G5 Quad and dual 30″/23″ monitor setup: Stretching the hardware to the limit?

Posted by Pierre Igot in: Macintosh
January 5th, 2006 • 5:11 pm

My G5 Quad came with a GeForce 6600 graphics card which, according to the technical specifications, is enough to drive a 30″ display on the first (“dual-link”) DVI port and a 23″ display on the second (“single-link”) DVI port. A good thing too, because that is exactly the dual monitor setup that I wanted.

With my G4, I was using a dual monitor setup with a 23″ monitor as the main display and a 17″ monitor as the secondary display. The ATI card in the G4 MDD had one ADC port and one DVI port. Since both monitors were ADC, I had to get an ADC-to-DVI adapter (made by Apple) to connect the secondary ADC monitor to the DVI port. Unlike an ADC connection, an DVI connection only carries the video signal, which means that the monitor requires its own power supply. (The ADC-to-DVI adapter consists of a power brick with cables going in and coming out. It also provides a USB connection between the monitor and the computer.)

Along with the G5 Quad, I bought a new 30″ monitor. My plan was to use the 30″ monitor as the main display, and the 23″ as the secondary display. With the GeForce 6600, that means that I have to use the ADC-to-DVI adapter to use the 23″ ADC display with the G5’s single-link DVI port.

I set this up as soon as I got the G5 Quad, and it all appeared to be working fine. Unfortunately, over the first few weeks of using the new setup on a daily basis, I noticed a couple of small, yet not insignificant problems that I didn’t have with my previous setup.

The first problem occurs exclusively on my secondary display (the 23″ monitor connected with the ADC-to-DVI adapter), and only within a specific area on the screen (roughly in the top-left quarter, although the actual affected area is smaller than that). The symptoms are the following: All of a sudden, when viewing web pages in Safari, whatever portion of the web page happens to be in that area is covered with brightly coloured pixels, some of which appear to be fluttering uncontrollably.

Obviously, if I take a snapshot of that portion of the screen with a screen capture tool, the end result is a perfectly normal looking picture. So I took a picture of the affected area close up with my digital camera:

Screen artefacts on secondary display

I first noticed the problem with the low-quality JPEGs used by Amazon.com on their product pages, which happened to be in the top-left corner of the page. If I moved the Safari window around, the affected area would not move, and the artefacts would disappear if the portion of the page that was now in the affected area did not contain a picture. In other words, the problem disappeared with only black text on a white background.

Later, I saw that the problem also affected anti-aliased text on a dark grey background and small GIF pictures, as shown in the picture above. I also discovered that I could actually make the problem disappear by moving other background windows around or hiding other background applications. So the problem appears to be related to the overall contents of the screen and to occur only under a number of conditions.

For example, right now, I have a number of background windows open on my secondary display in addition to the foreground Safari window containing the same site as illustrated above. (The picture represents a portion of the TERMIUM web site once I am logged in.) The problem is there, but only if the TERMIUM window is in the foreground. If I click on one of the Interarchy windows that’s currently in the background to bring it to the fore, most of the TERMIUM web page remains visible in the background, including the affected area, and yet the problem disappears. It reappears as soon as I click back on the TERMIUM web page to bring it to the foreground. The problem also disappears altogether if I move the Interarchy windows to the main screen (i.e. remove them from the secondary screen altogether) and bring the TERMIUM web page back to the foreground.

It is rather strange. It’s obviously an intermittent problem that only occurs with a given set of circumstances, and I am still in the process of trying to identify the required circumstances more specifically.

My suspicion is that the problem is related to the overall amount of power required by the screen as a whole. I am no engineer, but I suspect that the amount of power required varies depending on the relative brightness of what is being displayed on the screen. If the overall screen picture contains more bright pixels, the display requires more power. If the overall screen picture contains more dark pixels, the display requires less power. Yet the fact that it only occurs with web pages in Safari (as far as I can tell; I have yet to encounter it with other types of windows) means that there is also a software dimension to the problem.

The reason for my suspicion regarding power issues is that this is not the first time I see this kind of problem on an LCD screen. Back in 2003, when I was trying to find a solution to the horrendous noise problems caused by the G4 MDD, I actually purchased an ADC extension cable from Dr. Bott. My plan was to hide the noisy G4 in a closet. Unfortunately, as soon as I connected the 23″ display to the G4 using that extension cable, I started seeing random dancing coloured pixels on the screen.

After discussing this with an engineer friend, we came to the conclusion that the cable was simply not strong enough to carry the signal for a 23″ screen over such a distance (contrary to Dr. Bott’s claims), possibly because of the loss of power related to the length of the cable. (The extension cable worked fine with the 17″ display.) I ended up returning the cable and getting a refund. I was later told by a Griffin Technology engineer that they were working on an extension cable that would work with the 23″ display, and I was even offered to be sent a test unit at the end of the development process, but this never happened. I suspect that Griffin Technology encountered the same difficulties as the ones I experienced with the Dr. Bott cable.

The other reason why I suspect some kind of power issue is the other problem that I have encountered with my new setup. The new 30″ display comes with two built-in FireWire 400 ports, which are connected to one off the G5’s FireWire ports with the FireWire 400 cable that is part of the connection between the 30″ display and the computer. When using these FireWire 400 ports on the 30″ display to connect a variety of external FireWire drives, I discovered that, with some of the drives, the use of these ports on the display would lead to the appearance on screen of a series of bright coloured pixels scattered all over the screen, this time moving along with the affected elements in the picture.

For example, there would be a few bright coloured pixels in the title bar of a Safari window, and these pixels would stay “attached” to the window when I moved it around. I would also see such pixels when running one of OS X’s built-in screen savers. Here again, the pixels would move around, along with the affected part of the moving picture, and change along with the changing of the picture itself. This is the exact same problem that I had with the ADC extension cable and the 23″ screen.

This problem only occurs when I connect certain external FireWire drives—even though the drives come with their own power supply. I still suspect that the sudden surge in power required when connecting the drives, even if it’s marginal, is related to the problems on the screen.

Fortunately, in that case I can make the problem go away by disconnecting the drives and then putting the monitor to sleep and waking it back up. (Simply disconnecting the drives doesn’t make the problem disappear.) And now I avoid connecting these particular drives to the FireWire ports on the display altogether. (They are older drives that I do not use on a regular basis anyway.)

As far as I am concerned, these two problems are not a deal breaker. I can easily avoid connecting the offending external drives, and the problem on the secondary monitor only occurs under a given set of circumstances. It could also very well be that the problem has to do with the fact that I have an ADC 23″ display, while Apple doesn’t use ADC connections any more and probably only tested the G5 Quad’s graphics card with the newer generations of Apple displays with DVI connections. It wouldn’t be the first time that Apple failed to properly test their products with a sufficiently varied range of hardware (including some of their own products).

However, my ADC 23″ display is still perfectly good (even though it’s not as bright as the 30″ display, as noted earlier), and I am not about to replace it with a DVI model just for the fun of it! I am a bit disappointed that all this expensive equipment doesn’t work perfectly right, and I have submitted the problem as a bug report to Apple and also written to the local Apple Education representative through whom I bought the system.

We will see what they have to say about it.


Comments are closed.

Leave a Reply

Comments are closed.