Sunday, May 16, 2010

What is causing the DV Series to fail prematurely?

I have seen dozens of 'solutions' and fixes for the DV series and none really address the root causes. Others damage or ruin motherboard. Below are some of the weirdest solutions I have seen:



1: Placing a penny between GPU and Heatsink.
2: Using a copper shim (only slightly larger in area than penny)
3: Excessive thermal compound
4: Wrong type of compound.
5: Placing a tea candle (in aluminum holder) on GPU
6: Putting Motherboard in oven
7: Pressing down on GPU real hard
8: Reflowing without flux?!


Outline what each does and why it still fails later.

1: This will increase the pressure on the GPU when held down with 5 screws, a temporary solution, but does not remedy. Provides no greater heat dispersal and only transfers heat through penny. Metal fatique may damage the 5th retainer screw and not support the GPU.
2: Copper shim does the same as above.
3: Spreads heat over a larger area but does little to cool.
4: Does nothing
5: Can accidently cause more damage, and is only a temporary solution, because the holder never reaches the tempetature that RoHS solder melts.
6: Could ruin board, plastics, labels and fan/USB and other connectors. Dangerous & Risky
7: A temporary solution that fails later because the two surfaces only contact, not bond.
8: Sometimes work, but solder may be tacky under GPU and no way to tell. Not recommended.


Typically it starts like this.

After about a year of average daily use (and shortly after warranty expires) the vents on the heatsink would begin to clog. This lint builds up over time. If you have new carpet, rugs, lots of activity, kids, pets, dust and smoke, the time may be shortened.

The operating temp starts to rise over time because of the lack of ventilation. Chips run hotter, fan stays on longer drawing even more lint. Eventually the solderballs lose contact with pad under GPU. Any oxidation prior to the manufacturing process only attributes to loss of contact.

This can also happen with the Southbridge chip if it too had oxidation, causing various problems such as no WIFI, keyboard, touchpad and USB not working.


The Solution:

1: Reball the GPU (if possible) if not continue...

2: Reflow the GPU with proper tools, flux, and temps

3: Verify silicone pad under GPU heatsink


4: Replace aluminum pad under CPU heatsink with copper and apply thermal compound.


5: Disassemble Heatsink and clean


What doesn't work.


Holding down the power button while battery and AC removed. Clearing CMOS. This only re loads the factory CMOS info from ROM


Water based flux. Usually clear and sold on ebay to fix X-boxes
The liquid solution is too thin and heat used evaporates the liquid needed before the de-oxidation process begins. This is used in manufacturing for thin component motherboard during the wave solder operation. Works in manufacturing for thinner boards, but not for this process. Plus sometime conductive and may create short under GPU or chips if all is not dried.

High viscosity flux. Muratic Acid based. Sometimes listed as Liquid Rosin Flux but too thin and does not stick to pad during de-oxidation process

Blowing lint back into laptop with can of air. The lint needs to be removed not blown back into the machine, it only collects back into the
fan/vents.


What Does Work.


Reflow GPU for video issues
Clean Heatsink Fan - Always when the system is opened.

Rosin Flux petroleum based sold in paste form. Usually in a semi-gel state, but liquifies when it gets warm. Sticks to pads and when heated, and slight smoke indicates de-oxidation is happening. When the flux smokes it is also cleaning the contact pads making a larger area for whetting. I have done this and it works!  It is non-conductive and will not create shorts under chips.


How did I test?
I took an old motherboard where exposed
copper was oxidized and discolored, then
applied rosin flux paste. Heated the area to
tempetature that solder melts and when it
started to smoke, removed heat and allowed
to cool. After cleaning the area with acetone
the pad was bright like new.


ASK your tech "What are his qualifications?"
A+ Tech is only certified to replace peripherals and reload Windows, drivers, no soldering is taught.

Computer Techs - 2 year degree in computers, but no electronics training.

Geeks - only have experience, no formal training. Formal training is needed to be able to ask questions when you have them. Otherwise you guess at your own questions. Similar to hackers but no hands on hardware.

Hackers- The true hacker, (not malicious programmer). They build things and have experience soldering, only some of them have formal training.

Electronic Techs - 2 year degree and some soldering.
 

12 comments:

  1. today i read your this paragraph.your guidence is very helpfil for me.i am also have a Computer 2 year degree.and 1 year laptop repairing traning.

    ReplyDelete
  2. This comment has been removed by a blog administrator.

    ReplyDelete
  3. Mr Mayo,
    I bought several of these machines on eBay after watching a reflow video on eBay. Figuring this to be an easy money maker, I jumped in with both feet. Little did I know that I was I'll prepared for the adventure. I attempted reflows on two boards, and may have trashed them both. Ihave a third board which acts like it has a power issue, and I haven't tried anything on it yet. Board 1 is a dv9000, board 2 is a 2500 and board three is another 9000 with no repairs attempted. Board acts dead after reflow attempts on the video and south bridge. Board 2 turns on and immediately off. I removed the video chip from it and ordered a new balled chip. Now to where I would like some help. :) with board 1, will replacing the video and south bridge help bring it back? Same question for board 2. On board 3 I noticed scorching on the power connector. I used the DMM on the chips you recommended another poster look at, but I'm unfamiliar with the pinouts of the chips and what you mean by direction. I am a controls engineer with many years experience working on computers. I was trained by the US navy in component level troubleshooting but that was many years ago. :) I think with a nudge in the right direction I could get these three running, and u seem to have a very deep knowledge of the electrical principles at work here. I'm hoping you can help!
    Mike

    ReplyDelete
  4. Catching up on previous posts missed----

    Andy: Replace with exact number to avoid adding to the confusion, and diverting from original task. If you use the same number, the unit may fire up, or indicate where/what the primary issue is. From your description, if the correct FET was used, and now the ringlight is flashing, that would indicate a capacitor in the front is is shorted. Typically in most model the FDS4479 is on top (CPU facing up), and the AO4407 is on the bottom.

