Why might these ICs be failing?

[tom66]

On the Badcaps.net TV repair forum we've been noticing an interesting trend.

Since about 2012 large numbers of chips on T-con boards (the part which controls the LCD panel in an LCD TV) are failing. In every case the chips are AS15 or AS19 from the same manufacturer, full part number EC5575, datasheet here.) Replacing the IC solves the problem in virtually every case (symptoms being distorted colours, solarised picture, poor contrast or overbright image.)

The IC is essentially a 15 channel voltage buffer (internally implemented as a unity gain op-amp.) When working, each output follows a nice smooth curve. When failed, typically 3 or more points of the curve will become distorted; but usually at least a few of the points still work.  Of note, the input sources are high impedance (from a resistive ladder), and they typically get pulled the same way, so the REF input matches the output. Removing the IC makes the voltages return to normal.

When the IC is failed it runs about 20C hotter than normal. Normal operating temperature is about 35C. 

The failure is typically described as sudden; one moment the TV is working, and the next, the image is bad. Sometimes, as the IC warms up, the fault will get worse. The ICs have been found to last between 3 and 6 years before failure, although hours of usage are unknown because they don't operate with the TV off. They've been found on boards from Chungwha PT, AU Optronics, Samsung, and Chi-Mei Optoelectronics... so it's not just one manufacturer screwing up the design.

Any semiconductor guys here might know what's going on? My personal guess is some kind of bias circuit for the input is failing, but it's failing on random channels, and not all at the same time, which is odd. Not sure why the failure would be so random, rather than for example only affecting the middle most channels (with the highest power dissipation.)

[MatCat]

Poorly made IC's, 3-6 years?  Sounds about right for a consumer change in TV set by then doesn't it...

[bench_knob]

Quote from: tom66 on Today at 05:34:30 AM>

On the Badcaps.net TV repair forum we've been noticing an interesting trend.


Since about 2012 large numbers of chips on T-con boards (the part which controls the LCD panel in an LCD TV) are failing. In every case the chips are AS15 or AS19 from the same manufacturer, full part number EC5575, datasheet here.) Replacing the IC solves the problem in virtually every case (symptoms being distorted colours, solarised picture, poor contrast or overbright image.)

The IC (EC5575) is essentially a 15 channel voltage buffer (internally implemented as a unity gain op-amp.) When working, each output follows a nice smooth curve. When failed, typically 3 or more points of the curve will become distorted; but usually at least a few of the points still work.  Of note, the input sources are high impedance (from a resistive ladder), and they typically get pulled the same way, so the REF input matches the output. Removing the IC makes the voltages return to normal.

When the IC is failed it runs about 20C hotter than normal. Normal operating temperature is about 35C. 

The failure is typically described as sudden; one moment the TV is working, and the next, the image is bad. Sometimes, as the IC warms up, the fault will get worse. The ICs have been found to last between 3 and 6 years before failure, although hours of usage are unknown because they don't operate with the TV off. They've been found on boards from Chungwha PT, AU Optronics, Samsung, and Chi-Mei Optoelectronics... so it's not just one manufacturer screwing up the design.

Any semiconductor guys here might know what's going on? My personal guess is some kind of bias circuit for the input is failing, but it's failing on random channels, and not all at the same time, which is odd. Not sure why the failure would be so random, rather than for example only affecting the middle most channels (with the highest power dissipation.)

Without a lot more in depth info about how the device is implemented in the circuit (ie, is it being sneak-stressed in some manner?) as well as specific manufacturing details about the IC itself, one may only speculate (via wild-a.ssed guessing) about possible causes. However, one can easily introspect that the silicon itself is being overly stressed simply by its (likely) being an SMD component, that requires high-temperature type 'baking' to afford mechanical/electrical circuit interface junction connection..soldering to PCB. A friend, Dr. Sun Yuen Wong, a multiple Ph.D. scientist; we had discussions regarding SMD component soldering techniques, all-in-all having the general conclusion, the new soldering strategies heat damage **ALL** SMD components shortening their MTBF.

The modern LCD TVs are rated at around five to seven years life-span, as compared to the through-hole component TVs (CRT TVs) which exhibited typical life-spans of 15 ~ 20 years.

So if ya got some silicon that's not quite up to what its supposed to be, such as might be the case in the IC you cite, ya might well see premature failures ahead of the rest of the components. 

bench knob

[SeanB]

If it is a batch problem likely it is either a contaminated lot of the chips ( failing machine leaked something that got on the chips and contaminated them) or a batch of poorly cured or wrongly mixed encapsulant that had impurities in it.  Or the IC was just not designed correctly, and has a high failure rate.

[tom66]

I just fixed this Sony with this problem. Before & after pic. Very obvious failure. However in some cases, the image just looks "off", almost as if it has too much or too little contrast, or the black level is too high. I think the reason colours are affected (rather than just greyscale) is because the colours have gains/offsets set on the T-con processor (to compensate for the response of the pixel, wavelengths of the colour films, bulb wavelength, etc.) so when these get distorted the picture colours go out of sync causing colour distortion too.

[SArepairman]

Well I would measure the temperature of the running chips and the failing chips (which you did), the output current of the chips, monitor for voltage/current spikes, how clean the supply rails are. I would also specifically look into system start and shutdown conditions, where there is most likely the most stress.. maybe something is sinking alot of current on start up.

Re solder in your bad chip and measure some parameters then solder in the good chip and measure some parameters. It would maybe give you a better idea of what is going on. It could be as simple as a bias resistor drift causing the chips to drive something too hard? Or maybe some decoupling capacitor ESR is too low and its just pushing some chip too hard on startup. Or maybe an inductive kick back from some kind of inductive element. These kinds of test could help rule out improper implementation vs semiconductor failure.

