Discussion in 'HP' started by 2.0, Sep 3, 2010.
@MobileArtist , when you try to adjust brightness, what happens?
Nothing happens, because my Dragon was never sent to SDD.
I bought a 3700M and was ready to send mine once he solved the "dullness" problem. It was another forum member who sent his.
Dear fellow owners of The Dragon,
My machine have served me well over the past decade, and I fail to find comparable product in the modern notebook market. So I'm still using it as my primary notebook, but lately I ran in a technical problem. The system started crashing all the time, especially when some load is seen on the video card, this and the fact that it runs fine in safe mode, led me, to think that the video card is busted. The board is Nvidia 8800GTS, so I found and bought one from presumably a HP NX9240 system. The problem is when I install this board in the HDX I get only grey screen and no POST.
this be happening due to some BIOS incompatibility or just the seller send me a dead board? The both boards look exactly the same, identical model numbers and pins also.
This is listing for the board I got.
It does sound like the Achilles' Heel of this computer has struck again… The infamous 8800M GTS. Fear not, there are technical adepts on this forum who will help you out.
It's amazing that this computer continues to garner interest, and very cool that this long dedicated thread has picked up some energy.
your model number is correct (454311-001) plus the red glue around the chips is another sign its the right card (other cards did not have that and would not work in the dragon ) but the ram (ddr3) not sure there I thought it was ddr2 ???
You can try: battery out, power out and hold down power button for 30 secs. Then plug in only the power (not the battery) and see if it will post. Also make sure that no external devices or monitors connected.
Additionally, if any of the top panel lights are lit soon after power on, hit fn+f4. At this point, if you have an external monitor, connect to VGA or HDMI to test if GPU is working at all.
Otherwise: full disassembly. Take out the CMOS battery on the board (CR-2032). Reseat the card. Double check all cable attachments for fit and proper insertion. Reinsert CMOS battery, re-assemble.
Hmm, 6-bit per colour will get you only 262,144 possible colours, but 8-bit can have 16,777,216 combinations. That is; each led's brightness is set by the data packet send to it and if this has 6 bits then, with two possible options for each bit (zero or one), you get 2^6 = 64 brightness options for, say, a red led (think of it as discriminate steps in power level used to drive the led). The same is true for blue and green, so you get 64*64*64 = 262K colours. Similarly 8-bit gets you that 16.7M value; (2^8)^3 = 16,777,216.
Quite unusual, but that N201J4-L01 datasheet even has a nice summary table of how the display expects the bits to be arranged for which output level:
Spoiler: N201J4-L01 color data input
Anyway, because of this exponential behaviour in colour counting it is tempting to increase bit rate; a mere 33% more data gets you 64 times as many colours and likewise 66% to 4,096 times for 10 bit (1.07B). But there's a law of diminishing returns here; our eyes won't be able to distinguish between many of these extra combinations. Furthermore, it also depends on the specific colour channel due to our sensitivity to various wavelengths:
More overlap means more nuances can be distinguished (since more input is processed by our brains). So a neat idea would be to have 10-bit for green, 8 blue and 6 red. You would have the combined data throughput of a normal 8 bit display, except resulting in a better image for the human eye:
Errr ... sidetracked, a bit .
Yes, that would work.
Thing is, that panel doesn't actually have native PWM control; these early CCFLs used variable voltage and/or current to have working backlight control. The LVDS configuration even lacks a PWM pin and, actually, all backlight pins entirely. Reason is that its backlight control is the responsibility of the inverter and, for this model, that is a device separate from the display. The sheet even states something to the tune of 'not supplied by Chi Mei'.
See if you make a photo of that backlight inverter and of the lcd cable itself. It should have a split-off from the mb to both lcd and the inverter. The inverter's cable ought to have 4 pins or more and one of those will be the pwm signal. There'll be a converter chip on that board that processes the pwm range into a voltage or current range and knowing which chip it is means you'll know what range to send to it. Conversely, knowing which panels were used in the systems that were running these FX3700Ms will tell you the range the gpu is transmitting over its pwm pin.
"…The 8730w comes equipped with your choice of a standard 17.0″ anti-glare widescreen display or a “DreamColor” RGBLED backlit screen with an 8-bit panel that can display 16.7 million colors…"
"…For graphics, HP included an Nvidia Quadro 3700M series video card…"
Analysis, well done.
Yes, the LCD cable branches off to the inverter - 12 pin connector. The inverter controller chip is a BD9828 7k81. I had a spare LCD cable and inverter in a box.
It's great to have another technical adept on this thread. I'm surprised by your observation that the WUXGA does not use PWM, which would mean that this screen is flicker-free. I've always thought that PWM was a bad invention.
Separate names with a comma.