RifTUP! FullHD Oculus DK1 upgrade kit

[mr.uu]

I would cut the cable connector to the mask right out and repurpose the whole
dk1 controller box, lvds cable, and 1280x200 panel as something else. Either a
low res vr mask with say the use of the other two lenses in dk1 or as a tiny
pc screen. I've done both. A tiny screen is quite neat when you are reclining,
not wanting to sit up to a monitor.

We want to use the Rift tracker, so we need the controller box. What you suggest is a complete DIY, which is not what we want. We want an easy do able Frankenrift (=DK with upgraded panel).

[cgp44]

I opened the dk1 controller box and saw the USB wires going straight through, no connection to the controller.
Perhaps I only saw two of the USB wires.

[OzOnE2k10]

@cgp44 - You know what, I never even thought of that. lol

Yep, you could quite easily use the stock panel for something else. Very handy in fact.
If you buy one of those cheap scaler boards with Composite / S-Video / RGB input, it would be perfect for using older consoles on, or for building a portable / mini cabinet.

I still have a Hydis panel + controller sat in a box here, but you can never have too many mini screens.

@mr.uu - I don't think cgp44 is meaning to necessarily use the stock panel for a DIY Rift.
I think he's just pointing out that it could be re-used for something else once it has been replaced with the 1080p kit.


You can see why Daniel went for the retro-fitting option though (stock DK1 cable), as it makes it easy for most people to RiftUp their panels.
(btw, somebody should trademark RiftUp - did you think of that name, Daniel? hehe.)


Yes, the USB connection simply passes straight through the DK1 scaler box, through the cable, and direct to the head-tracker.
The tracker is powered directly from the USB port too, it doesn't use the DC power socket at all IIRC.


It sounds like Daniel will be giving the option of an HDMI socket in the headset / controller as well as the connectors for the stock DK1 cable.

When extending the USB through an HDMI cable, I imagine most cables will work perfectly fine as long as they're not ridiculously long.
(using the original "DDC" pin pair, like in my guide thread).

I think a 5 to 8 meter HDMI cable would be more than enough for most rooms, and allow a bit of slack.


That's a point - @Daniel - I'm guessing there will be an EDID EEPROM on the controller box board itself?
Or, is the EDID part of the amp / EQ chip itself, or does the small MCU chip handle it?

Also, what sort of functions will the MCU on the board do, apart from switching between DVI / HDMI?

Really getting excited about this again now the Rift-up kits are getting closer.

OzOnE.

[mr.uu]

The two data lines are directly routed, the power line for some reason not.

I personally still think that we should go the way Daniel propose. IF there is a problem in using the old cable from controller box to the rift for hdmi transfer, we could go the way Oculus was going with their 1080p prototype - HDMI cable (picture) plus usb cable (power and tracker). Then we could use the DK leftovers for something else. IF the tracker works that way. I am pretty certain, that we soon will know.

[OzOnE2k10]

Oh yes, looks like the stock controller also controls the power to the tracker?

http://d3nevzfk7ii3be.cloudfront.net/ig ... f2A2A.huge

Never noticed that before, but should cause any problems.

I think it will be fine to route the USB via the HDMI cable, as the DDC / EDID pins will be spare anyway (EDID is on the controller).
Or, maybe leave the option for a separate USB socket at the headset too?

Is this starting to complicate things a bit too much though? Not sure?

We'll wait to see what Daniel comes up with.

The tracker is definitely just straight USB / power.

I'll have to test later on if the stock controller does enable / disable the tracker power though - can't say I've noticed the "USB connect / disconnect" sound in Windows before?

Actually, I'll give that a try now to see if the USB power works with the DK1 controller power button.

OzOnE.

[rozsnyo]

The control box for DK1 has no function which is required for the tracker to work. It is just taking the 4 usb signals (ground,power, d+,d-) to the headset. The decision to use a separate 4 pin plug at the control box side remains unknown for me - maybe they wanted the cable to look symmetrical and source just 2 type of connectors, not three. At the headset end the cable must be split as the tracker board is separate piece.

