profile so something is different about the sync signal on your rev 3.

Your symptoms seem to match those that @hansliss is seeing with an A2000 See: https://github.com/c0pperdragon/Amiga-Digital-Video/issues/11

Can you post a photo of the source summary page in the info menu.

[hansliss]

@kipper2k

i tried swapping the jumper and i ended up with the blinking

Blinking means the sync timings don't match those in the profile so something is different about the sync signal on your rev 3.

Your symptoms seem to match those that @hansliss is seeing with an A2000 See: #11

Can you post a photo of the source summary page in the info menu.

I found schematics for the A500 in an "Amiga 500 Introduction" (https://museum.syssrc.com/static/Amiga_500_Introduction.pdf) that does indeed show both /CDAC and /CSYNC (pin 34 and 32) missing from the Denise socket. It doesn't say which revision they depict but it seems likely that it's an early one.

[hansliss]

@kipper2k

i tried swapping the jumper and i ended up with the blinking

Blinking means the sync timings don't match those in the profile so something is different about the sync signal on your rev 3. Your symptoms seem to match those that @hansliss is seeing with an A2000 See: #11 Can you post a photo of the source summary page in the info menu.

I found schematics for the A500 in an "Amiga 500 Introduction" (https://museum.syssrc.com/static/Amiga_500_Introduction.pdf) that does indeed show both /CDAC and /CSYNC (pin 34 and 32) missing from the Denise socket. It doesn't say which revision they depict but it seems likely that it's an early one.

From what I can gather, on the A500 rev. 3, the missing signals are available so it might be possible to patch them in to the Denise socket - this is done entirely at your own risk, obviously. Find and double-check schematics first!

/CDAC, pin 34 on Denise, is available on Gary, pin 26 - Gary is U5, a 5719 near the center of the motherboard. /CSYNC, pin 32 on Denise, is available on U41, pin 9 - U41 is a DIL-20 74HCT245 just next to Denise.

[kipper2k]

k, its working, i need to tidy it up and then i will show some pics etc

[kipper2k]

using this... /CDAC, pin 34 on Denise, is available on Gary, pin 26 - Gary is U5, a 5719 near the center of the motherboard. /CSYNC, pin 32 on Denise, is available on U41, pin 9 - U41 is a DIL-20 74HCT245 just next to Denise.

I followed traces to find better places to connect a pin header to.

the CDAC signal is tied into pin 26 of Gary, if you follow the trace down you will see there is a via just left of the capacitor, that is where i added the pin header. Same idea for the sync signal, i just followed it until i found a via to add a header there too.

The adapter board could be updated to break out pin 32 and pin 34 to a 2 pin header on the right side to allow an easier install on a Rev3 A500. Another option would be to remove the motherboard and add the 2 wires on the underside, then no changes to the adater board would be needed

[hansliss]

For the A2000-A (rev. 4), and perhaps even A1000, it may be possible to do the same thing by patching U77 pin 8 to Denise pin 32, and pin 15 on the MMU slot through an inverter to Denise pin 34. I may test this after I've verified my adapters on an A500.

[kipper2k]

Here is a picture of the underside of the Rev3 A500. The yellow wires show what is needed to be done to get the HDMI working. Doing this means you do not need to modify the HDMI adapter.

Ignore the black wire mod, that was on this board already

[hansliss]

For the A2000-A (rev. 4), and perhaps even A1000, it may be possible to do the same thing by patching U77 pin 8 to Denise pin 32, and pin 15 on the MMU slot through an inverter to Denise pin 34. I may test this after I've verified my adapters on an A500.

I got it working in my A2000-A! Instead of the MMU slot, I took CDAC from RN17 pin 3, through a 74*04 inverter and to pin 34 on Denise. My setup was a bit too precarious to keep (with a caseless PSU and cables everywhere) so I'll experiment further later, but it looked like a perfect picture.

