Barco DP2K 15C Intermittent Boot Up

[Guy Walker]

Good evening all, and thanks for the previous help and advice on our shoestring DCP set up. Extremely helpful. Our rig consists of a Doremi DCP2000 and Barco DP2K 15C and we've undertaken lots of work to get it really purring along. We've overcome quite a few technical hurdles and I think we're nearly there...but...we've got an intermittent problem. The majority of the time (3 times out of 4, say?) the Barco boots up smoothly and we get a steady green tail light and we can connect with Communicator. All diagnostic tests come back green. But sometimes it gets stuck: tailight flashes green, blister panel on the Barco remains unlit, cannot connect via Communicator. Actions taken so far: reseated and cleaned all the cards in the main cage; reseated and cleaned all the back plane connectors; tried a spare ICP card (booted up but no certificate). Has anyone encountered a similar issue and have any recommendations? Diagnostic package attached for interest...

Thanks again

[Frank Cox]

Heat? Got enough ventilation going through the lamphouse?

[Steve Guttag]

Normally, when a Barco S2 takes an extended time to boot up (more than 5 minutes, it is missing something. Often, it is the SDI (or IMB) board. In fact, the projector will boot up, eventually, without them and give a green tail light. If it does come up, after a period of time, if you go to Diagnostics, Tests, Video Path, you'll likely find that the SDI board appears missing.

Likely, your long-term solution is to change the signal backplane. The connectors on that thing get into a state where even cleaning/deoxit will not overcome it.

When you apply DeOxit (either D5 or G5) to the card's gold fingers, do you let it "cure" (haze over)? If not, please do so.

[Leo Enticknap]

I'm thinking CCB or, +1 to Steve, the backplane. From the projector log:

9215=Jan 1 00:00:10 localhost user.warn kernel: Page-cache hash table entries: 16384 (order: 4, 65536 bytes)
9216=Jan 1 00:00:10 localhost user.warn kernel: POSIX conformance testing by UNIFIX
9217=Jan 1 00:00:10 localhost user.warn kernel: CODETEST: PCI device not located
9218=Jan 1 00:00:10 localhost user.info kernel: PCI: Probing PCI hardware
9219=Jan 1 00:00:10 localhost user.err kernel: PCI: Cannot allocate resource region 0 of device 00:00.0
9220=Jan 1 00:00:10 localhost user.err kernel: PCI: Cannot allocate resource region 0 of device 00:01.0​

There then follows dozens of repeated "Cannot allocate resource region" errors, until it finally moves on to other stages in the boot sequence. There are many other previous instances of the sequence, too. The cards in the cage connect to the backplane via PCI connectors, so I'm wondering if a backplane fault is causing "device not located." Note: my understanding is that during the initial boot, log entries are timestamped Jan 1 / midnight before the ICP has finished booting, during which time the CCB cannot read the secure time from it.

The only other thing I noticed was this:

9359=Jul 13 15:53:48 localhost cron.warn crond[79]: time disparity of 28681432 minutes detected

That's over 54 years! I'm not sure what two clocks the disparity is between. The ICP's clock is clearly OK, because the timestamp is now correct (July 13). Maybe the clock in the Enigma card is way out, and this might have something to do with the extended boot time?

If there has been data rot in any of the flash memory components that hold software and firmware code, manually force reflashing every component of the 2.10.121 bundle might help.

[Marco Giustini]

I always wonder why my PC can stay on 10 hours a day for 10 years NOT requiring a card re-seat and a Barco Projector needs a new backplane every now and then...
Fai enough, my PC doesn't have 45 fans shaking the hell out of it. Still...

[Steve Guttag]

I blame whatever card edge connector Barco is using on the signal backplane. NEC and Christie use the same ICP card and while they too do get to a "reseat" point...it is, seriously, something like 100:1 on the ratio with Barco. And, eventually, a signal backplane is needed because the edge connector, even after reseat doesn't do the trick.

I'm sure one of the touchy thing is the very low-current combined with the bazillions of connections on the ICP, in particular that makes this a higher failure point.

Next, the entire Barco cinema projector install base is but a sample size to a laptop production run. If there was this sort of issue on a consumer good that could have millions to billions of units, they'd have to address it (with the manufacturer of the component as the liability would be a fortune. Barco (and other cinema companies) just have to make it 2-3 years before they can straight faced tell you to buy a new signal backplane. Then again, I'm sure Barco is going to hurt massively on the TI TRP style DMD debacle...even if they are getting something out of Texas Instruments for it (and they should), it is still going to hurt.

