Laptop design flaw investigation

[alxnik]

Hello Everyone

I was given a laptop to try to service which seems to have a design flaw.
The laptop is the Acer ICONIA 6120 Dual-Screen which although is a normal laptop, instead of screen/keyboard combo has 2 touch screens and a virtual keyboard. The problem with it is that the upper screen registers so called "ghost touches" (more info below). I researched the thing in the interwebs as well as I have done my own tests. Many people seems to have the same problem so I almost certain that this is a design issue, however acer seems to quietly deny this issue so the manufacturer is a dead end. People haven't found any fixes except from a guy who goes around forums asking for 20 euros to give the fix. Since I don't trust this guy, I will not go with this solution.

Now to the problem itself. The upper screen registers touches when the fingers of the user are not even close to the device, in a rate of about 30-40 touches/sec (roughly) which means that it gets crazy after a while. Below are the results of my tests on this.

1. When cold booted, the problem needs 3-4 hours to manifest (at least)
2. When the problem manifests, a hot reboot, a boot to a different OS (as in linux, not windows) doesn't fix the problem so it doesn't seem a software issue
3. The problem is only on the lower edge and the right edge
4. Installing/Uninstalling the acer drivers makes no difference
5. When the problem manifests, shutting down, removing/inserting the battery and rebooting, seems to fix the issue for about an hour or so.

Some people are suggesting to add some insulation between the touch interface and the screen but there is anecdotal evidence that this doesn't work.

In conclusion, I'm not afraid to open it up and do additional tests or mods in order to fix it, however I have no idea even theoretically, what could be the problem. I'd love to hear some (even wild) guesses on what could be the problem, in order to have a path to follow. Any additional tests I could do would be welcome

[Stonent]

Some touch screens have firmware that feeds the touch controller. I'm not sure how they could deploy an update to fix an issue like this. My guess if it takes a few hours is some kind of overflow to occur in the controller and the data wraps around to an invalid area. Could be bad use of pointers in the original C code.

There's a possibility that the bios on the computer uploads the code at power up. If that were the case Acer could fix it easily.

[peter.mitchell]

Most likely a grounding issue, very common on capacitive touchscreens.

[JoeO]

Do the free stuff first.

Just let it sit there powered on.  After 4 hours, does it fail?

Run it without the battery.  Does it fail with out the battery.

Take out one memory SIMM if you can.  Does it fail?  If not try running it without the other SIMM.  Does it fail?

If it is say 3 years or older, if you can get to the processor, try some new thermal paste.  This may cost you a few drachmas.

Let us know if it gets fixed.

[Stonent]

Most likely a grounding issue, very common on capacitive touchscreens.

Never underestimate the ability of RoHS solder to break something.

[alxnik]

Some touch screens have firmware that feeds the touch controller. I'm not sure how they could deploy an update to fix an issue like this. My guess if it takes a few hours is some kind of overflow to occur in the controller and the data wraps around to an invalid area. Could be bad use of pointers in the original C code.

There's a possibility that the bios on the computer uploads the code at power up. If that were the case Acer could fix it easily.

Altough it sounds plausible, wouldn't a hot reboot fix it? Also it has the latest bios so I'm thinking that if it was easy, acer would have fixed it by now (1 year from latest bios). Lastly, I think it has something to do with enviromental factors too (heat?), as if I cold reboot it, it only needs 1 hour to restart the problem.

Most likely a grounding issue, very common on capacitive touchscreens.

I read that other people are thinking the same thing. So what is a grounding issue to a laptop screen? How do mod the thing to fix it?
Interesting thing in the design of the whole setup is that the laptop has a power supply followed by a DC power filter in a seperate enclosure. This gives me an impression that at the last moment they found out that noise from the input was unacceptable. Might this affect the touch in any way?

Do the free stuff first.

Just let it sit there powered on.  After 4 hours, does it fail?

Run it without the battery.  Does it fail with out the battery.

Take out one memory SIMM if you can.  Does it fail?  If not try running it without the other SIMM.  Does it fail?

If it is say 3 years or older, if you can get to the processor, try some new thermal paste.  This may cost you a few drachmas.

Let us know if it gets fixed.

Haven't tested these yet, the laptop is less than 2 years old and my guess is that these problems would manifest in other ways too. Nevertheless, I guess it doesn't hurt to try

[Towger]

Haven't tested these yet, the laptop is less than 2 years old and my guess is that these problems would manifest in other ways too. Nevertheless, I guess it doesn't hurt to try

Less than 2 years old + Grease still in Europe = Still in warranty. 