There's a bit of a mess with the hot glue but I didn't want to risk anything coming loose. I'll add some Kapton tape on the case below this kludge just in case.

[kipper2k]

awesome, my a1000 is an NTSC version and i think it is different, the schematics are very poorly photocopied and i am having a hard time reading it. From what i can tell sync is on U7a pin 2 and cdac pin 15 of the 86 pin header

[hansliss]

awesome, my a1000 is an NTSC version and i think it is different, the schematics are very poorly photocopied and i am having a hard time reading it. From what i can tell sync is on U7a pin 2 and cdac pin 15 of the 86 pin header

Can you somehow send me the schematics you have, so I can compare them to the various schematics I've collected?

[kipper2k]

here is a link...

www.kipper2k.com/a1000.zip

[hansliss]

here is a link...

www.kipper2k.com/a1000.zip

Ah, yes. I saw that one. One of the worst ones.

You can find the uninverted CDAC on R107, on the side not connected via a nearby ferrite bead to pin 8 on U91 (looks like it might be a 74S51), and also of course on pin 15 on the expansion connector. It doesn't seem to go anywhere else.

/CSY is on Agnus pin 39 and on U6A (HC244) pin 8, which should be near Denise. The pin number is hard to read, but input pin 8 on an HC244 corresponds to output pin 12, so it seems very likely it's an "8".

[hansliss]

/CSY is on Agnus pin 39 and on U6A (HC244) pin 8, which should be near Denise. The pin number is hard to read, but input pin 8 on an HC244 corresponds to output pin 12, so it seems very likely it's an "8".

I think you might be right about pin 2 on U7A as well, although that one is buffered through U6A. Not entirely sure what that label on pin 12 on U6A is supposed to be. :)

[kipper2k]

thanks for lookin, i ill try it out tomorrow

[hansliss]

You can find the uninverted CDAC on R107, on the side not connected via a nearby ferrite bead to pin 8 on U91 (looks like it might be a 74S51), and also of course on pin 15 on the expansion connector. It doesn't seem to go anywhere else.

/CSY is on Agnus pin 39 and on U6A (HC244) pin 8, which should be near Denise. The pin number is hard to read, but input pin 8 on an HC244 corresponds to output pin 12, so it seems very likely it's an "8".

I have to add a caveat here: This is unlikely to apply to a PAL A1000. On mine, R107 is connected to a nearby capacitor, and I can't see any trace of that on any of the A1000 schematics I've found so far. I can only conclude that the PAL A1000 has completely different schematics. The motherboards looks different too - there's a continuous row of 18 resistors on the bottom right (plus one capacitor and another resistor). My motherboard is made in Japan, part number 252277-01, from 86.08.08.

[hansliss]

You can find the uninverted CDAC on R107, on the side not connected via a nearby ferrite bead to pin 8 on U91 (looks like it might be a 74S51), and also of course on pin 15 on the expansion connector. It doesn't seem to go anywhere else. /CSY is on Agnus pin 39 and on U6A (HC244) pin 8, which should be near Denise. The pin number is hard to read, but input pin 8 on an HC244 corresponds to output pin 12, so it seems very likely it's an "8".

I have to add a caveat here: This is unlikely to apply to a PAL A1000. On mine, R107 is connected to a nearby capacitor, and I can't see any trace of that on any of the A1000 schematics I've found so far. I can only conclude that the PAL A1000 has completely different schematics. The motherboards looks different too - there's a continuous row of 18 resistors on the bottom right (plus one capacitor and another resistor). My motherboard is made in Japan, part number 252277-01, from 86.08.08.

So, I've traced this elusive CDAC signal on my PAL A1000, 252277-01 from 86.08.08: It originates from a 74S51 under the metal cover near the top, U3M, pin 3, which is connected to a ferrite core nearby (it's just opposite U8K pin 3/4), and at this point the signal is called (probably, the schematics are almost useless) "DAC". The actual CDAC signal can be picked up after this passes through R97, which is located three rows of chips below the middle, next to U3M, A 74F74. CDAC is on the top pin of this resistor, just opposite pin 7 of U3M. After this, it goes directly to pin 15 on the Zorro bus, and nowhere else.