@mr.uu - the power line might be switched according to power status of the "monitor" part of the controller - one can verify that by looking at device list with powered off / on DK1

@OzOnE2k10 - yes, the RiftUP name is my idea

Currently the RiftUP design takes HDMI+I2C (for EDID) up to the head, and a separate I2C which is used for controlling the switch, buttons and led. With OzOnE2k10's (and others) ideas to use a HDMI cable with both hdmi+usb (on EDID pins) I can make some soldering point locations prepared for putting/expecting the usb from there, so one just needs to do the other piece (usb+hdmi -> our combined hdmi). And that piece then WILL contain a discrete EDID eeprom chip. The design I have now does not need one, as EDID is simulated from RAM in the hdmi receiver chip.

Result: headset board input style: DK1 lvds cable / separate HDMI + USB / merged HDMI&USB (some soldering required to change one into another).

[OzOnE2k10]

Hi,

I just checked my DK1 controller, and it does have what look to be some transistors for controlling the USB power.
It's not a big deal at all of course.

I haven't actually plugged the PC back in yet (my Rift was boxed up, and I borrowed the monitor for something else), but the tracker is definitely powered via the mini-USB port as you say.

Interestingly - the 5 Volts from the USB port to the tracker side is only enabled when the DC mains adapter is plugged in.
So, I guess Oculus just wanted the tracker to be enabled only when the board has DC power?

It doesn't seem to matter whether the power LED on the controller is ON or OFF, only that the DC power is plugged in.
Also, it doesn't need the DVI or HDMI to be plugged into the PC, so it's not because they wanted the tracker to power up only with everything connected.

Again, it's just an interesting observation, as I doubt it will affect anything at all for the RiftUp..
It just means that the tracker will be powered up as soon as the mini-USB is plugged into the PC.

I'll have to trace the circuit just out of sheer curiosity.
(The USB +5V power goes to one side of transistor Q8, then is switched whenever the DC jack is in.)


Oh, I forgot EDID was in the headset. Makes sense though.

The "merged HDMI+USB" option with soldered jumpers is perfect.

tbh, if people have enough modding skill to want to extend the cable, then they are very likely to have a soldering iron.
Shouldn't be difficult to do for the average modder either.

Just need to leave the option open for extending the cable, and people can use separate HDMI / USB cables if they wanted (pretty sure I'd go for the merged option though).

OK, good stuff!

OzOnE.

[OzOnE2k10]

Mystery solved.



EDIT: Reduced image size a bit.

I just traced this from my DK1 controller board.

Looks like all it does is that the RTD2483 scaler chip can detect whether or not USB power is plugged in (pin 103).

The scaler can also enable / disable power to the tracker (pin 104).

No big secrets there, and hopefully won't affect anything for the RiftUp.

That is unless the SDK can actually enable / disable USB power via I2C on the HDMI / DVI port, so it can detect which tracker is related to which display output (by seeing which tracker board re-enumerates when USB power is cycled)?

Hmmm, time for a bit more testing.

I honestly don't think this will change anything for PC's with a single Rift on, but it may be a feature used by the SDK for multiple Rifts on the same PC?

I'll check the latest SDK to see if it checks any of this stuff.
It doesn't sound like something they're currently using, or you'd probably hear the USB "disconnect / reconnect" sound every time you start up a Rift game / demo?

OzOnE.

[cgp44]

How are we going to access the hdmi wires for the USB?

Either have a base board for that. Or cut into the hdmi cable and go looking for
(connectivity test on hdmi plug) the wire that is the designated one for each USB wire.

[cgp44]

Hi rozsnyo,

Regarding a box of 120 pieces of 5.9er sharps. I would go for it as long as you have the
numbers here . We need a mechanism here where a maker such as rozsnyo can determine numbers (definite orders / possibl interest) to judge the commercial risk. Such a mechanism would be more than sending messages, we need a table where
Many logged-in prowlers/viewers can click, and a count tallied.

Whatever the OEM variation as long as they work for FHD, you can cope with no?
Is it a matter of sourcing a suitable connector and pcb. Perhaps you should find out the various
Connector pcb landing variations and make a pcb layout that does two or three of them.

