EEVblog #630 - Soniq LCD TV Troubleshooting Repair

[EEVblog]

Dave troubleshoots a Soniq L32V12A 32" HD LCD TV

[Legit-Design]

Those spring loaded thin probe tips make it look so easy.

Just saw Dave jump back while measuring something... After that he went under the bench to take powercord out?
Looks like there was near electrocution situation. Good thing nothing happened! 

Did you forget to discharge the mains filter cap? Thats why you added safety note about one hand behind back? 
I think it was before you were measuring the ESR on the live stream.

[bktemp]

It looks like the main processor/display controller is not working correctly.
You started at the mains input, I would start at the other end: Measuring the signals going to the lcd panel. If there is something wrong, you know it is not the display.
The signals to the lcd are typically LVDS: 1 differential pixel clock and 4 differential serial data lines running at 7x pixelclock rate.
For a 1366x768 panel the pixel clock should be around 75MHz, depending on the framerate. It may be transmitting odd and even pixels in parallel then the clockrate is halfed and the number of data lines is doubled. Maybe you can use and show some of the advanced trigger features of your scopes to detect gaps in the clock signal.
If this clock is not continuously present while the tv is on and displaying an image, there is something wrong with the main processor. Then you should look for other signals failing/changing at the same time.
You could turn this repair video into a fundamental video of how to probe highspeed logic signals using a state of the art scope trying to find a fault in a digital logic system.

[PeterG]

Why not scrap the Soniq tv and buy your mum a nice LG or similar with a better screen.....

Regards

[nitro2k01]

Maybe it's just the different emissivity but I think the CPU heatsink looked suspiciously cold compared to the surrounding PCB area. Maybe you should do some contact temperature probing, as well seeing if adding some airflow makes a difference. I've famously been wrong with my hypotheses before, but maybe some dickhead designer used some form of glue which, instead of conducting heat from the CPU/main chip to the heatsink, isolates it.

[SeanB]

Just the bright aluminium doing tricks with emissivity. The surrounding hot board says the BGA is definitely getting toasty warm, so most likely the balls are cracked on the chip or on the carrier board. That will explain the symptoms, and the sudden stopping of it responding initially when Dave touched the back side and stressed the board.

Might be worth doing a reflow of the BGA from the back of the board, but this one is likely a terminal case.

Sell the power supply on eBay and cut your losses. Take the panel apart and see how good the backlights are still and give them with the board, packed in a plastic sleeve and wrapped in bubble wrap then inserted into a cardboard mailing tube ( or some 50mm PVC drain pipe)  longer than the backlight tubes and capped with tape and more bubble wrap.

I got free a nice LG 22 in flatron display, power supply is fine, dead microcontroller on the small input PCB. Going to scrap it for metal and the backlights, just in case I need them on my other monitor. No screws to open the case, more just prize the 2 parts apart with a screwdriver.

[tom66]

I think you should do tests of those regulators when it isn't working, as well as when it is. Also, that 1.89V is very high for a 1.8V regulator with a 2% tolerance typical. A high voltage reading can indicate an oscillating regulator due to bad regulator or output capacitor. (It can also indicate something like a low 2.5V regulator, so it's important to check part numbers. The voltage is usually indicated by the last two characters -33 or -18 for example.)

I had a cheap 37" LCD TV which had a horrible whine on the audio and it would sometimes only boot with graphical noise all over the panel. It turned out to be a bad capacitor on the output of a buck converter. Testing the output voltage showed 1.79V, for a 1.8V rail that is fine, but it had about 300mVp-p of ripple on it. Once the TV warmed up the ripple disappeared.

Also I would try using freeze spray or hot air to see if you can invoke a temperature dependent fault.

The 9G Pio Kuro on my wall right now had a bad regulator on the analog A/V board, bargain of the century to get it with that fault.

[peter.mitchell]

I think you should do tests of those regulators when it isn't working, as well as when it is. Also, that 1.89V is very high for a 1.8V regulator with a 2% tolerance typical. A high voltage reading can indicate an oscillating regulator due to bad regulator or output capacitor. (It can also indicate something like a low 2.5V regulator, so it's important to check part numbers. The voltage is usually indicated by the last two characters -33 or -18 for example.)

I had a cheap 37" LCD TV which had a horrible whine on the audio and it would sometimes only boot with graphical noise all over the panel. It turned out to be a bad capacitor on the output of a buck converter. Testing the output voltage showed 1.79V, for a 1.8V rail that is fine, but it had about 300mVp-p of ripple on it. Once the TV warmed up the ripple disappeared.

Also I would try using freeze spray or hot air to see if you can invoke a temperature dependent fault.

The 9G Pio Kuro on my wall right now had a bad regulator on the analog A/V board, bargain of the century to get it with that fault.

I was thinking this too Dave, you shoulda scope'd the rails.

[synapsis]

Last weekend a friend at work gave me a 50" plasma TV that had the intermittent issues then went dead. He knew about bulging capacitors and took a look around, but didn't find any.

I went through it and found the 5V standby rail was dead. Turned out a 1000uF cap had bulged *underneath*, the ESR went up to 16 ohms (LCR'd it after I took it out) and blew a Zener diode.

I found it with my Flir, but it would've been nice to have a meter like Dave's for that!

BTW, Samsung A/V equipment is nice in that they label the pinouts of all the connectors/test points on the silkscreen. This TV was one big power supply.