The DAC signal, which can be found at the bottom pin of the same resistor, goes into the nearby U3M pin 3, and also to U2G pin 11, a 74F374 which is much closer to Denise. I'm not competent to judge whether this signal will work as well or if it has to be the actual CDAC. In any case, it'll still have to be inverted, of course.

The main problem, as it turns out, is that there's no room in the A1000 for the RGBToHDMI anyway. Denise is just next to the PSU, and the Pi would need to be where the PSU is. So I'm just going to reassemble this A1000 now and forget about it. Maybe the A600 solution could work. I haven't looked at that.

[hansliss]

I tried this now in my PAL A1000 (252277-01 from 86.08.08), with Dupont cables between the Denise adapter and the Pi since the Pi won't fit next to the PSU, and it kinda worked. I took CDAC from the resistor near U3M and inverted it with a 74LS04, and I took /CSY from pin 8 on U6A. Unfortunately, the picture is quite glitchy, but with those long Dupont cables, I'm not very surprised.

I guess next step is to try what I should have done in the first place: try to move the PSU out of the way and see if putting the Pi directly on the adapter will eliminate the glitches. If they don't go away, there's still something wrong with the CDAC and /CSY signals.

[hansliss]

I tried this now in my PAL A1000 (252277-01 from 86.08.08), with Dupont cables between the Denise adapter and the Pi since the Pi won't fit next to the PSU, and it kinda worked. I took CDAC from the resistor near U3M and inverted it with a 74LS04, and I took /CSY from pin 8 on U6A. Unfortunately, the picture is quite glitchy, but with those long Dupont cables, I'm not very surprised.

I guess next step is to try what I should have done in the first place: try to move the PSU out of the way and see if putting the Pi directly on the adapter will eliminate the glitches. If they don't go away, there's still something wrong with the CDAC and /CSY signals.

I've tried this now, and the picture is still not clean. I'll move the patches to the underside of the motherboard tomorrow.

[c0pperdragon]

Actually, I think inverting the _CDAC may not be necessary. Having the wrong phase may cause the hsync detection to be come unreliably (causing horizontal jitter) and the horizontal pixel alignment will be off by just one pixel on the output screen. Both can be fixed by adjusting the settings in the RGBtoHDMI software on the Pi.

[c0pperdragon]

The glitches in your picture look very much like the ones I had when I wired up everything on breadboards. Your cables are just too long for clean transmission.

[hansliss]

Actually, it was a bad solder joint somewhere on the adapter. After moving the patches to the motherboard, I tried a known good adapter and it worked, so I reheated all the SMD solder points on this one, and after that it worked, too.

When I do these patches on the motherboard itself, I prefer to do it "right" and produce the proper signals, in case I want to switch to a Super Denise later or something.

I can still see a couple of red pixels on the Kickstart screen on the screenshot, which I didn't see on the actual screen. I'll look into that. But overall, it works just fine.

[hansliss]

I can still see a couple of red pixels on the Kickstart screen on the screenshot, which I didn't see on the actual screen. I'll look into that. But overall, it works just fine.

The A1000 RGBToHDMI image has some flickering on vertical lines (specifically on the black-to-blue transition on the Kickstart screen) even after I moved the CDAC and CSYNC patches to the motherboard. Two different Denise adapters produced the same symptoms. The MB patches are now very similar to the ones on the A2000, but on the A2000 the image appears perfect, with no artifacts. I wonder if there's something else in the motherboard design that produces bad signals.

In any case, I've reassembled and set the A1000 aside for now. I'm not installing RGBToHDMI in it at this point.