Are these 120 pc boxes from a Chinese source that does a reasonable price or is it from the
markup man?

[rozsnyo]

Some progress for today.
Schematics done (copy paste is not so easy when the component numbers overlap... so there was a lot of cleanup required).
Given that we got quite a lot of space behind the 5.9, the component spacing is more relaxed.

[stryker7314]

Some progress for today.
Schematics done (copy paste is not so easy when the component numbers overlap... so there was a lot of cleanup required).
Given that we got quite a lot of space behind the 5.9, the component spacing is more relaxed.

Good work Rozsnyo! Looking forward to your kit!

[rozsnyo]

So.. a day to stuff the components to their proper place.

As this version features automatic USB source selection (DK1, USB, HDMI), usb hub and a processor programmer, it would make sense to condense this design into a 4.7" space and then make a non-specific LCD interface connection (so far there is 5.9" plug on the other side of the board). There is also a one more hdmi switch, to select rift/hdmi, as I would like to offer this board in a cut/crop fashion to non-RIFT use.

What are the odds of getting 120 orders for this model? Or shall we make lets say just 20 and confirm the suitability of 5.9" on tha amount? I expect that we strip down the design to bare minimum for the "mass produced" upgrade kit.

This model is highly likely to drive other MIPI screens, so its higher cost is definitely weighted with higher value.

I hope you do not mind that I post progress data here, feedback is definitely welcome!

[Hornet]

Wow this looks really nice and sophisticated! Tesim tesim

[mr.uu]

If the boards do work you will sell them all - 20, 120 or 1200.
Out of >50.000 sold DKs owners there are certainly more than 2% who do want a 1080p rift immediately AND can afford it (150-200€).
Just my two cents...

[Bloump]

Hello there! just wanna raise a hand, if you sell, i buy ..probably was already told, but to avoid risk why not indiegogo or similar? you only proceed once you get the money..
ah, please, make it stupid proof to upgrade

[KBK]

Yes, the USB connection simply passes straight through the DK1 scaler box, through the cable, and direct to the head-tracker.
The tracker is powered directly from the USB port too, it doesn't use the DC power socket at all IIRC.

Yes, that is true. IIRC, The tracker re-regulates to two different voltages, through two different regulators. One for the sensors, one for the CPU. (IIRC, from the spec books, one is a floating reference....? (I'd have to go back and do lots of reading again.)

[stevetb]

So.. a day to stuff the components to their proper place.

As this version features automatic USB source selection (DK1, USB, HDMI), usb hub and a processor programmer, it would make sense to condense this design into a 4.7" space and then make a non-specific LCD interface connection (so far there is 5.9" plug on the other side of the board). There is also a one more hdmi switch, to select rift/hdmi, as I would like to offer this board in a cut/crop fashion to non-RIFT use.

What are the odds of getting 120 orders for this model? Or shall we make lets say just 20 and confirm the suitability of 5.9" on tha amount? I expect that we strip down the design to bare minimum for the "mass produced" upgrade kit.

This model is highly likely to drive other MIPI screens, so its higher cost is definitely weighted with higher value.

I hope you do not mind that I post progress data here, feedback is definitely welcome!

Rozsnyo,

You will have no problem hitting the 120 unit mark, no worries there. I would suggest as soon as you can to put together a group for your first batch of 120. Many of these users, such as myself, would gladly help front some of the upfront costs. I sent you a message of my interest but never heard a reply. I am very interested and absolutely want to be in the first wave of these modification kits, very excited here!

Where do I send my money?

-Steve

[remosito]

do a kickstarter/indigogo campaign? Provide limited amount of 120 kits for people to back. If they sell out real fast. add another 120? and so on???

I would be really surprised if the kits don't sell like hotcakes after it appears on reddit and oculus website. Only dampener I could see is a April-June DK2 launch confirmation by Oculus at GDC...

[ripcurl123]

don't let rumours and speculation ruin this thread i cant see dk2 or consumer being released in april june ,and if it did the waiting list will be stupid going by dk1 last year. there is a big demand for just 1080p at the moment ,and whos to say dk2 wont be delayed remember last year, as to quantity 120 got to be minimum ,
ps i dont want to come across as dissing oculus i think palmer has done a brilliant job bringing us this tech and i for one appreciate it..
vr forever he he
just my two cents
dave

[Fredz]

I wonder if it would be possible to switch the pixels faster than the refresh rate with your board.

