Two more hardware standards Apple play fast 'n' loose with - DVI and Display Port

Another example of a customer splitting an equipment order into bits and not buying the monitors we recommended (saved a full fifty quid on each unit!) has combined with Apple's very poor implementation of 1920x1200 resolution to bit us in the backside.
The newest iteration of MacPro workstations ship with a display card that has a DVI and DisplayPort (with a DP->DVI breakout adapter). Aside from the problem of non-standard blanking as implemented in OS-X's drivers (see my blog entry about Kramer DVI routers) there is a very funny (but consistent effect) if you boot one of those computers with two monitors extended over fibre - you get one display at low-res and the procedure to get two monitors running at 1920x1200 is;

1. Boot the machine with a single monitor connected to the DVI port - increase resolution in increments to 1920x1200 @60hz
2. Reboot
3. Check the resolution sticks.
4. swap the monitor to the Display Port output
5. Reboot
6. Wind up the resolution as per 1. and if OS-X detects the extra monitor turn on display mirroring
7. Reboot
8. If both monitors come back up at 1920x1200 then turn off mirroring and ensure that both monitors are still at 1920x1200
9. Reboot
10. Make sure it's all sticking!

Compare this to the procedure for bringing up a PC-based Avid (running on an HP 8400/8600 workstation, nVidia card);

1. Set both displays for 1920x1200 @60hz

It is so clear that Apple assume you have the machine under your desk and you're using two of their monitors on the pre-made cables they supply. That's not how broadcast facilities are configured and if Apple wants to see FCP used more in film & TV they need to make their implementations of signal standards more robust.

Beware of your DVI's blanking width!

You think you know a certain signal standard but then some nuance of interpretation jumps up and bites you on the backside! Here are excerpts from an email conversation, mine are in italics, the manufacturer's in bold.

....we have a pair of VS-66HDMIs that pass signals fine at resolutions sub 1920x1200 but when you hit that resolution the output flickers and won't lock. Taking the router out of the circuit but using the same cables allows the monitor to lock to the signal fine. The two sources are; Apple MacPro with nVidia GForce GT120 graphics card Apple G5 with ATI Radio 9600 Pro graphics card. Suffice to say we've done the usual powering down etc and tried different cables in case something is on the hairy edge of spec but with different sources that seems unlikely.

Thanks for your e-mail, the details of which I sent off to our R&D people. I have now had a response, in which they said;

"When we mention 1900x1200 we mean narrow blanking. With regular 1900x1200 blanking the bandwidth is higher than the chipset capabilities."

When I asked if the VS-66HDCP matrix (6 x 6 DVI matrix) was any different the answer came back;

"They are the same and use the same chipset. You may be able to set the PC's output to narrow blanking."

So I guess the next question is whether you can adjust the PC’s to provide narrow blanking.

OK, thanks for looking into it xxxx. I had a chat with Apple Engineering late yesterday (we’re a re-seller) and they say that no stock graphics card that has shipped with either a G5 or MacPro in the last five years supports narrow DVI blanking at 1920x1200! In the case of nVidia cards that setting is exposed in the PC driver (I checked on a couple of Windows machines and that is indeed the case – although the installed default was standard rather than narrow blanking) but there is no way to get to it with OS-X.

It seems you should change your advertising to read ‘doesn’t support Macs at hi-res’ or ‘not for use with Macs & monitors greater than 23” display’ something like that. Given that an awful lot of people in the creative industries use Apple computers this isn’t an unusual requirement of a product and if someone reads the copy “Up to UXGA, 1920x1200, 1080p.” You’d forgive them for assuming that it will work with their Apple computer running at 1920x1200.

It puts us in a spot as we now have to source another pair of switchers to sort out our customer.

So there you go - I'm pinning my hopes on the Gefen equivelent. They seem to recognise that a lot of people in the TV industry use Macs!