[SirCathal]

@hansliss Du you made any progression? because I will try the Adapter in my Pal A1000 at the Weekend. Can you post a picture where I can see the exact positions of the two Wires on the Mainboard and on the Adapter?

[hansliss]

@hansliss Du you made any progression? because I will try the Adapter in my Pal A1000 at the Weekend. Can you post a picture where I can see the exact positions of the two Wires on the Mainboard and on the Adapter?

Sorry, the A1000 is back together for now, and I haven't even ordered the new, smaller PCBs yet, which is the next step for the A1000.

But my descriptions above should be enough for you to find the correct points. Apparently you don't need to bother inverting CDAC if you're only using it for RGBToHDMI.

[hansliss]

@hansliss Du you made any progression? because I will try the Adapter in my Pal A1000 at the Weekend. Can you post a picture where I can see the exact positions of the two Wires on the Mainboard and on the Adapter?

Note that the patches I made are for a specific version of the A1000, one without the daughterboard covering half the motherboard. I believe it is newer than the ones with the daughterboard.

[BattleR0ntti]

For the A2000-A (rev. 4), and perhaps even A1000, it may be possible to do the same thing by patching U77 pin 8 to Denise pin 32, and pin 15 on the MMU slot through an inverter to Denise pin 34. I may test this after I've verified my adapters on an A500.

I got it working in my A2000-A! Instead of the MMU slot, I took CDAC from RN17 pin 3, through a 74*04 inverter and to pin 34 on Denise. My setup was a bit too precarious to keep (with a caseless PSU and cables everywhere) so I'll experiment further later, but it looked like a perfect picture.

You wouldn't happen to have the partnumber for the inverter? This looks like the solution needed for my old rev.4 board, but there seems to be a lot of 74*04 inverters out there.

[hansliss]

For the A2000-A (rev. 4), and perhaps even A1000, it may be possible to do the same thing by patching U77 pin 8 to Denise pin 32, and pin 15 on the MMU slot through an inverter to Denise pin 34. I may test this after I've verified my adapters on an A500.

I got it working in my A2000-A! Instead of the MMU slot, I took CDAC from RN17 pin 3, through a 74*04 inverter and to pin 34 on Denise. My setup was a bit too precarious to keep (with a caseless PSU and cables everywhere) so I'll experiment further later, but it looked like a perfect picture.

You wouldn't happen to have the partnumber for the inverter? This looks like the solution needed for my old rev.4 board, but there seems to be a lot of 74*04 inverters out there.

Oh dear. It's upside down on the underside of the PCB and the Amiga is fully assembled... :) I'm pretty sure it was an old LS I had lying around.

[hansliss]

You wouldn't happen to have the partnumber for the inverter? This looks like the solution needed for my old rev.4 board, but there seems to be a lot of 74*04 inverters out there.

Actually, you may not need to invert the signal, according to @c0pperdragon. Much easier if you don't have to do that.

[SirCathal]

@hansliss Today I tried to get the adapter to work in my Pal A1000. Unfortunately, I wasn't quite successful yet. I put an OCS Denise in the adapter and set the adapter to ECS, as this should be necessary in the A1000 (without that, I simply got no picture). Then I soldered a cable from pin 8 on UA6 to pin 32 of the Denise socket under the moterboard. (The Picture was taken before I soldered the cable under the board)

The result is an image with individual white flickering pixels and cyan shadows next to black pixels.

https://user-images.githubusercontent.com/69156692/110226165-3d8e0a80-7eed-11eb-8280-889ec5aa7cec.mp4

I've tested several adapters and PIs. the combination works perfectly in an A500. Do I now have to bring a signal to pin 34 of Denise? And do I understand correctly that this comes from Pin15 of the expansion port? Can it be that the A1000 generally has timing problems with the adapter? I placed it above the power supply, but to rule out that the tower of GPIO extenders is the reason for the error, I also connected everything directly. But that didn't bring any improvement.