The panels are most probably 60Hz (possibly even a bit more, like the 71Hz of the DK1 panel), but could it be possible to send the signal at a faster rate (say 120Hz) but only display at the lower supported rate (60Hz) ?

This would help a bit in reducing end-to-end latency. For now it's 1ms (USB) + 3 x 16.6ms (rendering, write display, pixel switching) = 51ms, but with this technique it could be reduced to ~42ms (8.33ms gain).

Ideally backlight strobing and LCD overdrive/real time compensation would be quite useful as well (for motion blur and ghosting reduction), but I don't know if the cost/complexity is worth it.

[shinseiryu]

Even if the DK2 is released soon I will order that and still order an upgrade kit for my DK1. I would like to loan/give my DK1 to a friend who cannot afford his own DK.

[Nick3DvB]

Ideally backlight strobing and LCD overdrive/real time compensation would be quite useful as well (for motion blur and ghosting reduction

You read my mind Fredz, I have no idea what's possible (without an OLED) but if you're building your own controller it must be worth trying something?

Mark Rejhon's article might be a good place to start: http://www.mtbs3d.com/phpBB/viewtopic.p ... &start=680

[cgp44]

how to have two headsets on one computer would be a good and long running thread.

[cgp44]

I've just got some brand new lcds from etradesuppy.

optimus G pro lg 5.5 inch fhd lh550wf1-sd01 us$32 <======== that's the one
htc one 4.7 inch fhd us$53
lg 4.3 inch venice 730 2-laner mipi 800x400 us$26

These lcds weight nothing the lg55 was 22grams, the smallest 18grams.

here are the lcds placed over the current DK1s warped image extents
The rumors that the oculus hd was the lg5.5 could be.
Regarding coverage I like the idea of a dual lg5.5 on DK1 optics.
A dual 4.7 would have the best true pixel density but has serious fov
edge cutoffs. Of course dual would require beasty GPU power and I have not
seen much cost reduction from the dualoppoly. They even remove the 'last years'
model from the stocks so you dont get the price reduction pressure of past years.
Specifically, I brought the gtx670 last year for nz$500 and was expecting nz$300 or
less for another sometime soon. But no, nothing on stock or same price. Noone is
selling their gtx670 just yet.

dk1 dual head possibles.jpg

Now I'm wondering if we use the 4.7er what is the best fov cutoff positioning. Perhaps
a greater fov cutoff on the nose side?

lg55onDK1.jpg

[cgp44]

I've just found the posts on page 40 of the other main diy thread (which died for about four months).

It contains the lg 5.5 datasheet lh550wf1-sd01 and the chip for hdmi to mipi.

For those who are looking into the details of mipi, and willing to do something
rather than wait for someone else- these are the best explanation and details
of the mipi programming level as in dcs commands to send and pinouts.
I have wasted weeks of time on the crap that is out there.

Of course rozsnyo is about to deliver big time and will make a killing, but
there are other routes to diy happiness.


These are how datasheets should be written, not the pathetic ones probably written
by the inhouse newbie who knows little and is not in touch with the latest build
by the typical introvert engineering team.

[rozsnyo]

Design of the headset prototype board is done! We manufacture one during this week.

[SaltySalmon]

Looking really good! Can't wait for all the magic to happen!

[DoppelGUNger]

Hello rozsnyo. I build my Oculus on tfp401 -> ssd2828-> lh550wf1. havent you have any init strings for LG lh550wf1.
p.s. sorry for the Google translation

[BrightEyed]

I'll be keeping a close eye on this too, am really interested in procuring a RiftUp kit when available. Looks like you've done great work on it so far!

[cgp44]

the lh55owf1-sd01's datasheet is on page 40 of the other diy large thread (up to 45 pages)

This is an excellent datasheet with the initialisation string clearly documented, but not explained.
This goes into OnoEk210's concerns about scaling. Does the IC drive do the scaling itself or do
we have to send another initialisation dcs init sequence, as there are timing string of bytes.