[Tothwolf]

I would definitely suspect faulty joints on the processor due to how it glitched out when Dave touched the side of the case. Maybe tap at it with a plastic spudger and warm it up a little with a hair dryer to see how it reacts? Just the difference in stress on the board from removing and reinstalling the mounting screws could have been enough to cause it to finally fail completely.

[open loop]

+1 for poor soldering/cracked BGA balls, liquid flux and heat the proc for 5  mins with a hot air gun, providing that heat sink comes off easy. Unless you know someone who has an IR rework setup.
We use at work something like:

http://www.pdr-rework.com/e6evolution

Much better than the cheap stuff you see on eBay.

[nathanpc]

What's the impedance of the LoZ mode? Looking at the photos from Dave's Fluke 117 teardown, the only big resistor that looks like would be used in the LoZ mode is this 1k 3W(?) resistor on the side:



And you would be pushing it pretty hard while discharging a 400V capacitor.

[Bored@Work]

What's the impedance of the LoZ mode?

3 kOhm

And you would be pushing it pretty hard while discharging a 400V capacitor.

The meter is specified for 600 V.

[tom66]

LoZ is implemented using a PTC,  which will increase in resistance as power dissipation increases. I imagine Fluke have tested it will withstand 1kV (or 600V) essentially continuously.

[Refrigerator]

Maybe sticking it in an oven would fix those problems, you know, the same way people fix GPUs and Xbox360 RRODs.

[nathanpc]

Thanks very much for clarifying.

[SeanB]

Low-Z mode has a voltage limitation if the voltage is above 40V, to a maximum of 30 seconds every 2 minutes or so, otherwise the PTC will overheat and burn out.

I have a cheap and rather elderly Stienel voltage check which has a PTC to limit current to the internal LED's and it has this warning for voltages over 40V, while below 40V the PTC will handle power continuously. nice in that it will handle AC or DC from around 5V to 400VAC in a single range, with only 3 neon's to show 110, 220 or 380 VAC or VDC with polarity given by the 2 red led's in the probe case.

[Bored@Work]

Low-Z mode has a voltage limitation

Not that particular Fluke.

[tom66]

Maybe sticking it in an oven would fix those problems, you know, the same way people fix GPUs and Xbox360 RRODs.

That rarely fixes any BGA soldering issue - the fault will return later. However, it is worth a shot, if only to prove it is the cause. After the more traditional tests have failed.

[Owen]

After watching Dave's Video and his differential probes, their comes a question in my mind, that I've always wanted to ask: When do I have to use those second ground bnc\plugs coming right from the BNC input of the probes? Some probes seem to have them some not, i always wondered why? In which situation do they become usefull? Thank you .

[Rasz]

Wrong video title, shouldnt be called Repair, more like 'Troubleshooting teardown with 20 minutes of talking about repairing'

Dont talk about measuring, measure. Put scope behind LDOs on the digital board, power up, look for a change when picture glitches/dies. Even better would be just changing all electrolytic caps on digital board blindly, power it up from separate PSU and check if it still glitches. Swapping all the caps takes less time than talking about swapping them

There is no need of reballing in case its solder joints under BGA, simply heating up main chip until it floats works in 90% of cases.

Dave wouldnt last 3 posts on elektroda.pl/rtvforum (forum for people servicing electronics professionally)

For people interested in real _repair_ videos I highly recommend this channel
https://www.youtube.com/user/1servicecore
Yes, its  in Ukraininan, but important stuff is rather self explanatory

Maybe sticking it in an oven would fix those problems, you know, the same way people fix GPUs and Xbox360 RRODs.

That rarely fixes any BGA soldering issue - the fault will return later. However, it is worth a shot, if only to prove it is the cause. After the more traditional tests have failed.

.

No. It rarely fixes Nvidia GPU BGA problems, because Nvidia had problems between BGA package and silicon, and not between package and pcb. Chip itself is faulty.


In case of TVs its usually shitty noPB solder thats the problem, no the chip.

[Sionyn]

thermal degradation of solder joints or Pad cratering that or a failing of the chip with small heatsink

[Feuerbard]

its chinese Mstar chip based TV , this soc used in many cheap entry level  brand tv like toshiba . lg  , samsung  e  t ?

if one dram in bga installed may be is Mstar Saturn 7 like in this SM

http://archive.espec.ws/files/LG%2026LE3300_26LE3308%20&%2026LE330N%20LED%20LCD%20TV%20SM.pdf

this chinese crappy tv have many problems in firmware , bad or corruption data in spi fplash and probably bad soldering on soc or dram

[AlphZeta]

Depending on the age of these LCD panels, sometimes the bond of the flat flex cable to the panel itself (i.e. the ones mated onto the glass) becomes unbounded which manifest itself as unstable images (usually flickering, showing weird color, etc.) and the image can go dark completely.

To test whether it is indeed flat flex to the LCD panel, you can apply pressure near the connection on the panel while the monitor is powered on and you should see picture re-appearing or flickering.

[Zad]

One technique that Dave needs to learn - tapping. Good old-fashioned methodical hitting stuff with a (insulated) stick.

Maybe it is because we have had half a generation of systems that run relatively cool, combined with the relative robustness of SMD that we don't get the same dry joint problems that we used to. With intermittent problems like this, tapping the components, PCB and interconnects with a nice soft insulated screwdriver handle is the primary fault detector / inducer.

Dave has a temperature controlled reflow oven, it would be worth stripping off the heatsinks and giving it a run through.