Identifying a SMT regulator (long)

[PlainBill]

I recently picked up a dead Memorex MLT1912 TV (19" Hi definition
with LCD display). It was represented as 'dead', and that's no lie!

Investigation revealed the standby LED was on, and no response to
either the remote or the control panel power button. This TV get
power from a 12V 'power brick'. This appears to be working properly.
Further regulation used N2596G switching regulators. Some had been
turned off via the control input, and determining the resaon was
getting tedious, so I switched to 'eye-tech' mode.

That revealed a suspicious burn mark on a connector on a daughter
board. Examining the adjacent IC through a magnifier I was able to
clearly see that the 'magic smoke' had escaped. So far, my attempts
to fully identify the IC have been fruitless. A quick summary of what
I have found:

SOT23-5 package marked with A17d (the d represents my best guess).
Pin 4 is shorted to pin 3. A tentative identification of the pinout
is:

1 Control? (CN3-4 via 4.7K R87)
2 Ground
3 Power out
4 Power in
5 Feedback? (Resistor network?)

There is a SMT ferrite bead in the trace to pin 4, and another after
the inductor connected to pin 3. It was suggested this is probably a
high frequency buck regulator, similar to an Anpec APW7101. Datasheet
at http://www.datasheetpro.com/2712_download_APW7101-1_datasheet.html

NOW the problem. Many manufacturers produce buck regulators in the
SOT23-5 package. Usually they will have several versions with a fixed
output voltage and one with a variable output voltage. All attempts
to identify one with a part number including A17? have failed; the
logical conclusion is this is a 'device marking', meaningful only if
the manufacturer is known. And I have been unable to find any
manufacturer using A17x as a device marking - but I'm still looking.

I tried to 'reverse engineer' this by tracing the output. This has
also been unsuccessful. I have identified several capacitors that are
filtering the output, but cannot find a connection to any IC on the
board - but I'm still looking. This is complicated by the fact that
this seems to be a multilayer circuit board.

Unfortunately, one of the ICs is a R8A66950BG. Since this is a ball
grid package, I can't probe the pins. I have been unable to locate a
data sheet, tech briefs indicate it requires 3.3, 1.25, and possibly
2.5 volts - the manufacturer is ambiguous about this.

I have been in contact with Memorex, they refered me to Fox
International. Fox International does not list a service manual for
this TV. They are 'researching', and 'will get back to me'.

While I'm not out of options, I'm hoping for new ideas on
troubleshooting this. Of course, someone correctly identifying the
part would be most welcome. Ideas on how to SAFELY determine the
proper voltage would also be helpful. Or an exhaustive database of
SMT markings vrs part numbers would be most appreciated.

Pictures of the circuit board and the IC are available at
http://www.badcaps.net/forum/showthread.php?t=6915

Thanks,

PlainBill

 

[David Nebenzahl]

I tried to 'reverse engineer' this by tracing the output. This has
also been unsuccessful. I have identified several capacitors that are
filtering the output, but cannot find a connection to any IC on the
board - but I'm still looking. This is complicated by the fact that
this seems to be a multilayer circuit board.

Unfortunately, one of the ICs is a R8A66950BG. Since this is a ball
grid package, I can't probe the pins. I have been unable to locate a
data sheet, tech briefs indicate it requires 3.3, 1.25, and possibly
2.5 volts - the manufacturer is ambiguous about this.

I have been in contact with Memorex, they refered me to Fox
International. Fox International does not list a service manual for
this TV. They are 'researching', and 'will get back to me'.

It's experiences like these that lead me to conclude that, more and
more, the entire set is to be considered one big FRU.

 

[Franc Zabkar]

A quick summary of what I have found:

SOT23-5 package marked with A17d (the d represents my best guess).
Pin 4 is shorted to pin 3. A tentative identification of the pinout
is:

1 Control? (CN3-4 via 4.7K R87)
2 Ground
3 Power out
4 Power in
5 Feedback? (Resistor network?)

There is a SMT ferrite bead in the trace to pin 4, and another after
the inductor connected to pin 3. It was suggested this is probably a
high frequency buck regulator, similar to an Anpec APW7101. Datasheet
at http://www.datasheetpro.com/2712_download_APW7101-1_datasheet.html

I tried to 'reverse engineer' this by tracing the output. This has
also been unsuccessful. I have identified several capacitors that are
filtering the output, but cannot find a connection to any IC on the
board - but I'm still looking.

Can you apply the voltage formula in the APW7101 datasheet using the
measured values of your feedback resistors?

Vout = 0.6 x (1 + Rf1 / Rf2)

You may arrive at a voltage which makes sense as a Vdd for one or more
of the chips. Other regulators may use a 1.25V internal reference
instead of 0.6V.

Unfortunately, one of the ICs is a R8A66950BG. Since this is a ball
grid package, I can't probe the pins. I have been unable to locate a
data sheet, tech briefs indicate it requires 3.3, 1.25, and possibly
2.5 volts - the manufacturer is ambiguous about this.

AIUI, 1.2V is required for the CPU core, 3.3V for I/O (analogue), and
2.5V for the DDR-SDRAM interface:

http://www.renesas.com/media/company_info/news_and_events/press_releases/2005/0926/20050926e.pdf

According to the technical brief, "the R8A66950BG requires only the
addition of a single 16-bit-width 256-Mbit DDR333 in order to
implement various kinds of processing supporting digital TV
broadcasting".

It seems to me that there should be a discrete SDRAM chip which will
allow you to confirm whether your SMPS is a 2.5V regulator. Otherwise
I'd measure any feedback resistors for the other two (?) regulators
and determine their output voltages that way.

