any way to extend this flat cable?

[lasergrazy]

Try this site they make complete cables, it's a German site http://en.esskabel.de/uebersichten/C320
Check the pin count and me sure the pitch of the pins and how the connector locks, screws I think
and they tell you if they have one.
And those guys are able to help you to http://meritec.com/products/flat-fle...isplay-cables/
Ask for FFC, flat flex cable with connectors what you need looks like an display cable.

I hope this helps.
Albert

[lasergrazy]

On your pdf there are a bunch of dimensions on page 3 and below that's all you need to figure it out

[Yerkat]

Thanks, guys. I will try contacting that website about getting a cable made.

The cable length required is exactly 100cm.

3. maybe the signal timing is important, the red/green/blue LEDs are timed perfectly to match the content on the DMD, if the DMD gets the data late because of the long cabeling, then the colors will not match the image.

The power cables for the red, green, blue diodes use a ribbon cable with a more easily sourceable connector still connected to the PCB. It's a 10 pin connector and ribbon cable for each color, 5 for - that get joined on the pcb and 5 for + the same way, my guess is the individual pins were not enough to handle the current and thats why they went with 10, judging by the width of the conductive paths for these on the PCB.
Since these led power cables need to be extended as well to the pcb their lengths can be extended the same amount, right?

That said I think how DLP systems work is the LEDs or r,g,b beams separated from a bulb are always on and the color wheel plus the mirror angles just block the light coming from other leds on time. I don't think they are modulated, just the current is controllable so people can set different color modes or switch between eco mode or normal.

I don't know at witch frequency's the DMD is working, maybe 3. is not a problem, but 1. is for sure a problem, the electrical engineer is going to the trouble to design matched pairs (you can see them an the picture of the PCB, the snake like traces to match the length of the pairs), i think you need this too

I'm trying to understand this point. Since the lengths are matched already on the pcb, why can't the extension cable have equal length lanes?
Thank you very much.

[Farbe]

Hey,

there is a color wheel?
wow i didn't expected this, it adds a component and a failure point.
i Expected that the LEDs are switched accordingly to the DMD image.

then the colorwheel timing and the dmd timing are critical.

if you need exactly 1m the signal propagation time will be around 5ns for copper cabeling

that means that if the DMD gets signals with a maximum of 1Ghz frequency (1ns cycle duration) that the signal arrives 5 Bits to late at the DMD

I don't think that the timing is that critical at the end.

The fact that any cabeling is a lowpass filter is worse than the propagation delay, so if the signals are high frequency (witch they are, a dmd uses dithering to make grayscale, so a lot more than 60fps/1080p)
if your cabeling has too much capacity/inductance than your signal gets distorted enough to be not working.
Thats the matched impedance thing, the traces on the pcb have a fixed distance to each other and to the ground plane underneath. To "control" the signal distortion.
You can do this in cabeling harness to, but its really complicated. First you need to know the wave impedance thats used on the PCB/DMD, your cabeling needs to be exactly the same to reduce reflections.
That requires very specific conductors and isolation and distance of the conductors, not easy.

I'm trying to understand this point. Since the lengths are matched already on the pcb, why can't the extension cable have equal length lanes?
Thank you very much.

You need a special cable with the exact same length on all pairs, if you make your own, thats difficult to do. I have tryed to make longer labels for LVDS Display long time ago, and it was very difficult, the length must be matched very closely, and see the above section about impedance matching.
You need the exact same length because the image data gets splitted between different pairs to reduce the frequency needed to transmit the data.
if one pair is longer than the others the data arrives late at the dmd, then the dmd can not identify what the data meant to be.



is there are really no way to get the pcb close to the dmd?

[Yerkat]

Hey,

there is a color wheel?
wow i didn't expected this, it adds a component and a failure point.
i Expected that the LEDs are switched accordingly to the DMD image.

That would make more sense, but maybe cost of modulating LEDs is too high, or generates more heat, or something else, because all of their 3 LED LCOS or DLP projectors have a color wheel. I am as clueless as to what the requirement behind the colorwheel in these systems is.

if you need exactly 1m the signal propagation time will be around 5ns for copper cabeling

Sounds fine.

The fact that any cabeling is a lowpass filter is worse than the propagation delay, so if the signals are high frequency (witch they are, a dmd uses dithering to make grayscale, so a lot more than 60fps/1080p)
if your cabeling has too much capacity/inductance than your signal gets distorted enough to be not working.
Thats the matched impedance thing, the traces on the pcb have a fixed distance to each other and to the ground plane underneath. To "control" the signal distortion.
You can do this in cabeling harness to, but its really complicated. First you need to know the wave impedance thats used on the PCB/DMD, your cabeling needs to be exactly the same to reduce reflections.
That requires very specific conductors and isolation and distance of the conductors, not easy.



You need a special cable with the exact same length on all pairs, if you make your own, thats difficult to do. I have tryed to make longer labels for LVDS Display long time ago, and it was very difficult, the length must be matched very closely, and see the above section about impedance matching.
You need the exact same length because the image data gets splitted between different pairs to reduce the frequency needed to transmit the data.
if one pair is longer than the others the data arrives late at the dmd, then the dmd can not identify what the data meant to be.

That sounds bad. To be honest I'm not that into electronic engineering to know enough about inductance, I'm more of a software guy. Arduino and simple boards is as far as my knowledge goes in this field.
Does this mean a "99% won't work" or still worth a try?

is there are really no way to get the pcb close to the dmd?