So laseredge (wherever he's gone) said that most of these use the IC driver chip (more a silicon strip that you
see on the edge) use Renesas 63311. We will see if that explains timings. On the troll for that.


It has the 30 pin GB042-30P-H10-EX000 where P is plug, substitute S for socket, H10 is the height
of the plug and socket together: 1mm . EX000 for either E1000 or E3000 but that I think is just an internal design
and not to do with dimensioning. The display has the plug/header, you get the S part.

Unfortunate etradesupply don't have them. I've see 5 to 10 miniums in alibaba, sent a 5 order with an initial
price of $2.

So Im desperate of GB042's and have even considered substitution by way of close up comparision of the
datasheet drawings. I looked at two other's on etradesupply datasheet links and the GB042's have an
uncommon fat centre hole.

[rozsnyo]

the lh55owf1-sd01's datasheet is on page 40 of the other diy large thread (up to 45 pages)

This is an excellent datasheet with the initialisation string clearly documented, but not explained.
On this goes into OnoEk210's concerns about scaling. Does the IC drive do the scaling itself or do
we have to send another initialisation dcs init sequence, as there are timing string of bytes.

So laseredge (wherever he's gone) said that most of these use the IC driver chip (more a silicon strip that you
see on the edge) use Renesas 63311. We will see if that explains timings. On the troll for that.

Good luck with those. They might seem nice but they are false. I am using different codes and the 5.5" was the hardest model to get to work.
The displays do not support scaling - feeding different resolution ends up in "garbage" being drawn on the display.

[cgp44]

rozsnyo, I've being thinking about those tiny 0.4mm 30 to 44 pin connectors.
They are so delicate that they can easily be damaged if you pull them wrong.
I have one with a pin missing. Perhaps they are designed to only connect once
and the replacement scene has this risk.

If you place the connector on the main board you are risking the whole board.
Please consider having a separate tiny pcb just for the connector that can be swapped.

[rozsnyo]

Yes, that will be in the next version - a totally simple hdmi to mipi screen.
This 5.9" is rather my "dev kit 2" and a candidate board for Rift DK1 upgrade - where single connection is not an issue.

Btw I had demonstrated my modular kit 10+ times with various screen swapping, connecting, disconnecting, and the plugs still live. But low mating count is not unusual - e.g. the FPC for iPad3 is rated for 10 matings and people send me some time the board back as it gets broken... resoldering a new plug is not an issue, so I think that will apply for 5.9" too. It is just not supposed to be a regularly connecting/disconnecting part. You have HDMI for that!

Oh, last update: I have wired the VSync singal to the processor's interrupt pin, so we can play a little with strobing the backlight (needs a little hack to raise the current though).

And here is a simple splitter, for those who would use my "custom HDMI", which offers hdmi+usb+power (usable when you are building a headset from scratch and you want to use single standard hdmi cable for both the screen and the tracker).

[cgp44]

just how do you resolder these micro wires on these connectors!
They are reflow only NO???

[cgp44]

rozsnyo, regarding the failed resolutions. Did you just send the lcd the new timing
signals or did you also try to reset the timing via those timing sequences (if they were
basically correct in their format, or was the init sequence structure (not just the data) completely off, i.e. perhaps wrong IC driver. I'm looking for the relevant renesas driver part.

[cgp44]

The renesas R63311 can do 9 resolution combinations.

http://www.rsp.renesas.com/en/news/2012 ... 120418.htm

[3dvison]

Hi rozsnyo,
Nice to see how the Rift-Up is coming along.
Do you have an update on when you might start taking orders ?
Or is that still far off ?

[rozsnyo]

The renesas R63311 can do 9 resolution combinations.

http://www.rsp.renesas.com/en/news/2012 ... 120418.htm

You misinterpret the text. The R63311 is capable of driving screens with the listed resolutions. It does not scale the image!!

[Fredz]

Oh, last update: I have wired the VSync singal to the processor's interrupt pin, so we can play a little with strobing the backlight (needs a little hack to raise the current though).

Great, thanks for the addition !