[AndyC_772]

Interesting thing in the design of the whole setup is that the laptop has a power supply followed by a DC power filter in a seperate enclosure. This gives me an impression that at the last moment they found out that noise from the input was unacceptable. Might this affect the touch in any way?

PSU noise certainly could affect the touchscreen, though it's highly unlikely this would be in a way which gets worse over time. I'd expect it to either work or not work.

More than likely, they designed the product without the filter, went EMC testing, failed, and had to come up with a fix at the last minute. That's the usual reason why ugly, obviously bolted-on filters end up being used in products.

It's easy to test, though. Run the laptop until it starts to misbehave, then pull out the power lead and run it off the battery for a while. If it still misbehaves, then it's not power supply noise. (At least, not noise from the external adapter... it could still be internally generated noise, but there's really no good reason why this should change with time).

Does it make any difference what the state of the battery charge is? Some mains supplies are actually quieter on full load than they are on part load. Charging a flat battery might cause more current to be drawn from the mains supply, which would put it into a quieter state. As soon as the battery is charged, the current draw reduces and the adapter gets a bit noisier. Highly unlikely, but not completely impossible.

What happens if you power the laptop from a different supply? Could you power it from a quiet lab supply?

I wonder whether there's some ongoing self-calibration going on which is eventually failing. There must be a sensitivity setting somewhere in the system; maybe there's a periodic calibration being carried out which is affected by temperature? When it starts to misbehave, stick it in the fridge for half an hour and see what happens.

[kxenos]

You said the problem appears always in the same spot? Does the problem appear when you have the device opened up? Also, is it possible to interchange the two touchscreens? The description sounds like temperature problem to me.

[ablacon64]

I'm about to get one of those to try to fix that problem, client will bring it in 2 weeks, for now the information I have is only that the touch screen goes crazy and that is a common issue. My guess.... Cold solder. I've seen this same issue on tablets with DC Jack cold solder problems. But since I'll get my hands on it only in 2 weeks, I can only wonder...

[alxnik]

Interesting thing in the design of the whole setup is that the laptop has a power supply followed by a DC power filter in a seperate enclosure. This gives me an impression that at the last moment they found out that noise from the input was unacceptable. Might this affect the touch in any way?

PSU noise certainly could affect the touchscreen, though it's highly unlikely this would be in a way which gets worse over time. I'd expect it to either work or not work.

More than likely, they designed the product without the filter, went EMC testing, failed, and had to come up with a fix at the last minute. That's the usual reason why ugly, obviously bolted-on filters end up being used in products.

It's easy to test, though. Run the laptop until it starts to misbehave, then pull out the power lead and run it off the battery for a while. If it still misbehaves, then it's not power supply noise. (At least, not noise from the external adapter... it could still be internally generated noise, but there's really no good reason why this should change with time).

Does it make any difference what the state of the battery charge is? Some mains supplies are actually quieter on full load than they are on part load. Charging a flat battery might cause more current to be drawn from the mains supply, which would put it into a quieter state. As soon as the battery is charged, the current draw reduces and the adapter gets a bit noisier. Highly unlikely, but not completely impossible.

What happens if you power the laptop from a different supply? Could you power it from a quiet lab supply?

I wonder whether there's some ongoing self-calibration going on which is eventually failing. There must be a sensitivity setting somewhere in the system; maybe there's a periodic calibration being carried out which is affected by temperature? When it starts to misbehave, stick it in the fridge for half an hour and see what happens.

PSU is in the inprobable bin for me, although I am in a "you never know" state. You have just shoved it further in the inprobable bin. Nevertheless interesting info, I think first order of the day is to put the psu on a electronic load and measure it on the oscilloscope, even as a test for test's sake.

You said the problem appears always in the same spot? Does the problem appear when you have the device opened up? Also, is it possible to interchange the two touchscreens? The description sounds like temperature problem to me.

Quite possible to be a temperature issue, although I can't see a heat source on the lid/upper screen. Haven't opened it up yet because I have a heavy hand on these things and end up breaking plastics here and there. I want to do it after I gather all the possible info. Changing the 2 touch screens isn't possible as they are attached to the respective upper and lower frames.

