LCD Monitor Modification

[vne147]

Hello everyone. I’m working on something and have hit a snag and could use your help to resolve it. I’m trying to help a friend with a project where he is modifying an LCD monitor and a graphics tablet for a PC. The end result of the modification is that he will be able to draw driectly onto the monitor instead of having to draw onto the graphics tablet. I’m pretty sure the tutorial he is following is here:

**broken link removed**

As part of the modification, he has taken apart the LCD monitor. Inside the monitor is of course the LCD itself and associated lighting but also inside are two PCBs connected to the back shell. One PCB is the power supply/inverter board and the other one has the LCD driver ICs.

The power supply/inverter board connects to the CFL lights on the edges of the LCD through a pair of wires. There are two edge lights so 4 wires total.

The other board connects to the LCD through a 34 pin data cable. Here is a picture of the cable:

https://www.electro-tech-online.com/attachments/imag0068-jpg.34858/

Since this monitor was never intended to be taken apart, the cables inside (data cable, CFL cables) are much too short for his purposes so he asked me to lengthen themfrom about 6 inches to about 5 feet.

I was able to make an extention cable for the data cable using some 2mm IDC connectors, a 2mm header and some 1mm pitch ribbon cable. I did not put any kind of suppresor around the cable. Should I have? For the lights, I just spliced and soldered extra wire to the CFL wires to lengthen them. Sounds simple right? Well, neither one of the “extension” cables works as expected.

First off, the data cable does work well enough to display an image but it is not the complete image. It is distorted and there are some horizontal lines that obscure the image. If I had to guess I’d say it seems like there is some interfence or problem with the clock signal that’s causing certain pixels to show up in the wrong place on the monitor.

Second, the CFL lights don’t light up all the way. It appears they only light up about halfway down the length of the tube. The half that is lit is still dim compared to what it should be and the other half isn’t lit at all. I would have expected some line loss with the extra cable length but this seems excessive.

My questions:

What have I neglected to consider?

Short of returning the cables to their original lengths what should I do to make the extension cable work both the CFL and data cables?

Please let me know if any of my descriptions are unclear or if you need some more information. Any insight or ideas would be much appreciated. Thanks in advance.

 

[superfrog]

I am bloody amazed that you get any data trough such a parallel data cable over these kind of distance.

even with a conservative 50Hz, 3*1024*768*8 we are talking about 1 Gb/s. Supposing that the cable is not differential pairs but simple ground we are talking 250Mhz. 500 if it is differential pairs. The driver is meant to deal with 15cm of cable, you multiply the capacitance of it by roughly 10. As I understant you removed the ferrites around the cable, so you are receiving high frequency garbage all over.

As for the Psu cable, if I remember right we are talking a few kV for CFL, so I guess the insulation might not be quite up to the job if it is paired cable.

 

[vne147]

Superfrog,

Thanks for the input. I want to make sure I understand your explanation correctly.

even with a conservative 50Hz, 3*1024*768*8 we are talking about 1 Gb/s. Supposing that the cable is not differential pairs but simple ground we are talking 250Mhz. 500 if it is differential pairs. The driver is meant to deal with 15cm of cable, you multiply the capacitance of it by roughly 10.

I'm a physicist, not a EE so bear with me. The numbers 3*1024*768*8 come from:

3 - the 3 colors, RGB
1024*768 - the total number of pixels
8 - the level of each color (0 - 255)

Therefore, 3*1024*768*8*50Hz = 943,718,400 bits/s ≈ 1 Gb/s

Is this correct?

Even if my derivation is correct I still don't understand why the driver board would be operating at 1 Gb/s. Wouldn't the 1 Gb/s only be necessary if you were pumping all those bits through one wire? That's kind of a side point that I just want to make sure I understand.

That aside, I think what your saying is that the increased wire length significantly increased the inherent capacitance of the wire and now the bits aren't rising and falling fast enough because the driver board was designed to output through a wire with a capacitance roughly equal to what a 15cm wire would have. Is this a correct interpretation?

If this is correct, other than redesigning the driver board is there a way to work around this problem?

As I understand you removed the ferrites around the cable, so you are receiving high frequency garbage all over.

This is not correct. The original cable in the photograph from my first post is still part of the circuit. I plugged the extension cable into the driver board and then connected the other end of the extension cable to the original cable. I connected the female IDC on the extension cable to the female on the original cable using a box header. I then plug the free end of the original cable (the lighter colored connector in the photograph) into the LCD. I left the ferrite suppressor on the original cable in place but I didn't add any new suppressors to the 5 ft extension cable. Should I have? Would it make a difference?

As for the Psu cable, if I remember right we are talking a few kV for CFL, so I guess the insulation might not be quite up to the job if it is paired cable.

It is a paired cable. I'm not sure how much voltage the inverter should be producing. I did measure the voltage on the PCB after I lengthened the cables and started getting the lighting problem and it was ≈ 350 VAC. Can you elaborate a little on how the insulation of the wire is affecting the voltage? Should I use a larger gauge wire or a thicker insulator? What can I do to mitigate this issue?


Thanks for your help.

 

[superfrog]

I (might) have messed up my computation,
It is indeed 1Gbit/s give or take a bit minimum for the interface, possibly much more in bursts so that the refresh is not apparent)

if it is full parallel it is 30Mhz minimum (32 bit data, 1 ground return, 1 clock)
if it is differential we are talking double that because 2 wires are used for each data bit on the interface. (16 bit *2 + 1 pair for the clock)

I just talked about capacitance because it is easy to understand, you would need to add the parasitic inductance and the linear resistance of the wire to have the complete story here. To make it simple, longer wires are much harder to drive cleanly.

I am not an expert for anything that is out of a chip, so I can't help you much for that, but if what is important is the voltage on the receiving side of the interface, then I suspect that having stronger drivers might do the trick.

Getting discrete compenents fast enough for that is probably on the hard side.
A buch of D type ff controlled by the clock signal would do and reclock your signal nicely. You could have as much driving power as you wish to force the bus to be fast enough (anyone that know something about that is welcome to prove me wrong, I am extrapolating from in chip knowledge).

More realistically, as you do not know which wire is the clock a bunch of fast CMOS buffers would probably do, but you would need to know if you interface is differential pairs or straight parallel, and which wire is ground and clock.

As you might be getting at this point, changing an interface driver without the spec of the interface is tricky.

Could you not try to keep your system working with existing cables, it would be so much simpler.

As for the insulation, it is a matter of conductance of insulators, at 50kV for example, wood is a fairly good conductor if I remember well. At 5 Kv, the insulator of a 240AC type cable might be a good enough conductor for HV signal. Hence, you would be loosing most of the power there.
For testing you could just separate your cables so that they have air in between, which has relatively high resistance with arcs only triggering at 30kV/cm if I remenber correctly