FB3 and RJ45 (cat5)

Printable View

Show 50 post(s) from this thread on one page

09-22-2010, 02:45
FourDee
I think we're on to something here.
The way I see it is this:

Transmitter box:
- ILDA in
- Address selection DIP-switch
- RJ-45 connector
Receiver box:
- ILDA out
- RJ-45 connector, autodetect if it's connected to a router or straight to a transmitter box (ie. crossed cables or not)
- Address selection DIP-switch tells it which transmitter to listen to
- Other functions could be implemented digitally
  - Inverting of x an y axis
  - Summing a color signals for single color projectors
  - ...
- If you want to get fancy you can even put an lcd on there for display and configuration of parameters
So a show setup would look something like this:

Code:

DAC > TRANSMITTER > | | | | > RECEIVER > PROJECTOR DAC > TRANSMITTER > | SWITCH | > CAT5 > | SWITCH | > RECEIVER > PROJECTOR DAC > TRANSMITTER > | | | | > RECEIVER > PROJECTOR | | > RECEIVER > PROJECTOR | | > RECEIVER > PROJECTOR
09-23-2010, 23:26
Things

Here is my solution:

http://dl.dropbox.com/u/203420/IMG_2235.jpg

No circuitry at all, just direct links between the DB25 and ethernet. The green indicator LED is for debugging, sort of. + and - power for it is sent through the 2 seperate ethernet cables. The LED will light only if both cables are plugged in, and in the right ports.

Saves running back and forth checking and blowing stuff up :)
09-24-2010, 05:35
andy_con

in an ideal world it would be good to run all 25 ilda pins down a single cat 5 cable
09-24-2010, 14:51
Things

Well yeah, I made these things for my purposes. I don't carry through any of the extra colour signals (FB3 doesn't have them anyway), however I did carry though + and - for everything, including colour (Even though it's just grounded in the FB3), so it does maintain a bit of the standard. My projector doesn't have an interlock or shutter though, so I didn't carry those through.
09-24-2010, 15:23
White-Light

Its a pity there don't seem to be any analogue to optical convertors, only digital to optical as I found this cable which sounds ideal (at least to me with no electroinic knowledge) and is reasonable at 66p per metre in its armoured form:

http://www.comms-express.com/product...e-fibre-cable/

This is the data sheet:

http://www.comms-express.com/assets/...20LT%20CST.pdf

Edit: I wonder if you could send digital optical from the pc to the FB at the projector end with an optical to digital convertor providing the flashback input at the FB end?
10-14-2010, 01:18
heroic

For the love of all that is holy please please please use standard ethernet framing. The chips are cheap, the magnetics are easily available, and if you use gigabit ethernet and you're only sending data in one direction, even the cheapest switches will never have collisions since all links are full duplex.

-J.,
designer of stuff that talks more data in a day than an FB3 does in a century
10-14-2010, 03:37
gashead

I've always thought it would be beautiful to come up with such a device, something with a LCD display and maybe a few buttons.

You could set the IP address of the device and then "encapsulate" ILDA over IP, perhaps you have a head end that connects to your USB port and or takes a ILDA in and then in turn this device connects to all your "remote demux".

Being a Network engineer by trade, this is cool shit! there aren't many boundaries to this, all someone needs to do is come up with the hardware to do it!

Just imagine it, if we had devices such as this we could have word wide simultaneous LEM's and run synchronized laser shows across the Internet!

Just one of the possibilities!

Who's game to start designing the hardware ? *retardo

N.
10-14-2010, 06:05
DMills

Ok, how does a line up like this look to people....

A single eurocard with a modest ARM chip (something with a built in ethernet MAC and PHY ideally), Texas make some latched 8 bit DACs that come 4 to a package, so two packages of those plus some 5534 or similar as line drivers will give 8 channels of colour control for full ILDA spec, for the main output there is that rather tasty line of TI 12/14/16 bit DACs (pick to suit requirement they all share a pin out).

The plan would be to run from +-15-24V so scanner supplies could be used.

If we use one of the on chip timers to directly drive the ldac signal AND to generate an interrupt then there will be effectively no timing jitter as the ldac signal is the critical one and there is no software involvement in its generation (The ISR just has to line up the next set of data on the IO ports.

These things often have a couple of async serial ports on board, so DMX and possibly midi? DMX could conviniently use artnet as a protocol (widely available accessories), and we could well look at using the sync serial port for an audio output....

Add a few GPIO pins and whatever other bells and whistles are available and it is all good.

I personally favour using rotary switches and dips rather then an LCD as it makes mounting the card in an existing projector simpler.

There will be quite a bit of smt involved, including fine pitch QFP for the processor, but it should just about be hand solderable or workable with a simple minded reflow oven.

