Have you seen this color problem before? Repair of an Advantest R3477 screen

[DaJMasta]

I've got an Advantest R3477 spectrum analyzer that seems to be perfectly happy, albeit a bit noisy, but the touch screen on the front shows a strange kind of horizontal shift of the red color channel to the right, maybe a couple dozen pixels, but not consistent everywhere.  The analyzer is usable the way it is, but the text in particular is hard to read, and I've verified the VGA output in the back the best I can (only RGB capable monitor around is old enough to not be able to manage the resolution, but in the boot screen it shows no color fringing in the misaligned pixels).

So far, I've disconnected and reconnected the cables a few times, reworked the solder joints on the connector on the LCD board, and just did a little reflow of that portion of the board with the hot air gun.  There are a couple of connectors that the signal goes through between the CPU module and the screen, but at least the last stage one has been checked and wiggled to no avail.  The panel is a LTM08C351F by Toshiba and there is a datasheet that at least gives a block diagram and connector pinout, but I figured before probing the signals going in on the scope, I'd see if someone had a good idea of what was going wrong.  http://html.alldatasheet.com/html-pdf/143915/TOSHIBA/LTM08C351/296/1/LTM08C351.html

I think it's some kind of signal quality or timing issue, but I was wondering if someone here had seen this before - I feel like it's the kind of thing that could be a very simple fix if you know where to look, and I haven't worked on many LCD boards.

[EPTech]

Hi there,

Looks like a problem with a green column driver. Notice the green in the text on the middle-left has two different shades of green. These drivers are often integrated on the flex of the display, so you probably need to replace it.

[capt bullshot]

From the datasheet, the panel is driven by a single clock and many parallel data lines. Your VGA output would be driven by a (RAM)DAC chip, that basically takes the same inputs (clock and many parallel data lines), but maybe at a different timing. So proving the VGA output does not necessarily prove the internal interface to the LCD (but it is quite presumably supposed to be OK).

So, if you can, first check the clk line and maybe the MSB of each colour data lines with an scope for timing - try to look for an visible edge (black / white) that is supposed to change all colour data lines at the same time, you should try to trigger on the HSync or VSync signal.
If this looks OK, then you'd go for a replacement panel ...

[DaJMasta]

Took a bit more of a look, in terms of system arrangement, it looks like the graphics processor sends an LVDS signal offboard, which goes through a backplane, which goes through a front panel board that includes an LVDS to FPD converter chip.  I probed pin 5 of the display to get the compound synchronization pin on the scope, and it seems to be pulsing more or less like I expected - a series of shorter pulses in a block separated by a gap before the next block.

The part that I had reflowed before was the timing controller, so while this may still be an issue, it shouldn't be bad joint related.  I assume there has to be chips on the flex going to the panel, since it doesn't have nearly enough connections to it to control even all of the rows, so I assume those would be unfixable with my equipment/parts.

Do you think there is any value to trying to replace the LVDS converter chip on the other board?  I can't say I have evidence that it's not working, but it sounds a hell of a lot simpler and cheaper than swapping out the panel and touchscreen.

[capt bullshot]

Do you think there is any value to trying to replace the LVDS converter chip on the other board?  I can't say I have evidence that it's not working, but it sounds a hell of a lot simpler and cheaper than swapping out the panel and touchscreen.

Does the LVDS have one lane per colour? If so, check these.

Yes there are chips located somewhere within the panel. Many panel have a small PCB attached to their backside, and a lot of flex PCB mounted driver chips on two sides of the LCD glass. This is where I'd locate the failure (either on the panels internal flex boards, or on that small PCB). In general I don't like the idea of swapping chips just to see if they are bad, I'd try to prove the converter chip by scoping its output signals. Maybe one is missing, that can be a solder joint or a broken pad driver, but I'd expect the whole chip to go bad (so you wouldn't see anything on the screen).

[DaJMasta]

One LVDS per color to the converter chip.  I did some proving on the individual color data lanes for all three colors and didn't find what I would suspect to be a problem - timing issues, inconsistent high or low voltages/etc, since it's not a 100% clear picture slid over but varies slightly over time (always the same side and same general distance, but varying continuously, slightly.)

So I opted to just replace the screen - the replacement module wasn't too bad, but an appropriately sized touch panel for it was as much as the display... and I'll need to install it onto the screen module myself, though this doesn't seem difficult.

The parts are a month out as I couldn't find a US supplier within almost double the price, so I'll swap it when it arrives and see how it does.  Thanks for the help!

