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

[MacTester57]

You could try the following:
-close the "gap" in your video cable shielding (between the weaved cable shields & the DVI connector) with copper tape (as kiwi suggested)
-Wrap the 12 & 24V supply cables around ferrite rings
-bundle up your cables as clean as possible with zip ties (separate signal and supply wires)
-ensure, that you have a straight and stable GND level for all your PSU's (no breadboard connections & the wires must be thick enough - ca. 1mm2 per 10A current)
-The faraday cage in the dome must be properly grounded to your wall outlet

It could be, that your three PSU's influence each other and cause some instability in the supply wires, because they all are switching power supplies.

good luck!

MacTester

 

[ersterhernd]

You could try the following:
-close the "gap" in your video cable shielding (between the weaved cable shields & the DVI connector) with copper tape (as kiwi suggested)
-Wrap the 12 & 24V supply cables around ferrite rings
-bundle up your cables as clean as possible with zip ties (separate signal and supply wires)
-ensure, that you have a straight and stable GND level for all your PSU's (no breadboard connections & the wires must be thick enough - ca. 1mm2 per 10A current)
-The faraday cage in the dome must be properly grounded to your wall outlet

It could be, that your three PSU's influence each other and cause some instability in the supply wires, because they all are switching power supplies.

good luck!

MacTester

MacTester, what would I do without you?

I did all of the above, with the exception (so far) of the copper tape, as I have to order that if needed.

In no order :

1. Ran a thick 14 gauge GND wire for the video cables. Soldered it direct to molex GND from the PicoPSU.

2. Tightly 'twist tied' a heavy wire around the mesh shielding of both the grey and black LCD wires. Soldered them that way, for 360 degree contact around each wire's mesh shield. Had to be wary of the heat, didn't want to damage the tiny internal wires in each cable. Ran both of these to the DVI GND pins.

3. Put ferrite rings on all 12V, 24V and USB lines. There's two rings on the 12V LCD supply wire.

4. Cleanly re-routed all voltage supply wires with zip ties away from the LCD cables.



The result: A CLEAN VIDEO SIGNAL!!!!


Thanks so much for the advice to both Kiwi and MacTester. EMI interference noise was a huge issue with this system, far beyond the two 17" iMac G4's that I build last year. Fortunately, with some excellent advice, I got it solved.


Ersterhernd

 

[MacTester57]

MacTester, what would I do without you?
...
Thanks so much for the advice to both Kiwi and MacTester.
...

You're welcome. I'm glad that it works now!

Do you have some actual pics?

MacTester

 

[ersterhernd]

Hi Mactester, pics are coming today.


Cheers!

Ersterhernd

 

[ersterhernd]

Build Log 15 -- SSD Installation

One of the great features of the new Haswell NUC is the 6.0 gb SATA port. I took advantage of this feature by way of an Intel 530 series SSD. It fit like a glove underneath the iMac. I'm currently running a Corsair mSATA as well, for total storage of 240 gb.

The bottom Apple cover plate still fits perfectly too.

Ersterhernd

 

[kiwisincebirth]

The result: A CLEAN VIDEO SIGNAL!!!!

Thanks so much for the advice to both Kiwi and MacTester. EMI interference noise was a huge issue with this system, far beyond the two 17" iMac G4's that I build last year. Fortunately, with some excellent advice, I got it solved.

Ersterhernd

Glad you solved issues. I suspect the reason why you had this difficulty is the amount of data being sent to the panel is significantly more.

The 20" LCD has 72% more pixels (compared with 17"). It is actually 81% more if you consider the 20" panel is 24 bit colour (on iMac G5 at least), only 18 bit on 17" panel.

Extra data means higher bit rates, and higher bit rates means higher frequencies, higher frequencies means each bit of data is smaller and more susceptible to interference.

I hope (when I get started) to avoid similar issues. Of course my build is quite different, but your experiance has highlighted how critical this is. PS Work on my build hasn't completely stopped, hope to be back underway in week or two.

Is that a photo of the bottom of a completed G4 build?

Kiwi

 

[ersterhernd]

Glad you solved issues. I suspect the reason why you had this difficulty is the amount of data being sent to the panel is significantly more.

The 20" LCD has 72% more pixels (compared with 17"). It is actually 81% more if you consider the 20" panel is 24 bit colour (on iMac G5 at least), only 18 bit on 17" panel.

Extra data means higher bit rates, and higher bit rates means higher frequencies, higher frequencies means each bit of data is smaller and more susceptible to interference.

I hope (when I get started) to avoid similar issues. Of course my build is quite different, but your experiance has highlighted how critical this is. PS Work on my build hasn't completely stopped, hope to be back underway in week or two.

Is that a photo of the bottom of a completed G4 build?

Kiwi

Hi Kiwi, I guess I spoke too soon. YouTube video kicks off the artifacting within a few minutes. Damn.