I'm about to get one of those to try to fix that problem, client will bring it in 2 weeks, for now the information I have is only that the touch screen goes crazy and that is a common issue. My guess.... Cold solder. I've seen this same issue on tablets with DC Jack cold solder problems. But since I'll get my hands on it only in 2 weeks, I can only wonder...

Would CERTAINLY like to hear from you in 2 weeks time if I haven't fixed it until then. Cold solder is certainly one of the things to watch for, but something must be very wrong in the QA process to be so common

[senso]

If it manisfests over time/with hot laptop, might be a cooking cap in the PSU/local decoupling to the chip, or dodgy tantalums..

[ablacon64]

Cold solder is certainly one of the things to watch for, but something must be very wrong in the QA process to be so common

Well, most laptop motherboard problems I've seen are due to cold or insufficient solder. There is a guy on youtube telling people to open the Acer Iconia and put a double tape to pressure the touchscreen against the LCD frame but some viewer posted that the problem came back after a while, so I don't believe the tape solved the problem in the first place, I guess the guy changed the screen status just by manipulating it. Sometimes it happens when we disassemble a dead laptop suffering from BGA cold solder an it starts working again, just because we removed the motherboard out of the chassis, but that doesn't mean the problem is solved.

[vk6zgo]

I'm a hardware person,so may be barking up the wrong tree,but can you swap the two touchscreens in software,so you can check if it really is a hardware fault,or something silly in how the Computer reads the  particular touchscreen?

[free_electron]

this smells of a static charge accumulation problem...

capsense touch screens work on charge dispersal principle. if the common charge on the display rises they re-calibrate. There may be a point in time when the calibration can no longer compensate... that corner/ side may have a bad grounding tab.

you say that a full powerdown / battery removal solves the problem.  i think it is not the power removal in itself or the battery removal , but something you do physically to the laptop at that point. like for hotboot you do not close the lid , but for a battery swap you do as you need to flip the machine upside down...

closing the lid may have the main machine body touching the top rim of the screen discharging it enough so that this works again. or simply because you flip the machine upside down it discharges through the work surface...

do a simple test : when the problem occurs. close the lid , then flip the machine upside down so the top panel touches the table. let sit for 20 seconds. then flip machine back over , open lid and power up.
see what happens...

[ablacon64]

capsense touch screens work on charge dispersal principle. if the common charge on the display rises they re-calibrate. There may be a point in time when the calibration can no longer compensate... that corner/ side may have a bad grounding tab.

Makes sense, one of the things that client told me is that the problem goes away for a few hours when he opens the laptop completely (both LCDs flat) and he cleans the LCDs with an anti-static cloth. I think it's not the cleaning that makes it go away, but the pressure, thus a grounding issue. Anyway, just wondering again since I've never seen this laptop personally.

[alxnik]

Cold solder is certainly one of the things to watch for, but something must be very wrong in the QA process to be so common

Well, most laptop motherboard problems I've seen are due to cold or insufficient solder. There is a guy on youtube telling people to open the Acer Iconia and put a double tape to pressure the touchscreen against the LCD frame but some viewer posted that the problem came back after a while, so I don't believe the tape solved the problem in the first place, I guess the guy changed the screen status just by manipulating it. Sometimes it happens when we disassemble a dead laptop suffering from BGA cold solder an it starts working again, just because we removed the motherboard out of the chassis, but that doesn't mean the problem is solved.

There are mixed reviews about the youtube guy.
However I started a test which runs for some hours now, I cracked open the lid frame (which actually is pretty simple) seperating the touch from the screen. This guy is using non conductive tape so I am guessing that by seperating them I simulate his solution. I've left it for some hours, will check it later again.

this smells of a static charge accumulation problem...

capsense touch screens work on charge dispersal principle. if the common charge on the display rises they re-calibrate. There may be a point in time when the calibration can no longer compensate... that corner/ side may have a bad grounding tab.

you say that a full powerdown / battery removal solves the problem.  i think it is not the power removal in itself or the battery removal , but something you do physically to the laptop at that point. like for hotboot you do not close the lid , but for a battery swap you do as you need to flip the machine upside down...

closing the lid may have the main machine body touching the top rim of the screen discharging it enough so that this works again. or simply because you flip the machine upside down it discharges through the work surface...

do a simple test : when the problem occurs. close the lid , then flip the machine upside down so the top panel touches the table. let sit for 20 seconds. then flip machine back over , open lid and power up.
see what happens...

From all the replies, I can definitely say that yours fit my test results so far best. My next tests will be along this theory for sure.

