Film-Tech Forum ARCHIVE: Excessive noise/jitter Barco DP2K-12C

This is topic Excessive noise/jitter Barco DP2K-12C in forum Digital Cinema Forum at Film-Tech Forum ARCHIVE.

To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=16;t=003047

Posted by Joachim Heden (Member # 9738) on 06-14-2017, 03:08 AM:

Dear all,

We have made a warranty claim on a Barco DP2K-12C due to excessive noise/jitter in the dark areas of the image.

We are getting some pushback on this from the Swedish importer / service company Videvox - basically they are saying the projector is within spec.

We have now benchmarked the projector against another DP2K-12C and the result can be seen in this video:

https://vimeo.com/221418661/12a2c514e6

The light engine and the ICMP have both been replaced - no change in the noise/jitter.

Has anyone seen this before? Any other thoughts on this?

Kind regards,

Joachim

Way Creative Films
Sweden

Posted by Carsten Kurz (Member # 5396) on 06-14-2017, 05:09 AM:

Very basic idea - did you try a full clone from a 'good sample' to this projector?

Because, after the ICMP and Lightengine have been swapped - signalwise, this is a new projector. No idea wether powersupplies or EMI could also have this effect.

I guess I would bypass the service company and contact Barco directly about this.

- Carsten

Posted by Dave Macaulay (Member # 813) on 06-14-2017, 06:55 AM:

It's not entirely a new projector signalwise - the signals all go through the original backplane. I have replaced a couple of backplanes to fix image issues, but never have seen this noise problem.
The backplane is not super difficult to replace. Just make sure you move the "sim card" over!

Posted by Magnus Eriksson (Member # 6085) on 06-14-2017, 08:45 AM:

I swapped that ICMP and light engine. And I swapped the back plane without success. And I work at another service company nowadays.
Hi Joachim [Smile]

Strange is that this noise also can be seen in other projector models. I have observed it on:
Barco: DP2K-12C, 15C, 23B and DP4K-19B & 32B
Christie: CP2210 & CP4220
NEC NC1200C
HDSDI/Enigma, IMB, IMS or ICMP in the projector makes no difference.
Choose test pattern "RGB-12bit-Shallow Horizontal Ramp" or "RGB-12bit-Full Horizontal Ramp" (Barco) and look from close distance on the left side of the screen (where gray goes to black).
Since it´s visible on all three brands I guess it´s a TI/DLP thing?
/M.E.

Posted by Joachim Heden (Member # 9738) on 06-14-2017, 09:12 AM:

Sorry, forgot to mention, the backplane has been replaced also - no difference.

Joachim

EDIT:

Hi Magnus,

Yes if you look closely, there is a hint if this noise/jitter in other projectors, but I think it should be obvious that our projector exhibits this FAR more than anything else - video link above.

If we put up a 4% white (very dark gray) on the whole screen, it's like standing in a blizzard.

Clearly, what our projector exhibits can not be considered normal...?

Joachim

Posted by Magnus Eriksson (Member # 6085) on 06-14-2017, 10:06 AM:

Is the light (measured on screen) about the same on both sites? Same PCF used??
What happens if you use the same mcgd file on both? Or use nominal?
/M.E.

Posted by Mark Gulbrandsen (Member # 72) on 06-14-2017, 12:12 PM:

WOW! I have never observed that sort of noise in any S-2 NEC projector and I have quite a few NC-1200 under my wings, but mostly NC-2000's. It really is unacceptable IMHO.

Mark

Posted by Joachim Heden (Member # 9738) on 06-14-2017, 12:56 PM:

Magnus,
Yes measured light is about the same in both cases 14fl -- Also, the camera settings (aperture, iso, shutter angle, framerate) are all identical between the two captured sequences in the split screen video (and no image manipulation has been applied to match the images).
Joachim

EDIT:

Oh, and I should add -- the perceptual difference in real life is GREATER than what is evident from the split screen video.

J

Posted by Marco Giustini (Member # 4544) on 06-14-2017, 01:58 PM:

this is very interesting. I have seen that noise before but not so severe to be honest.
As it's been said, new ICP, engine and backplane should have cleared it. The engine also includes new signal wires.

What about mains, interference from LPS or other devices on the machine? It'd be interesting to know what happens when you get the new projector, please keep us posted.

Posted by Joachim Heden (Member # 9738) on 06-14-2017, 02:04 PM:

Magnus, and anyone else following this thread,

Although the measured light on the screen is the same (give or take) your question got me thinking...

The only thing that is "unusual" about our setup is that the screen is very small, only 4,20m wide (this projector is being used in a post production environment) so we are running it with an 800w lamp dimmed all the way down...

This is all layman's armchair reasoning, but perhaps the projector doesn't "like" to be run at this low of a light output level... if the noise/jitter is somehow constant/nominal independent of the light output a higher light output would "drown it out"?

Does this seem at all plausible?

I guess we could try tomorrow to pop a 1200w lamp in and crank it all the way up, bring the screen levels down by throwing an ND filter in front of the lens and see what that does.

Does this seem like it would be worth a try?

J

Posted by Carsten Kurz (Member # 5396) on 06-14-2017, 07:24 PM:

Well, normally: DLP noise is only strong in dark areas. From that follows, it is usually masked by it's low luminance for our eyes. If you raise the light level on the DLP, the noise luminance will rise as well, making the issue even more visible.

Again, try to do a full backup of this system, full clone from a 'good' system, and apply full clone to the problematic system, just to rule out weird configuration residuals.