Going to dig into this deeper I guess. A solution lies somewhere, but this is getting to be quite taxing on the ol' brain. You are very correct when you mention the word critical. The integrity and viability of the entire build rests on solving this one issue.

I wish you luck with yours. Hopefully it works out easier than mine, I've spent nearly ten days on this issue now.


EDIT: From a cold boot, the system will run error-free for hours in normal use, as long as no YouTube vids are played. Once the video playback begins, in about 10-15 mins the artifacting starts. Perhaps noise is 'building up' in that timeframe due to the increased video bandwidth of the playback.


Cheers!

 

[MacTester57]

The SSD installation looks good.

Hi Kiwi, I guess I spoke too soon. YouTube video kicks off the artifacting within a few minutes. Damn.

Oh no, that sucks!

What the h... ?? My thoughts:
-Add the copper tape
-wrap more than one winding around the ferrite rings on your supply & GND wires (about 5 windings)
-shorten all your supply wires as much as possible
-are all the TMDS pins properly plugged to the pins on your DVI connector and don't wiggle?
-are all red / green TMDS "couples" twisted as good as possible, before they are plugged to the DVI pins?
-because symmetrical signals are not affected by ferrite rings, you could try to put all your DMDS wires through a ferrite ring... (all pairs of them through the same ring)

Are the artifacts on the entire screen or all in the same "sector" of the screen? This would mean, that one of your TMDS channels does not work properly.

Is the iMac screen properly detected in "About This Mac" > Graphics/Monitor? You could install SwichResX and export your EDID (post it here if you want). Then we can see, if your EDID is read properly from your graphics driver...

Is the noise still present in your internal mic recordings?

MacTester

Edit:

EDIT: From a cold boot, the system will run error-free for hours in normal use, as long as no YouTube vids are played. Once the video playback begins, in about 10-15 mins the artifacting starts. Perhaps noise is 'building up' in that timeframe due to the increased video bandwidth of the playback.

Do VLC videos work?? or are the artifacts only in YT videos?

 

[ersterhernd]

MacTester I think I solved it this time. Turns out my hardware was just fine as is.

I discovered that pre-Snow Leopard, OSX shipped with a default gamma setting of 1.8, not 2.2 as is now. I recalibrated a new setting by changing the Generic RGB profile to 1.8 and saved. Haven't had an artifact (HD videos included) now for over two hours.

I guess this old screen, as nice as it is, just wasn't robust enough to push the richer 2.2 gamma setting out of the NUC HD 5000 graphics. I seriously doubt that any amount of hardware tweaking was going to fix that. The touch CCFL dimming that is incorporated makes adjustment a snap dependent on ambient light conditions.

The screen (fonts especially) is a little less vibrant, but not noticeable. I'm happy to say that this is beginning to appear like a successful build. I hope I don't have to post about any further trouble with this issue, its really beat me to death trying to solve it, but for the time being persistence appears to have paid off.


Cheers!

Ersterhernd

 

[kiwisincebirth]

Hi Kiwi, I guess I spoke too soon. YouTube video kicks off the artifacting within a few minutes. Damn. I wish you luck with yours. Hopefully it works out easier than mine, I've spent nearly ten days on this issue now.

EDIT: From a cold boot, the system will run error-free for hours in normal use, as long as no YouTube vids are played. Once the video playback begins, in about 10-15 mins the artifacting starts. Perhaps noise is 'building up' in that timeframe due to the increased video bandwidth of the playback.

Cheers!

I agree with MacTester, it sucks, and I am very curious about video played via other applications causes the same issue ? I can offer one other thought/suggestion/suspicion.

Maybe the issue is caused not by EMI, but by a noisy power supply that transmits the interference along the DC wires that power the LCD. From my understanding you have two power sources for the LCD.

The up converter 24V provided to LCD backlight inverter, it would seem unlikely this would be the issue since this would cause backlight issues, which you don't have.

The 12V/5V (?) from Pico PSU the provides power to the LCD logic board. It is possible that noise is affecting the LCD's electronics, causing the issue. A Pico PSU is normally plugged into a motherboard, and the motherboard typically has a lot on onboard power filtering, to protect itself, so maybe the PICO (built to a budget) is less stringent about the quality of its power.

Remember in a un-modified G4, the LCD is powered from the G4 motherboard, which may provide additional filtering, that you now don't have as going directly to the Pico PSU.

Remember these are only guesses.

Suggestion.

Provide some filter capacitors to the power cables that go to the LCD, either at the output of the Pico, or where you join to neck wires. You need a large electrolytic (for low frequency), and a small 0.1uf ceramic capacitor for high frequencies.

Note:

My understanding of ferrite cores is they are used to eliminate EMI induced into wires, not sure if the eliminate noise already present in the wire. The other thing is ferrite cores are rated by frequency so may be ineffectual depending on the frequency of the noise. Again just my opinion.

Good Luck
Kiwi