An Amazing Find for My Next Project -- iMac G4 20"

[kiwisincebirth]

Windows7 with intel hd 5000 driver works flawlessly with the D54250wyk and 20" iMac screen. Must be an OSX issue with the intel graphics.

Hmmmm now what???

Ersterhernd

Just read all your posts on this issue, an I was going to suggest something to try, until I read the above post, so not so sure now my advise makes sense. Well I suggest anyway as it can't hurt.

My suggestion would be to better shield the red/green wires coming from the LCD cable plugging into the DVI socket. To be exact each red/green pair should be twisted together more (rather than be loose), before being plugged into the DVI plug. Secondly each shield wire should then wrap its corresponding red/green pair, before being connected to the DVI pin. I know this sounds tricky to do, but even a small improvement could help

The reason I thought this was that having the wires loose makes it more likely for interference between different pairs of wires, and when you said in an earlier post that "adding a second extension cable fixed the problem", made sense to me, because the additional cable length would reduce the overall signal strength meaning less likelihood of signals jumping across wires.

Just my 2c
Kiwi

 

[ersterhernd]

Just read all your posts on this issue, an I was going to suggest something to try, until I read the above post, so not so sure now my advise makes sense. Well I suggest anyway as it can't hurt.

My suggestion would be to better shield the red/green wires coming from the LCD cable plugging into the DVI socket. To be exact each red/green pair should be twisted together more (rather than be loose), before being plugged into the DVI plug. Secondly each shield wire should then wrap its corresponding red/green pair, before being connected to the DVI pin. I know this sounds tricky to do, but even a small improvement could help

The reason I thought this was that having the wires loose makes it more likely for interference between different pairs of wires, and when you said in an earlier post that "adding a second extension cable fixed the problem", made sense to me, because the additional cable length would reduce the overall signal strength meaning less likelihood of signals jumping across wires.

Just my 2c
Kiwi

Kiwi you were exactly right. I taped absolutely everything, each set of 3 pins with a small piece of electrical tape for insulation. Took two hours. The real culprits were the vedid 5v / hot plug detect / ground wires on pins 14,15,16. They were cross talking with the reds and greens plugged in around them.

Mystery why only HD 5000 graphics were affected. HD3000, HD4000 and Win7 showed no artifacting whatsoever. Must be a stronger signal from the new NUC. Good to know.

On to good things now. I spent all weekend on this, but found the elusive issue with your help, thanks.


EDIT: Spoke too soon. Artifacts are back. Damn.

Ersterhernd

 

[kiwisincebirth]

Post a photo of the wireing if you can.

 

[ersterhernd]

Hi kiwi. After the last reoccurrence of the artifacting, I totally disassembled the wiring.

I removed the old solder and resoldered the 12v, GND and also the vedid/hot plug detect loop with a new 1k resistor. I ensured that the joints used as little solder as possible for minimal resistance.

Next, I did what you said and carefully twisted each green/red pair of signaling wires as tightly as possible with the corresponding shield wire twisted around each pair. I even went one step further and placed a small piece of electrical tape between each triplet. Finally, I connected the five wires (vedid,GND,hot plug detect,clock,data) with an insulating layer of electrical tape between them and the 12 signaling wires.

It was a long tedious job after a frustrating day, but the results looked positive. In first tests the display was clear of any artifacting. But would it stay that way???

The result, the system ran HD 1080p YouTube video all night last night as we slept. This morning there is still zero corruption on the display. Your suggestion appears to have worked. It appears that the twisted triplets provided enough improvement to clear the issue.

I hope this doesn't change, but with each passing hour of operation, I'm getting more confident that it'll keep working. One thing for sure, the video signaling from the HD5000 is much fussier about the path on which it travels than HD3k and HD4k were, with OSX anyway.


Thanks so much for your help. This part of the build has been a real challenge.



Ersterhernd

 

[MacTester57]

Good to hear, that it works now! Kiwi is right, twisted and shielded wire pairs are always a good idea on high frequency symmetrical signal cables.

MacTester

 

[ersterhernd]

Good to hear, that it works now! Kiwi is right, twisted and shielded wire pairs are always a good idea on high frequency symmetrical signal cables.

MacTester

This is interesting too. When I run the system now with the video wires dangling down near the upconverter (unaffixed from their permanent location), the pixelation comes back. Must be EMI from the upconverter. When put back up top, the video corruption instantly goes away.

The NUC board will be within 2cm of the upconverter, I'd better add a shield plate to the bottom exposed side of it. See the pic below.

Ersterhernd

 

[MacTester57]

Must be EMI from the upconverter

Are the weaved outer shields of your video and inverter cables connected to GND? I did that with a short piece of black wire, which is connected to the outer shield of the DVI connector and to the weaved cable shield (see pic above). If not, then the outer shield will be useless.

I'd better add a shield plate to the bottom exposed side of it. See the pic below.

Yes, I would do that.

MacTester

 

[kiwisincebirth]

Next, I did what you said and carefully twisted each green/red pair of signaling wires as tightly as possible with the corresponding shield wire twisted around each pair. I even went one step further and placed a small piece of electrical tape between each triplet. Finally, I connected the five wires (vedid,GND,hot plug detect,clock,data) with an insulating layer of electrical tape between them and the 12 signaling wires. ...

Ersterhernd

Glad to see your hard work paid off. I suspected the EMI might be the issue. Differential signalling uses very low power, and does rely on the wires bing twisted, or shielded.

It is easy to forget the basics of shielding. My first 8 bit computer came in a plastic (only) case, If you placed an AM radio next you could literally hear your programming running, different programs would make different (and identifiable) noises. Some guy wrote a program to play "Happy Birthday" via any nearby radio.

The other tool at your disposal is to wrap the DVI plug, around the exposed wires with something like this.

http://solutions.3m.com/wps/portal/...cs/Products/Product_Catalog/~?N=7234470&rt=c3

 

[ersterhernd]

Thx for the replies Mactester and Kiwi. Time will tell if I need more shielding or not. It isn't to the point where I can close up the base yet, but that will be the final deciding factor of success.

This 20" project is a very different animal than the two 17's were, that's for sure.


Cheers!

 

[ersterhernd]

Are the weaved outer shields of your video and inverter cables connected to GND?

MacTester

I did that tonight as per your photo, MacTester. It gave an improvement for sure. I had no idea the GND connect was that easy to do.

Loaded elgato eyetv tonight and watched full screen TV on the 20". Not a single artifact.





Cheers!