[hansliss]

Do I now have to bring a signal to pin 34 of Denise? And do I understand correctly that this comes from Pin15 of the expansion port?

Yes, you need to bring CDAC to pin 34 as well, inverted or not. If you have a 252277-01 motherboard, read my posts earlier in this thread to see where to find it. And move the jumper back to "Denise".

I still have small defects in the picture that I don't have on other Amigas, but I'm going to try the suggestion given in #36 and use a different kind of 74* ICs for the A1000.

[kipper2k]

I designed a new board so you wont need to use all those risers and make the connections easier. will be a couple of weeks

[SirCathal]

I designed a new board so you wont need to use all those risers and make the connections easier. will be a couple of weeks

Please let me know when it's ready!

[BattleR0ntti]

Actually, you may not need to invert the signal, according to @c0pperdragon. Much easier if you don't have to do that.

Thank you @hansliss !! I followed your instructions to the letter and it works! I opted to leave out the inverter and just used the two jumper wires for this and the result was a crystal clear image. Without this little fix the adapter just produces black image, but now it's great.

I did notice a little weird behavior on some Workbench icons when using ClassicWB 68k though. On some icons there are some red pixels vibrating back and forth/up and down, but I'm not sure why this is. These little pixel vibrations are not showing up when using Diropus, shell or even during some games, just the classicwb icons. This does not happen when booting form a Gotek and using plain WB1.3 or 3.1, so maybe it's just a ClassicWB icon problem.

Anyway, thank you again! :)

[hansliss]

Actually, you may not need to invert the signal, according to @c0pperdragon. Much easier if you don't have to do that.

Thank you @hansliss !! I followed your instructions to the letter and it works! I opted to leave out the inverter and just used the two jumper wires for this and the result was a crystal clear image. Without this little fix the adapter just produces black image, but now it's great.

Nice!

I did notice a little weird behavior on some Workbench icons when using ClassicWB 68k though. On some icons there are some red pixels vibrating back and forth/up and down, but I'm not sure why this is. These little pixel vibrations are not showing up when using Diropus, shell or even during some games, just the classicwb icons. This does not happen when booting form a Gotek and using plain WB1.3 or 3.1, so maybe it's just a ClassicWB icon problem.

This sounds a bit like what I noticed on my A1000. I'm awaiting PCBs for a test version of the adapter, on which I will use 74VHC* parts for that one, and see if it goes away. It's not exactly annoying though, just a minor thing.

[SirCathal]

I'm a little further, but there is still one problem. But I'll start over ...

I have pin15 from the expansion port, where the CDAC signal I miss is very far away from Denise. looked where the track connected to pin 15 runs so long. Then I found a via in the middle of the board, where you can get the signal from above as well as from below.

Here on the top of the board (roughly in the middle)

Here on the underside of the board (roughly in the middle)

That then brought a certain degree of success. Because the purple shadows have completely disappeared. What remains are the white, flickering dots. However, these appear primarily in HiRes images, i.e. on the Workbench:

Again better to see in the video: https://user-images.githubusercontent.com/69156692/111081043-a84cd080-8501-11eb-88c4-855efe6dfd3c.mp4

It is hardly noticeable in LoRes mode. Here is an example of a small demo from the good old Red Sector Demo Maker :) Here you only see a white pixel briefly in two places:

In addition, I said goodbye to my GPIO extender and switched to a 5cm ribbon cable. The cable has no effect on the pixel flicker, it doesn't change when I plug the Pi directly onto the adapter. With a 15cm cable it gets worse, so I shortened it to 5cm. The back wall is 3D printed and the original one is packed away intact.

Does anyone have any ideas what I could do against pixel flicker?

Edit: @ian Is it possible to solve the problem with another setting option in the software?

[IanSB]

@SirCathal

Is it possible to solve the problem with another setting option in the software?