Here's the exact situation I'm in right now:

The projector dimensions is about 20cm x 20 cm. The manufacturer made the PCB the size but there's a lot of unused space on the PCB.
By moving the heatsink to a new location which used to be reserved for a battery for a powerless operation I shrank the size of the optical engine to 13cm x 20cm which is perfect for my application.
But because of the PCB I can't use the enclosure and meet the size constraints. I'm in the product phase where I need to present a prototype to select people for evaluation. It is not clear yet whether this will be a profitable product or not, so I can't just spends thousands to get someone to design a simpler smaller board with all the buttons and inputs stripped, no OS with GUI and just DC in, HDMI in and AUX out left.
Sure a long ribbon cable and and more cables for LED powering going to another box is not pretty but for the sake of demonstration it will do because it does not affect the main point of the demonstration, but a larger enclosure for the optical engine caused by a large PCB will be a deal breaker.

[Farbe]

Hello,

can you post a Pic from the whole board?

Maybe the 7cm can be shaved of from the board, if the unused circuitry (Battery charger/Audio Amp) can be removed.

i don't think you will find someone that makes you a cable in prototyping quantities, and if you find someone than it will be very very expensive

i think the only option is to make connection boards to adapt the connector to a standard connector like pinheaders or something.
that makes for a less expensive try
i would try to use standard IDE/ATA cable thats 120ohm impedance and can be matched to the existing impedance.



if i researched correctly your DMD has 16 LVDS channels
LVDS is good, i think it is possible to do it.
The DMD´s from Texas have 120ohm impedance at the LVDS inputs, so in theory this cable will be very good.
Now you only need a breakout board with the connector and decoupling capacitors for the power supply pins.

This should be possible


BTW: i don't see a color wheel in your pics, are you sure that there is one?

[Yerkat]

Hello,

can you post a Pic from the whole board?

Maybe the 7cm can be shaved of from the board, if the unused circuitry (Battery charger/Audio Amp) can be removed.

I will post macro photos of the board from both sides today when I get my DSLR.

That said I'm pretty sure we can't just shave the board, I was thinking about it myself but the power connector, hdmi port, two controllers and led connectors are on that side. But I will post photos soon anyway.

EDIT:

i think the only option is to make connection boards to adapt the connector to a standard connector like pinheaders or something.
that makes for a less expensive try
i would try to use standard IDE/ATA cable thats 120ohm impedance and can be matched to the existing impedance.

if i researched correctly your DMD has 16 LVDS channels
LVDS is good, i think it is possible to do it.
The DMD´s from Texas have 120ohm impedance at the LVDS inputs, so in theory this cable will be very good.

Sounds good. Thanks for the research, couldn't have gotten here without you. Here are the controllers and DMD used in this projector:
http://www.ti.com/product/dlp4710
http://www.ti.com/product/dlpc3439
http://www.ti.com/lit/ds/symlink/dlpa3005.pdf

I haven't actually disassembled the optical engine and had a look on the DMD but there's only one other DMD doing 1080p and it doesn't match with the controller so it leaves this one DMD.

BTW: i don't see a color wheel in your pics, are you sure that there is one?

Not really, now that you mentioned it the color wheel is mentioned only by third party sources and the rainbow artifact to the eye and cameras could be caused by flickering LEDs instead.

By the way I contacted the two companies selling connectors and ribbon cables which were linked to, one hasn't responded yet ( meritec.com ), the other one ( esskabel.de ) said they couldn't help but it agreed it looked similar to the "Hirose DF18" connector. https://www.hirose.com/product/en/pr...D0.4V%2881%29/

[Farbe]

Hey,

mhm
i changed my mind and i think it will not be possible, the integrated LVDS driver is to weak to drive a 1m cabeling

Take a look at the datasheet from the DMD driver IC, it states clearly that the LVDS drivers that drive the datelines to the dmd (eg the cable) are not capable a driving more than 6 inches (around 15cm) of 100 ohm impedance transmissionline at the needed data rate.
That changes everything...
Now you need adapter boards with amplification to increase the length that can be driven.
thats not easy and requires serious engineering

cabling without amplification = not worth a try

sorry to tell you that.


but it is however possible to sever the board in halve.
i requieres work but i think its possible.
if you draw a diagonal line starting between the GB and R led connectors to the USB port, cut the board and reconnect the required signals.
The top stuff is only power supply stuff, and the fet to switch the Red led, therefor if you don't need the sd card and the battery connector, than it may be possible.
If i am correct, than the top IC (U14) is only for the charging of the battery, the LED on the left side is for showing the state of charge/on/off state, the BJT/fets are to control it, unnecessary, the IC (U15) is only the speaker amp, Q35/Q38 is to switch the power form/to the battery and to the LED driver.
The unused connector is labeled SPK2 (speaker 2) on the other side of the board is maybe the DAC or another audio amp

Hope it helps,
best Regards Farbe

[Yerkat]

damn...

Thanks for taking the time to think of ways on how to shrink the size of the board, however you seem to be suggesting to get rid of most of the right side of the board ('right' in the first image), however we need to get rid of the left side to make room. Thats where the hdmi port and power ports, the TI LED controller, the molding power choke thing and the LED power connectors are. And the (maybe) DAC (cant find info on that part number). Getting rid of the top-right side of the board wouldn't help because thats where the long lens assembly is. The left side has plenty of room because the heatsink can be moved farther right as the battery location is freed. I probably confused you by the first image of the topic where the image is upside down.
Or am I missing something?

EDIT: Heres the optical engine and the PCB that goes on top of it next to each other in correct angle. The flat \dmd cable is long enough to let me flip the board vertically, but not long enough to twist the cable and flip the board horizontally, I think.

[Farbe]

Hey,

yes, i mixed the orientation.

its not possible to sever the board at the required location.

the only thing that i can think of is flipping the board bottom side up, and twisting it at 90 degree to the right, but i don't think that the existing cable is long enough to do this.

Maybe its possible to change the optical path? Like using a mirror to change the lens form "up" to left?