But, since the problem is widespread to many different systems... i dont know.

Perhaps there is an internal data-sheet which suggest some component that is to be used with the chip (like a specific capacitor, transistor, etc) that is causing a problem. It seems unlikely that the design of the chip is flawed, I mean its a multimillion dollar operation to produce it.


Out of curiosity, are there known cases of design flawed semiconductors? Another user in a nother thread mentioned to me that certain data-sheets have incorrect parameters due to a flawed measurement methodology regarded as a "industry standard"... so the datasheet that comes with the chip might specify parameters incorrectly which lead to an over drive of the chip.

or maybe someone just signed some papers because they wanted to go home and see their wife after pulling a 70 hour work week during a "crunch time". (ah the mentality of bosses eh)

[edavid]

Does the voltage divider always get loaded down?  If so, that means the input MOSFET gates or protection diodes are getting damaged.  It should be easy to check if there is excessive voltage on the input pins (perhaps during power up/down).  If so, the board design is at fault, if not, it's the chip design.

[amyk]

This is the buffer for the ladder DAC that drives the pixels, so it's not surprising that a faulty channel causes that type of colour artifact to occur. As you mention that the failures are only occurring with one manufacturer, I think the problem is related to bad design; there are other pinout-compatible 14+1-channel buffers which presumably have been used on other TVs, but no reports of failure with them?

http://www.st.com/web/en/resource/technical/document/datasheet/CD00062587.pdf
http://www.ic72.com/pdf_file/a/65081.pdf

What supply voltage are they being run at?

[tom66]

They are run at 15.5V below the maximum rating. Same voltage that the panel uses for source drive.

I have been diagnosing this problem for a while on badcaps but this is the first one I have done myself; recently, it appears there's been a surge in failures. We've seen 5 in one week all the same chip.

I just figured out the service menu for the Sony, the hour count was 8,950 hours. That counts the time the TV is powered on and displaying a picture... so not long at all.

Only seen reports of the AS15 & AS19 ICs failing. AS15 fails a lot more often.

The odd thing is (I typoe'd this in my first post) it's seemingly always multiple outputs that go bad at once. And yes it forces the resistive divider to track the bad outputs.

In my case the VGAMA12~21 were bad; there are 21 gamma channels generated using 14 op-amp channels, I think a couple are averaged together in places using resistors. The upper half of the channels of the IC were failed. So in terms of die space, it was the same area of the chip.

Just a guess: Maybe some kind of transient event on turn on/off is causing destruction of one of the inputs, subsequently causing failure of inputs near it due to thermal damage? Or some kind of  shared bias for those channels went bad. It sure is odd.

[Kjelt]

In every case the chips are AS15 or AS19 from the same manufacturer

You can check if the same chips are used in another manufacturers tv set and if they encounter the same problems.
If there are other tv manufacturers using the same chip and have no problems: suspect the design, be it on the supplyvoltage (glitches, peaks etc.) or inputs/outputs.
If there are no other tv manufacturers using the same chip: suspect the chip (manufacturer).
Anyway it is always difficult to find these causes, but as Sherlock said: when you have eliminated the impossible, whatever remains, however improbable, must be the truth.

[John Coloccia]

This wreaks of an issue with poor input protection and/or static.  Have you ever seen what a chip looks like after it take a static hit or something blows up because of bad input protection?  You may as well smash that area of the chip with a hammer...it takes everything in the vicinity out.

[edavid]

Can you check if there is any voltage applied to the IC inputs or outputs when the TV is in standby?

[tom66]

In every case the chips are AS15 or AS19 from the same manufacturer

You can check if the same chips are used in another manufacturers tv set and if they encounter the same problems.
If there are other tv manufacturers using the same chip and have no problems: suspect the design, be it on the supplyvoltage (glitches, peaks etc.) or inputs/outputs.
If there are no other tv manufacturers using the same chip: suspect the chip (manufacturer).
Anyway it is always difficult to find these causes, but as Sherlock said: when you have eliminated the impossible, whatever remains, however improbable, must be the truth.

Personally on the forum I've seen more than 20 different boards from four different manufacturers (of LCD panels) with the problem.
Chance of such a multitude of screwups is low...

Note that the LCD panel is what is failing. Sony buy the AU Optronics panel for their TV. In the 40" models they use Samsung FHD panels which also have this problem.

Can you check if there is any voltage applied to the IC inputs or outputs when the TV is in standby?

After a couple of seconds after entering standby, less than 1V is applied on the Vdd or REF pins on this T-con board.

This wreaks of an issue with poor input protection and/or static.  Have you ever seen what a chip looks like after it take a static hit or something blows up because of bad input protection?  You may as well smash that area of the chip with a hammer...it takes everything in the vicinity out.

That's part of my thinking but would ESD be such a concern in an IC which has no exposure to users?  It could of course be factory damage - but why would it last such a long time?

[edavid]

Can you check if there is any voltage applied to the IC inputs or outputs when the TV is in standby?

After a couple of seconds after entering standby, less than 1V is applied on the Vdd or REF pins on this T-con board.

I think you need a better measurement than that.  If there is current flowing through the ESD diodes on the inputs or outputs in standby (which only requires 0.5V), that could certainly lead to long term damage.

[tom66]

What method would you suggest?

[Stonent]

A guy at work gave me his old 32" 720p Westinghouse and after about 20 minutes the screen would start streaking with red lines and then blocks. I replaced the T con board and its been fine. The box indicated the board fit about a dozen different brands of TV. I'll take a pic when I get home.