No, the main options are to try adding a capacitor to the PiClk signal or replace the 74LVC parts with 74VHC parts as discussed in other threads.

[scrappysphinx]

Sorry this might be a bit off topic but @kipper2k a few years ago you made modified hdmi cables with rca jacks (pic attached below) which you were selling on Amibay, if it were possible to purchase one of these (or maybe if you started making them again) would we be able to route audio through the hdmi too?

[c0pperdragon]

This cable actually looks a bit like an "april fools" device to me, as there is no way to insert analog audio into the digital HDMI data stream without substantial active electronics.

[hansliss]

This cable actually looks a bit like an "april fools" device to me, as there is no way to insert analog audio into the digital HDMI data stream without substantial active electronics.

I thought so first too, but it seems to be made for a specific HDMI interface that routes analog audio through a couple of unused HDMI pins: https://bigbookofamigahardware.com/bboah/product.aspx?id=1956

[philipkonkyl]

Hi I have a similar problem with my amiga 500. I have installed the Pi but I get a black screen with the Denise setting and a stuttered image with "Super Denise" (see picture)

I do get the OSD menu, but no matter what setting I try in the OSD there is no change to the picture.

It does say in the beginning "No Sync" or something like that.

I don't know if my amiga is a rev3 or if it has the sync signals.

I appreciate any help -- I would love to play sensible soccer on my projector :-)

[c0pperdragon]

Looking at the picture, the sync signal is not comming through to the Pi. That you have at least some image at all can only be due to some inductive coupling from the color signal and the csync input. Before doing more guesswork, please check your board revision. You find this information on the right side of the board in the box with some other print, like "Amiga 500" and "Rock Lobster". Should read something like "Rev.5" "Rev.3" or similar.

[philipkonkyl]

Looking at the picture, the sync signal is not comming through to the Pi. That you have at least some image at all can only be due to some inductive coupling from the color signal and the csync input. Before doing more guesswork, please check your board revision. You find this information on the right side of the board in the box with some other print, like "Amiga 500" and "Rock Lobster". Should read something like "Rev.5" "Rev.3" or similar.

My Mother board is a Rock Lobster, but i simply can't seem to find the "rev.X"info

[c0pperdragon]

So, yes this is definitely a Rev.3 board. Further up in this thread are tons of information how to make the adapter work with this. But you do need to do a bit of soldering here.

[philipkonkyl]

So, yes this is definitely a Rev.3 board. Further up in this thread are tons of information how to make the adapter work with this. But you do need to do a bit of soldering here.

OK Thank you so much @c0pperdragon ! I will try the soldering on the bottom. 2 small questions. Should I use Denise or super Denise ? And the Ram you see on one of the pictures is that extension module if yes - how much has been added. Again sorry for my newbie questions :-)

[c0pperdragon]

The picture above from kipper2k with the two yellow botch wires on the underside is exactly what you need. Then you will get the needed signals for the adapter board when jumpered to "Denise".

About your RAM expansion: This is obviously some aftermarket 512KB extension. So your Amiga should have the regular 512KB of chip RAM plus 512K of slow RAM.

[philipkonkyl]

The picture above from kipper2k with the two yellow botch wires on the underside is exactly what you need. Then you will get the needed signals for the adapter board when jumpered to "Denise".

About your RAM expansion: This is obviously some aftermarket 512KB extension. So your Amiga should have the regular 512KB of chip RAM plus 512K of slow RAM.

@c0pperdragon Thank you so much for all the info. !!!

[Bunta714]

@philipkonkyl

Were you able to get the modification working correctly? I have a Rev 3 board that is identical to yours (oscillator crystal above the CPU), and our model is slightly different from the one that kipper2k did the mod to (oscillator crystal below the CPU, I think).

I copied his exactly and I'm still getting only a black screen when switching to the HDMI from the Pi. I'm wondering if maybe I did it incorrectly, or if there is a difference in the boards.