[Frank Cox]

TI TRP style DMD debacle.

I don't have a Barco projector so out of general snoopiness:

What's this?

[Leo Enticknap]

I don't know what TRP stands for, but DMD is digital micromirror device: the tiny mirrors on the imaging chip that sits on each of the three formatters in the light engine. I suspect that TRP is a more formal name for what was informally known as "tilt 'n swivel." When the Barco SP4K line was launched in 2019, it used a new Texas Instruments DMD that enabled 4K to be crammed onto a .98" DMD chip (Series 2 4K projectors had to use a 1.38" chip, which was not only more expensive, but required more expensive components elsewhere in the projector to support it) by using mirrors that not only tilted to reflect the light from the lamp or lasers, or not, but were able to rotate as well. This allowed each individual mirror to be closer to its neighbor, and also increased the dynamic range possible.

The problem is that these "tilt 'n swivel" DMDs had an unacceptably high failure rate, resulting in dead pixels (usually green). I had to replace the light engines in most of the SP4Ks I installed in the two years or so after launch, and in some cases, replace them multiple times. These projectors were under warranty and so, as Steve notes, Barco took a serious hit, both financially and reputationally. There were also other significant design flaws with the first iteration of the SP4K-C light engine, convergence drift being the big one.

My understanding is that TI has now come up with a .98" 4K chip that is tilt only, and therefore a lot more reliable; and that SP4Ks are now shipping with it.

As for the Series 2 backplane, I'm not going to come down too hard on Barco, because compared to NEC, servicing and repairing them is straightforward. In contrast, there are some fans in an NC900 that can take 4-5 hours to replace. However, the Barco backplane vulnerability is exacerbated by the fact that replacing one is a long, difficult, and fiddly process. IMHO, only the lens holder is a more difficult part swapout. The part is cheap (in the low three figures for an out of warranty projector, the last I knew), but you need to factor at least three hours of Level 2-certified tech labor to do the job: longer if the interior of the projector is very dirty, there are corroded or seized fasteners, etc. etc. Once a projector reaches the age at which it needs a new backplane, you are very likely to encounter this. If a CCB or an ICP fails, for example, a theater manager plus me working remotely can do the replacement. If the backplane borks out, there is no way of avoiding a field service call, which will cost a lot more than the part itself.

[Steve Guttag]

TRP = Tilt-Roll-Pivot. And yes, it is the small (.98") 4K DMD that was used on the SP4K-C and the NEC "ML" series of projectors (so they have the problem too but at nowhere near the sales quantity). The SST series will replace the TRP but the entrance angles are different so, you have an entire optical path to consider when making the change. It will be a nightmare.

As for stuck pixels...I've had all three colors but red is the winner (in a negative sense) for me.

I have not had the convergence drift that Leo mentions though I think the nature of the laser technology and the use of the standard "C" lenses with far worse MTF (transfer function) that isn't as good as the larger "B" lenses, particularly those developed for 4K will play tricks on what you see. Using binoculars/spotting scope will differ from what you see as you go down to the screen. The USL VCC will also differ from what you see down at the screen. Heck, even walking from the back of the room to the front will have one perceive convergence differently. It is freaky as the colors have a 3D-like nature to them. If I have a second tech, the way to converge (quickly too) is to have one adjusting and one watching...you can dial it in pretty fast. With a picture running, they look fine. But...watch out on the edges/corners, particularly on the lesser lenes. I've found running high-contrast lenses improves things. Christie cheats on their Cine-Life plus as they will do electronic warping to fake convergence in the corners.

As to the S2 signal backplane. I can be in and out in just over an hour (unless the projector is unserviceable due to its enclosure). Unfortunately, I've gotten too good through practice. I use the proper torx driver on the screws rather than fiddling with a slotted driver and know which fasteners to remove and when...to leave the bottom card divider (above the SMPS) in as a screw catch until you need to get that last one out...etc. I don't need to take a picture anymore on where the cables go by the fans either.

Yes, NEC is the most miserable to service. They just stack crap on top of crap...have their cables too tight, and use multiple of the same connectors without any color coordination. Barco, conversely, is the best for service...most everything is built on the outside so you merely need to remove a cover or two to get at it. The cables are often color coordinated (with a color zip tie for an identifier, if not outright colored cable/connector) or otherwise uses different sized connectors. They also use a minimum variation of fasteners so you need minimal tools to get at things. Christie is somewhere between the two...depending on the item being serviced.