I'm a hardware person,so may be barking up the wrong tree,but can you swap the two touchscreens in software,so you can check if it really is a hardware fault,or something silly in how the Computer reads the  particular touchscreen?

Being a software guy, I think I have excluded all the possible software faults, although I don't think it's possible to swap the touch screens in software.



By the way, I got around testing the PSU, and at 50%-90% loading it had 1Vpp ripple before the filter and a nice <100mVpp after the filter (the filter seems mandatory). Even at 90% there seems to be no serious voltage dip.

[amyk]

Could it be EMI from the CCFL inverter that's charging up the screen?

By the way, I got around testing the PSU, and at 50%-90% loading it had 1Vpp ripple before the filter and a nice <100mVpp after the filter (the filter seems mandatory). Even at 90% there seems to be no serious voltage dip.

1Vpp on a 20(?)V supply is only 5%, and it's going to be regulated down to far lower voltages by a bunch of buck converters so that's reasonable.

[alxnik]

Could it be EMI from the CCFL inverter that's charging up the screen?

By the way, I got around testing the PSU, and at 50%-90% loading it had 1Vpp ripple before the filter and a nice <100mVpp after the filter (the filter seems mandatory). Even at 90% there seems to be no serious voltage dip.

1Vpp on a 20(?)V supply is only 5%, and it's going to be regulated down to far lower voltages by a bunch of buck converters so that's reasonable.

I like the EMI theory. I will go to the shops later to get some tape of some kind. I think I will add some copper tape in the mix to test it, although I haven't checked if it is CCFL or LED.

Yes it is a 20V (19.6 to be exact) PSU, and most probably it doesn't directly power anything, but nevertheless seeing 1Vpp on the scope is kinda ugly/stressful 

[Stonent]

I don't think anyone uses ccfl anymore.

[fluxcapacitor]

Could it be EMI from the CCFL inverter that's charging up the screen?

By the way, I got around testing the PSU, and at 50%-90% loading it had 1Vpp ripple before the filter and a nice <100mVpp after the filter (the filter seems mandatory). Even at 90% there seems to be no serious voltage dip.

1Vpp on a 20(?)V supply is only 5%, and it's going to be regulated down to far lower voltages by a bunch of buck converters so that's reasonable.

I like the EMI theory. I will go to the shops later to get some tape of some kind. I think I will add some copper tape in the mix to test it, although I haven't checked if it is CCFL or LED.

Yes it is a 20V (19.6 to be exact) PSU, and most probably it doesn't directly power anything, but nevertheless seeing 1Vpp on the scope is kinda ugly/stressful 

The copper tape might do the trick.Put it on the egde of the panel and onto the side of the hinge/support.Fixed an old pos terminal some years ago that had similar problems with calibration.

[ablacon64]

There is a topic in notebookforums saying it's software related: http://www.notebookforums.com/t/241914/acer-iconia-6120-6886-484g64ns-real-ghost-touch-solution

[alxnik]

There is a topic in notebookforums saying it's software related: http://www.notebookforums.com/t/241914/acer-iconia-6120-6886-484g64ns-real-ghost-touch-solution

I've read through it and according to my experience they are not correct  . The acer software seems to delay the problem for a while (drivers which calibrate the thing?) but it certainly was not fixed.

After leaving the laptop on for ~20 hours with the touch seperated from the screen, no ghosting appeared. About 4 hours ago I added some aluminium tape along the edges which I grounded. So far no ghosting. I am leaving on a business trip so I left it on and will check on it again in 3 days. If it is either EMI or bad grounding I think it should be fixed. Fingers crossed!

[amyk]

I was curious about what touch panel controller they were using (try to cross-ref with other models/manufacturers and see if they don't have this issue) so tried to find a schematic and apparently one exists but only on sites wanting to sell one or foreign-language forums you have to figure out how to download from...

[fluxcapacitor]

There is a topic in notebookforums saying it's software related: http://www.notebookforums.com/t/241914/acer-iconia-6120-6886-484g64ns-real-ghost-touch-solution

Some forums say its a software problem,some say ts a hardware problem.Some claim to have fixed the problem ,only for it to return a few weeks later .

Here is the service manual/guide for anyone interested :

http://uploaded.net/file/4a19i3sx

EDIT: i have tested the file for viruses and spyware, but i still wouldn`t trust any pdf .Open in sandboxie or a VM if you use windows.