Tektronix vs Sony

This afternoon it was a stand-off between two giants of the broadcast hardware world - in the red corner was the test and measurement colossus Tektronix with their WVR7100 video waveform rasteriser. In the blue corner was Sony with their new LMD-2450W monster HiDef monitor.
The problem is that the XGA output of the Tek doesn't drive the XGA input of the Sony. Every other monitor was fine and every other device (well, mine and Simon's laptops) could light up the Sony at 1024x768 @60hz. After much puzzling I called Lee at Tektronix and he confessed that they didn't have a proper XGA timing device, rather than relying on some capacity on an FPGA and so consequently it runs at approximately 60hz - the Sony (it transpires) is very fussy. I spent the afternoon stripping off the syncs using a Procon processing DA and re-inserting it as separate H&V syncs, composite syncs and finally syncs-on-green but to no avail. I'd assumed it was a synchronisation issue.
Anyhow - while truffling about on the web I found this great Tektronix document on DVI test and measurement.
You'd best do a save as on that PDF - for some reason it won't open in the browser

2nd o/p of MacPro display cards

We're exhibiting at Broadcast Live! and whilst setting up the equipment I had real difficulty getting a MacPro to feed a 23" cinema display - actually because it was a 7.5m DVI cable I'd stuck a DVI DA just before the monitor (Apple displays are very fussy about long cables - much more so than Dell/HP/Acer etc. - but hey, they're design classics darling so why conform to the standards?!) and would occasionally see a flash of the OS-X logon screen but v.unreliable. I knew the DA was OK having checked it on an HP8400 workstation and it wasn't until Graham reminded me of the problem I had at The Joint back in January (link in the title of this post) that I thought to try output two of the card - all good! It turns out the DVI DA doesn't return correct DDC data and so XFree86 (as implemented in OS-X) does funny things.

Extending DVI over fibre

I've spent the last couple of weeks designing and supervising the install of Optomen Television's new editing facilities in Kentish Town, North London. I decided this time to try extending all the DVI feeds over fibre - I've done it before on more expensive projects but up until now it's been prohibitively expensive for SD-only Avid rooms. However, Lindy are now carrying extenders that look remarkably like the ones I've used in DI projector rooms before. In fact Lindy are sourcing them from the same OEM as others (Gefen for example) but selling them for sensible money. They have proved good - there are a couple of things worth noting;

• Pay attention to the fibre allocation - you need two duplex multi-mode circuits (which we ran in loose-tube cable terminating it in dedicated wallboxes) - on the receiver the inputs are crossed-over WRT the transmitter - they presume a TX-RX cross-over (but the manual makes no mention!).


• The nVidia FX1500 cards that ship with the Avids have both RGB (SVGA) and DVI on their 29-pin DVI-I connectors but the card detects if a monitor is connected and mutes it's other output - so you can't used these splitters as the card doesn't support simultaneous analogue and digital output - bummer!

So I'd recommend these (with the normal warning about doing the fibre properly!)

You down with DDC (Yeah you know me)

Display Data Channel is the way both analogue (i.e. SVGA) and digital (DVI) computer monitors communicate back to their graphics cards what their abilities are - what resolutons they can run at and what refresh rates. Additionally the monitor will send it's name so that the driver can load colour correction lookups etc.
Now, Scene Double (link above) are my favorite KVM extenders - I've banged on about them before (see here) and they now have a very cute trick where the extender can back-propogate the DDC data from the monitor - very useful particularly when extending Mac monitors (where you can't force generic monitors like you can under Windows). Now, Ray Gordon - the lead engineer at Scene Double helped me through a problem today - I had a MacPro being extended over cat5 to a Dell 2407 monitor. The Graphics card (nVidia FX4500 - so no slouch!) refused to drive the monitor at anything above 1920x1080 even though the Dell's native res is 1920x1200. It transpires that OS-X has a problem that relates to the underlying X11 (Apple version of Xfree86) graphics engine which caches the DDC data from the first display on the card. If it has had a sniff of a different monitor (or the same monitor fed over DVI) it won't allow you to set a different resolution/refresh rate and have it stick through a reboot. The answer is on single monitor systems to hang the extender off the 2nd o/p - then it's all good!