But - 'real hardware' can be complex. It could be possible the lamp ballast is doing 'weird' things when running low, causing that noise, and that could vanish when you pull the current up, which would be no contradiction, but simply to opposing effects.

Do you have a luminance meter? Measure peak white when testing different setting.

From the video, it also occurs to me that while the noise level is higher, the noise frequency ('resolution')is lower on the bad projector (around a factor of two, I'd say). Are you sure the grey areas are within the same pixel area and have similar resolution?
Also, the 'good' projector is working on a perforated screen, while the 'bad' one on an unperforated. The perforation also masks the effect somewhat. That said, without being able to weigh in on luminance, camera settings, etc. the noise level on your system does appear excessively high indeed.

What types of lenses do both machines use?

- Carsten

Posted by Joachim Heden (Member # 9738) on 06-15-2017, 01:03 AM:

Dear Carsten,

In the video, both projectors display the same custom test pattern so yes, even though our screen is much smaller, the relative size of the samples are the same (same pixel density). I'm attaching an illustration of the test pattern indicating what the camera is shooting in both samples (please note that I've bumped up the contrast of the test pattern in this illustration for the purpose of clarity here)

So it is indeed strange, that on our projector, the "noise" seem to be "bigger". We will do a quick test this morning increasing the light output to see if this has any effect when the lamp/ballast is running at full capacity.

Kind regards,

Posted by Joachim Heden (Member # 9738) on 06-15-2017, 05:22 AM:

UPDATE:

Increasing the light output so that the lamp/ballast is operating att full power does not make any difference.

We will start looking into all the other suggestions in the thread as they apply, but it seems we might have a lemon on our hands, and the warranty claim should proceed.

Kind regards,

Joachim

Posted by Carsten Kurz (Member # 5396) on 06-15-2017, 06:12 AM:

Hmm. Could you create a test pattern with low greys, and some lines/ single pixel patterns in it? Or place the camera on a tripod and take two pictures of exactly the same area, one with the grey patterns, one with some lines (e.g. Barco Focus pattern)?

I'm not an expert on DLPs, but in general, DLP dark noise is not actually noise, but intentional dithering. This, however, should always take place on the pixel level. It shouldn't combine pixel clusters. So, by overlaying/comparing clearly visible single pixel structures, you should be able to find out wether this is actually single pixel noise/dithering.

The way DLP works, I would think that things like thermal noise etc. should not be able to excite the micro mirrors enough to become visible. They need a hard kick to move.
As such, I would think any noise must be introduced by some intention e.g. from the formatter boards, like the mentioned dithering. But again, I am not an expert, and DLP/formatter power supply or noise on control lines could possibly also cause this.

You can also zoom wide to maximum to enlarge everything as much as possible.

- Carsten

Posted by Steve Guttag (Member # 268) on 06-15-2017, 07:17 AM:

How similar are the software versions in the various test projectors? Newer isn't always better either.

Posted by Dave Macaulay (Member # 813) on 06-15-2017, 07:33 AM:

What is your test pattern? Is it a .png file loaded to the ICP, or is it played from a DCP or via an alt content input?
The ICP patterns do not involve the media block, so it does matter.
Dithering is a possibility but it should not be "lumpy" like in the images.
Is the noise static or dynamic?

Posted by Joachim Heden (Member # 9738) on 06-15-2017, 08:49 AM:

Dave,

For the purpose of the benchmark test in the video link, both projectors were playing the test pattern from a DCP.

On our projector (the bad one) we have also tried the .png into ICP, and external scource (davinci resolve via SDI) -- it all looks exactly like the DCP. No difference.

Also -- all standard internal test patterns of the projector that include dark grays exhibit the same noise/jitter.

J

EDIT:

To answer your other question: The noise is dynamic in every sense --it's constantly moving and on a micro level it appears to be 100% random (and do not appear to be at "pixel level/size). Stepping back to more of a macro level there appears to be some "periodic" element in the spatial character of the noise -- however, that character changes (quite drastically) with the tiniest adjustment to the brightness.

Also, projecting pure black gives us no noise -- zero -- but as soon as we go the tiniest amount over pure black, the noise "switches" on.

J

Posted by Marco Giustini (Member # 4544) on 06-15-2017, 01:40 PM:

you seem to have covered all the bases to be honest. Make sure the SW version is the same between the two machines.

Also remember that applying a full-clone will not wipe the existing configuration, hence you will end up with both machines' macros, lens files etc. Sure, your macro selector will only have the macros from the cloned machine. I just don't like the mess that usually comes out from the cloning process, I usually first wipe all the configuration files and then apply the clone - but this is me of course!

Posted by Magnus Eriksson (Member # 6085) on 06-16-2017, 08:10 AM:

I was thinking this morning (that happens occasionally):
This "bad" 12C, it has been running with ICMP (two different individuals) all the time, right? The good projector you compared with has a "normal" ICP if I remember correctly.
Has the bad 12C ever been tested with a standard ICP? If not it could be worth a try.
/M.E.

Posted by Tia Marklund (Member # 9739) on 06-18-2017, 09:16 AM:

Hi Joakim!
"The only thing that is "unusual" about our setup is that the screen is very small, only 4,20m wide (this projector is being used in a post production environment) so we are running it with an 800w lamp dimmed all the way down..."

When raising the light output, did you try another lamp? For some reason USHIO withdrew the 800W and replaced it with a 900W. Probably not the root cause, though...

Posted by Steve Guttag (Member # 268) on 06-18-2017, 11:09 AM:

The DXL-9BAF is actually an 850-watt lamp.