[DaJMasta]

Well I ordered the screen, it sat around for a month or more, and installed it eventually.... but it didn't solve the problem.  So I ordered a DS90CF364AMTD, the IC that does the LVDS to LVTTL translation on the board with the power switch and LED in the front, and swapped in the new one.... same issue.

It looks like the glitch I'm seeing is actually part of the signal coming from the processor board itself, so while I don't expect to replace it, if I find one for cheap I may swap it and see if it makes a difference.  The form factor is standard, so I'm sure there are options, but since the OS is a cut down windows XP, it probably only includes the drivers for the board being used, and it may be a real pain getting it to work with a different CPU module.  I don't have a differential probe and don't really have experience probing LVDS, so at the moment I am not trying to figure out if there's some issue with that signal.

Though it occured to me a while back, the error I'm seeing has an interesting quirk: when there's enough green intensity in a row (think average green level), then the whole row's colors come right.  When the green value is very high, the green value for the pixel is fine.  When the row doesn't have a lot of green and the pixel doesn't have a lot, it doesn't work.  You can see it in the bands of good color that appear in rows with a lot of bright colors (yellow in this pic), but I wonder if something like swapping the color palate of the window to something bright would actually improve overall color because of this quirk.

For now, at least, this will go unrepaired.  I'll post again if I find a cheap CPU module replacement to test.  I don't believe there's another chip off the CPU board that touches this LVDS signal.

[james_s]

That's certainly strange, and you said the VGA signal is fine? Makes me wonder if there's a bad ground somewhere causing some interaction between the parallel data lines for the colors. Maybe a timing issue? These sort of displays are usually pretty simple, there will be groups of parallel lines for R, G and B values and a pixel clock that latches a set of values into each pixel then moves onto the next. It's bugging me a little just wondering what could be wrong.

[DaJMasta]

So that post got me thinking again... when I checked the VGA output, I didn't really have a device capable of displaying the signal.  Best I could scrounge up was a 640x480 max color display using a 9 pin connector (from an amiga!)  And while I didn't see any color alterations in the jumble of pixels it displayed, it wasn't necessarily a good test.

So I tried to find a way to get a VGA signal on one of my displays (HDMI, DisplayPort, DVI)... and ebay came through for about $5.  Hooked it up today and verified the output - the VGA output has fine color, whereas the LVDS driving the panel in the front does not.


I took a loot at my images of the CPU board to try to find a LVDS repeater or something that could be bad if the signal is coming from the module fine.... but I don't see it - there's one chip I can't read next to the FPGA that handles the general interface stuff (including fan monitoring), but if I remember right, that's just a micro.... and since it's LVDS into the front board before it changes it to LVTTL to drive the panel.... and I've already replaced that chip..... I do think this means the issue is on the PC module, and that would mean that the VGA and front screen outputs are on separate channels coming out of it.  I will look into replacing the CPU board.... eventually.

[DaJMasta]

Found it and fixed it!  I was working on replacing the ETX computer module (and the board to board contacts can be a bit dicky at times, and I found a little chip on the underside of the board near the backplane connector.  I had accounted for the function of quite a few others, but it looked about the size and shape of the LVDS to LVTTL translator on the board going to the LCD data cable, so I looked up the part number.  It was an LVTTL to LVDS transmitter.... a DS90CF363A.  Digikey only had the B variant, but they seemed to be the same function and package, so I bought one and replaced it.

The colors are fixed!  And LVDS is fast enough that a little bit of stray flux on the first attempt cause some intermittent full-screen color glitches 


Why exactly they need to use one of these when the CPU module should be able to drive a flat panel directly.... I'm not sure.  Since I was expecting the signal to come straight from the CPU board, I hadn't even really considered that there would be a transmitter chip on the carrier board, but here we are.

[james_s]

Nice job! It's always satisfying to hear the result.

Maybe they used the same CPU board with multiple panel types, or thought they might? Maybe it has something to do with the need to simultaneously output VGA and drive the internal panel, who knows.

[DaJMasta]

Could be, I'm actually replacing it with a R3671 which uses exactly the same SOM-4470 mainboard module, though the carrier board is different (and since it's still a 700MHz P3, it's a lot more sluggish on the R3671 with the extra demodulation and analysis options.... so it's being swapped).


Actually, there's actually 4 passives and a little note on the bottom of the board that give you instructions for configuring different FDD/LPT chips on the module, so perhaps there was an earlier variant used that didn't support LVDS directly.

[james_s]

They may have previously used a TTL interface panel, those were common at one time before LVDS started being used.