Anyone have a TCP/IP stack for an arm7 or similar? The software will be the biggest part of the job.

There will need to be traffic both ways as the projector end needs to send its buffer fill status back to the host (Plus presumably advertising its presence on the network), and responding to ARP requests.

Do we really want hardware conversion from ILDA at the control end? It is modestly hairy to do right as many ILDA ports lack really adaquate low pass filters to produce something that has a well defined upper bandwidth limit and I could see strange aliasing being possible, surely better to just write drivers that let the cards just hang off the laptop ethernet port (Universally available, well tested, simple)?

We could have optionally fit jumpers to allow one of the spare pairs to carry the E-stop signal so a single cable solution would be possible.

Heroic, do you know of a small 100Mb/s switch available as a single chip in small quantities? Being able to fit a loop through for the ethernet would be handy.

Just my thoughts, on something I am noodling with in my spare time.

Regards, Dan.
10-14-2010, 08:59
heroic

Quote:

Originally Posted by DMills

Ok, how does a line up like this look to people....

A single eurocard with a modest ARM chip (something with a built in ethernet MAC and PHY ideally), Texas make some latched 8 bit DACs that come 4 to a package, so two packages of those plus some 5534 or similar as line drivers will give 8 channels of colour control for full ILDA spec, for the main output there is that rather tasty line of TI 12/14/16 bit DACs (pick to suit requirement they all share a pin out).

I think I'd probably use one of the Cortex-M3 chips that are widely available now, if I were going to run TCP/IP on the micro. But... I'm kinda lazy, so...

Quote:

These things often have a couple of async serial ports on board, so DMX and possibly midi? DMX could conviniently use artnet as a protocol (widely available accessories), and we could well look at using the sync serial port for an audio output...

Don't unnecessarily complicate things in version 1.

Quote:

Add a few GPIO pins and whatever other bells and whistles are available and it is all good.

I personally favour using rotary switches and dips rather then an LCD as it makes mounting the card in an existing projector simpler.

I personally favour using the on-chip USB port as a programming interface and maybe having a couple seven-segment displays so you know if it's tx or rx, what IP address it has, and which 'channel' it's on.

Quote:

There will be quite a bit of smt involved, including fine pitch QFP for the processor, but it should just about be hand solderable or workable with a simple minded reflow oven.

Anyone have a TCP/IP stack for an arm7 or similar? The software will be the biggest part of the job.

If you use a Wiznet chip or similar, the IP stack is embedded on the ethernet MAC. Indeed, you wouldn't even need an ARM then. You could do it with something trivial like an AVR or similar.

Quote:

There will need to be traffic both ways as the projector end needs to send its buffer fill status back to the host (Plus presumably advertising its presence on the network), and responding to ARP requests.

Not significant amounts of traffic compared to the data stream.

Quote:

Do we really want hardware conversion from ILDA at the control end? It is modestly hairy to do right as many ILDA ports lack really adaquate low pass filters to produce something that has a well defined upper bandwidth limit and I could see strange aliasing being possible, surely better to just write drivers that let the cards just hang off the laptop ethernet port (Universally available, well tested, simple)?

And incompatible with Pangolin software. Lowpass filtering an ADC is easy.

Quote:

We could have optionally fit jumpers to allow one of the spare pairs to carry the E-stop signal so a single cable solution would be possible.

No, you use a timeout; if no control packet is received within a 10 msec window, kill the output and break the interlock loop. Don't screw with ethernet wiring.

Quote:

Heroic, do you know of a small 100Mb/s switch available as a single chip in small quantities? Being able to fit a loop through for the ethernet would be handy.

No. Everything's gigE these days.
10-14-2010, 11:02
DMills

You know if the objective is simply to do a point to point ILDA over cat 5/6 (as opposed to over ethernet) there is a much simpler way to skin this.

TI do a line of audio ADCs and DACs that have a 'dsp' mode in which they time division mux 8 channels onto a single data line, so that would be X,Y,R,G,B,I, + 2 user defined....

The silicon needs a serial bitclock and a word clock for synchronisation as well as (sometimes) a system clock for internal operation.

At the transmit end, you have some signal conditioning logic, and the clock generators driving the following signals down the 4 pairs:

Wordclock (~50Khz as LVDS or RS485, no DC component so easy to transformer isolate if required).

Bitclock (256 * fs) = ~12Mhz as above.

Data, as LVDS, may have a DC component.

The last pair can be used to carry a simple sync serial stream using the above clocks for things like shutter and interlock.

You could (almost) implement this with HC series jellybean logic, no real need for computation.

But something ethernet based would allow the use of off the shelf switches and things like wifi links, so is probably worth pursuing.

I didn't know about those wiznet parts, cool!

Regards, Dan.

Show 50 post(s) from this thread on one page