    Mike: Seems a bit much to ask in a post, but I will attempt to answer. 1: sometimes you can over heat the GPU or Southbridge and cause many of the solderballs to bridge together under the chips, this effectively shorts the board. Though I have had some success bringing them back to life with using water based (WB) flux and reflowing. I don't care for the WB flux because it boils at a lower temperature and can be misleading in visually indicating process is working. Though it may work.

    2: Reballing in my situation is the last hope. Many can be saved without reballing, and only after above board has killed a board.

    3: Testing FETS: if you stick to the middle pairs with your meter probe, you should be fine. see the
    Tips on testing MOSFETS. FEB 13 page In most cases there is a dimple at pin 1 then pin 5 is across from Pin 4 (the gate). But that is only on single type MOSFETs. NEVER BRIDGE PIN 3/4 If you are still having trouble with them, most FET producers have the datasheets on their sites.

    AO semiconductors : www.aosemi.com
    Fairchild: www.fairchildsemi.com
    or sites such as Newark Electronics (www.newark.com) have a matrix of cross reference numbers. In most cases lately HP/HannStar have been using AO & Fairchild FETs.

    Also watch for boards that are using small dots of black hotglue to hold GPU's. That glue does not hold the chip down and overheading will cause the chip to rise and solderballs bridge and short. The boards using red hotglue around the GPU will have to have some glue removed opposite the opening. This allows the flux to flow under the chip. Well explained in the forthcoming DVD.

    ReplyDelete
  5. Scorching on a board power connection is indicative of a short.

    Example: If the FDS6679 is shorted, it will cause the charger to shut off, and you have to unplug from wall and mobo and wait 5 sec, then plug back into wall- BUT NOT INTO MOBO.

    It is only an indicator. At that time replace the FDS4479. Then try again.

    Most Intel board problem is with ceramic caps.
    Most AMD problems are GPU or GPU/Southbridge combo.

    I hope that helps some.

    ReplyDelete
  6. Mr Mayo,
    Thank u for ur prompt reply. On board hannstar daoat5mb8e0 rev e I found fet 6679 read open when diode testing the pins u recommended. On board daoat7mb8e7 rev 8 4407 read open between the recommended pins. I'm assuming replace?

    Mike

    ReplyDelete
  7. C: you have to check both directions.

    Only one direction should read infinity. If the other direction reads low resistance, replace that one.

    ReplyDelete
  8. When u say 'directions' I'm assuming u mean test with + on 2/3 then swap leads so + is on 6/7? Your post on testing these indicated to test bases on N or P gate polarity. Also, with testing the capacitors using an ohms function I see fluctuations in the resistance reading. My meter has a cap mode which gives a reading of .21 or whatever. I believe that is telling me the cap is ok.

    You also stated that improperly heating the gpu can cause pooling of the balls and short the gpu. Is this a destructive result or will correcting the shorts restore operation. I'm unsure how applying flex and reheating would fix that issue

    Thanks,
    Mike

    ReplyDelete
  9. C: When I say direction, it means From and To. Example, if POS lead is on 2/3 and NEG on 6/7, switching would put NEG on pins 2/3 and POS on 6/7. There was no mention of setting to resistance or cap mode. Only diode mode because (in most cases) a MOSFET is a diode with a switch. Sometimes they have zener diodes in the circuit too as with the FDS (Fairchild) MOSFETS.

    Shorting the GPU (when solder pools develop) will render the board dead and will require reball or replacement. On occasion I had once used water based flux and it corrected the pooling. But I think the result was a fluke, so I don't recommend it. This was after it had pooled. Very tricky times and temperatures.

    ReplyDelete
  10. C:

    Correction: Testing caps
    With the meter set to Resistance, you should get infinity. This is just a general test and values may read resistance in one direction depending how it is used. More so it is to see if the cap is shorted.

    My meter does not have capacitance mode, so I am not familiar with testing it that way. I use mine only to check for shorts. If values are not constant, it may mean that the cap (or another part in the path) is leaking/draining improperly.

    ReplyDelete
  11. Mr Mayo, thanks for all of your time! I will keep u posted as to my progress withthese three boards. I put flux on the gpu and it's memory chips as well as the southbridge last night and heated everything up hoping that it may correct the pooling but I haven't checked to see if it posts yet. I removed the gpu from one board and ordered a new one with balls already on it. My third board(the one u haven't tried anything with yet) wasn't tuning on because I forgot the ribbon cable. Lol so that one powers up but no video. I'm debating trying the flux and heat method with it, but I suppose it can't get any worse. I pulled a small animal out of the cooling fan on it so I'm assuming they ran it until it dies completely. Lol

    ReplyDelete
  12. Mr Mayo,
    Still no further along with these boards. After attempting a reflow on board 1, it turns on and then right back off. Board 3 used to turn on backlight and after reflow doesn't even do that. I heated the board enough to remove some of the red glue from around the chip, then coated the edges with flux. Heated the chip to around 215C several times. Added attic silver, reinstalled fan assembly with the ribbon cable lol used a battery to provide power. Hooked up the LCD panel and hit the button. Backlight doesn't turn on, fan runs at slow speed. Couple LEDs turn on but still no video. I don't know what I'm doing wrong here and would appreciate any suggestions.

    ReplyDelete

I cannot repair laptops for you via the internet, I can only answer questions related to posts. This is because I have not had experience on every motherboard that is out there. Bear with me.

If you do need it repaired. Contact me for quote (US/CDN Only)

Note: Only a member of this blog may post a comment.