CurtPalme.com Home Theater Forum

[CIR Engineering]

Lumagen Update

I talked to Patrick today and he feels that he's only days away from releasing a big new FPGA firmware. The new FPGA will include 3D frame sequential output modes that no longer are limited to 810 Vactive lines allowing full 1080 Vactive (also the min Hactive of 1280 will be removed so less will be allowed). This will provide the ability to choose any output resolution between 1080x1920 as long as the total pixels are within the clock limits of your model.

Another HUGE UPDATE with this FPGA is the ability to convert side-by-side 3D into frame-sequential (and frame-packed for 3d digital projectors) output modes. What this means is that 3D sources like DTV, DISH, and cable (depending on market) will now be supported as well as network media players like the Dune and PCH. Side-by-side (sbs) is typically a standard 1080i frame (or 720p sometimes) with the left and right half of the frame containing a separate image for each eye. This works with sat, cable, and any other legacy HDMI/DVI standard (1.0~1.3) because it fits within the allocated bandwidth for a regular 2D station. Essentially each eye gets 1080x960i.

Since 1080i sbs is actually 1080x960 per eye, the new FPGA can handle an output mode that is 1080x960p 120Hz. This mode will show 3D without having to do any scaling on these sources

Pretty sweet if you ask me.

Next on the agenda is top-bottom conversion to frame-sequential.

craigr

[CIR Engineering]

I watched the FIFA 2010 highlights 3D bluray disc last night on the G90. This is a 720p 60Hz frame-packed disc so the output mode for the Lumagen is 720p 120Hz frame-sequential. It looked really fantastic!

It was also cool to compare to everything else I have watched because with the 720p 120Hz mode there was no scaling of course.

In the USA you can find the FIFA disc for less than $15 on eBay (I paid 13.00 shipped) and it is really a good demo. I am glad I will have it here for the meet on the 26th It was super fun for me to watch because I saw almost all the games during the WC last year.

craigr

[JohnHWman]

[CIR Engineering]

[JohnHWman]

[CIR Engineering]

[JohnHWman]

Sure Craig,

Just let me spend some time to check the G90 schematics first to see what can we do.

John

[CIR Engineering]

[CIR Engineering]

G90 Specific Update

John and I have been working today on modifying the G90 Vsize circuit to allow for more stretch. This will accommodate extra huge vertical blanking timings to help reduce ghosting.

John sent me a very good idea to accomplish increased V stretch in an email earlier today. One issue is that the Vbias affects other waveforms like SAW and LIN and altering these must be avoided. John skillfully managed to avoid these problems, but his mod would have altered both 2D and 3D stretch equally and I wanted to avoid 2D impact. Ie, I did not want to hurt 2D performance in any way. Most notably, I did not want to lose the ability to run 818x1920 72Hz 2D 2.35. This could have been a real problem because on most G90's to run this resolution you need to run Vsize Course at or near zero. Increasing Vsize range could have ruined this ability...

What I did was alter Johns approach a little to achieve both objectives. The way I did this was to modify just the G90 Vsize Fine instead of the total Vsize bias or instead of altering Vsize Course. Vsize Fine is a pretty useless adjustment because we don't need such precise increments for the RGB Vsize and I never use it anyway (you shouldn't either).

So with the current mod if I set Vsize Fine to null (value 127), than the mod is not even doing any work and the G90 behaves totally stock! But with the new mod I have increased the power of Vsize Fine so that it stretches the image with the same magnitude as Vsize Course. So essentially what John and I have done is make the Vsize adjustment have 2x the range by converting Vsize Fine into a second Vsize Course adjustment This has the added benefit of leaving all existing G90 memories unaffected as long as you have Vsize Fine set to 127~128 because this is equal to zero Vsize Fine stretch

The down side of the mods is that it will drive the vertical amps harder. I have tested the mod and it works well. At this moment I am taking temperature readings for the vertical amps on the DF board to compare them with the stock Vsize range. So far the results are excellent in terms of heat. The heat sink for the amps equalized at steady state with temps at 55C / 131F which should be fine. I am right now running the projector is stock form to get the base line temp for comparison to that. John however feels there is nothing to worry about with 55C and thinks it is fine as long as it stays below 100C which is not even on the radar

I am going to let me G90 mule run the new mods with the Vsize Fine and Course set to MAX. I will leave it run for many hours straight today to see if it bursts into flames If all goes well and someone else wants to try the mods I would be willing to share them per Johns approval. You will need to be able to do some basic surface mount rework and change out four SM resistors on three different boards. It is a pretty basic mod.

As far as the 3D performance with the mods goes... it has indeed further improved and reduced ghosting. Unfortunately, with 16x9 these still exists some ghosting towards the bottom of the screen, but it is again reduced when compared to the other improvements I've made. So comparing this to what I started with, ghosting is really getting to be less and less of a problem. I think at this point we are nearing the limits of how little ghosting this phosphor can do.

I talked to Patrick at Lumagen earlier today and he is going to get me a test firmware TODAY that will allow me to run whatever resolution I want. I will be trying a 1080x960 later today with a ton of extra vertical pixels. I like this idea of 1080x960 because of three reasons. 1) there is no vertical scaling 2) the horizontal scaling is an even multiple of 1920 at exactly half 3) with satellite side-by-side the Htotal is 960 so with satellite 1080 3D there is no scaling at all.That will most likely be my final resolution for 3D and I think this project will be finished with the experimental phase.

Kind regards,
craigr

[CIR Engineering]

By the way, a similar mod should be possible for any other projector I would expect. I think on the Marquee that the anamorphic mod could possibly be slightly altered and you will get the same affect.

craigr

[overclkr]

Dude you got SKILLZ.

[CIR Engineering]

[CIR Engineering]

Lumagen Update

Yesterday Patrick posted a public "beta" firmware that is now available for download.

http://www.lumagen.com/testindex.php?module=updates

Here is the description:

[JohnHWman]

That's good news Craig !

Keep us updated on your progress with your 3D tests results, especially about your 1080x960p96 3D mode (or should I write 960x1080p96 as is it 960 pixels wide by 1080 lines height ???)

John

[CIR Engineering]

[CIR Engineering]

Well, I just tried the new Radiance Beta FW and on my system the frame sequential 3D output mode is totally broken. Kudos to Patrick for trying to rush the FW out, but at least here it is not working. You still might want to try it, or just wait until next week.

craigr

[Alan Head]

[CIR Engineering]

[CIR Engineering]

[Alan Head]