- Franc Zabkar

 

[jdiaz5513]

Here's a TSOT23-5 with the marking A17: http://www.micrel.com/_PDF/mic7300.pdf

It's possible the "d" isn't even supposed to be a letter at all?

If this helps you, I'd appreciate it if you took a look at my head-
scratcher:
http://groups.google.com/group/sci....72/f46dfde4d7ae8aaa?lnk=raot#f46dfde4d7ae8aaa

-JD

 

[Baron]

PlainBill Inscribed thus:

I recently picked up a dead Memorex MLT1912 TV (19" Hi definition
with LCD display). It was represented as 'dead', and that's no lie!

Investigation revealed the standby LED was on, and no response to
either the remote or the control panel power button. This TV get
power from a 12V 'power brick'. This appears to be working properly.
Further regulation used N2596G switching regulators. Some had been
turned off via the control input, and determining the resaon was
getting tedious, so I switched to 'eye-tech' mode.

That revealed a suspicious burn mark on a connector on a daughter
board. Examining the adjacent IC through a magnifier I was able to
clearly see that the 'magic smoke' had escaped. So far, my attempts
to fully identify the IC have been fruitless. A quick summary of what
I have found:

SOT23-5 package marked with A17d (the d represents my best guess).
Pin 4 is shorted to pin 3. A tentative identification of the pinout
is:

1 Control? (CN3-4 via 4.7K R87)
2 Ground
3 Power out
4 Power in
5 Feedback? (Resistor network?)

There is a SMT ferrite bead in the trace to pin 4, and another after
the inductor connected to pin 3. It was suggested this is probably a
high frequency buck regulator, similar to an Anpec APW7101. Datasheet
at http://www.datasheetpro.com/2712_download_APW7101-1_datasheet.html

NOW the problem. Many manufacturers produce buck regulators in the
SOT23-5 package. Usually they will have several versions with a fixed
output voltage and one with a variable output voltage. All attempts
to identify one with a part number including A17? have failed; the
logical conclusion is this is a 'device marking', meaningful only if
the manufacturer is known. And I have been unable to find any
manufacturer using A17x as a device marking - but I'm still looking.

I tried to 'reverse engineer' this by tracing the output. This has
also been unsuccessful. I have identified several capacitors that are
filtering the output, but cannot find a connection to any IC on the
board - but I'm still looking. This is complicated by the fact that
this seems to be a multilayer circuit board.

Unfortunately, one of the ICs is a R8A66950BG. Since this is a ball
grid package, I can't probe the pins. I have been unable to locate a
data sheet, tech briefs indicate it requires 3.3, 1.25, and possibly
2.5 volts - the manufacturer is ambiguous about this.

I have been in contact with Memorex, they refered me to Fox
International. Fox International does not list a service manual for
this TV. They are 'researching', and 'will get back to me'.

While I'm not out of options, I'm hoping for new ideas on
troubleshooting this. Of course, someone correctly identifying the
part would be most welcome. Ideas on how to SAFELY determine the
proper voltage would also be helpful. Or an exhaustive database of
SMT markings vrs part numbers would be most appreciated.

Pictures of the circuit board and the IC are available at
http://www.badcaps.net/forum/showthread.php?t=6915

Thanks,

PlainBill

Not much point putting pictures on a site where you have to register to
see them !

 

[Franc Zabkar]

SOT23-5 package marked with A17d (the d represents my best guess).

Perhaps a clue as to the marking code, "A17d", may lie on page 3 of
this datasheet:
http://www.micro-bridge.com/data/Axelite/AX3503.pdf

The "7" could designate the year, 2007, and the "d" may indicate the
week, 30, in which case the "A1" would be the important part of the
code.

Is this consistent with the date codes on the other chips?

Pin 4 is shorted to pin 3. A tentative identification of the pinout
is:

1 Control? (CN3-4 via 4.7K R87)
2 Ground
3 Power out
4 Power in
5 Feedback? (Resistor network?)

If the input voltage is shorted to the output pin, then I'd be worried
about damage to the downstream parts.

Unfortunately, one of the ICs is a R8A66950BG.

FWIW, the following parts lists have the same R8A66950BG chip (p/n
0IMCR02243A). You can actually buy it for $8 !!!

LG Electronics/Zenith DU-27FB34C:
http://finder.mcmb2b.com/lg.php?model=100159
http://bestbuy.partsearch.com/Model/LG+Electronics+Incorporated/LG+Electronics/Zenith/DU27FB34C.aspx

LG Electronics/Zenith H20J54DT:
http://finder.mcmb2b.com/lg.php?model=100183
http://bestbuy.partsearch.com/Model/LG+Electronics+Incorporated/LG+Electronics/Zenith/H20J54DT.aspx

LG Electronics/Zenith H27H49S:
http://bestbuy.partsearch.com/Model/LG+Electronics+Incorporated/LG+Electronics/Zenith/H27H49S.aspx

Unfortunately none of the "power management" ICs look like yours. :-(

Maybe you could track down a compatible chassis by searching for TVs
that use the same major parts. Can you identify the panel?

Can you find another TV brand/model that uses the same remote codes as
yours? You could take your remote to a store and try it on the display
sets, or search the http://www.oneforall.com/ web site.

FWIW, the part number for your remote appears to be
"HS-Y3719-BLK-320".

If your TV has an FCC ID, then you could possibly obtain schematics
here, or at least determine who actually made it:
https://gullfoss2.fcc.gov/oetcf/eas/reports/GenericSearch.cfm

- Franc Zabkar

 

[PlainBill]

Can you apply the voltage formula in the APW7101 datasheet using the
measured values of your feedback resistors?

Vout = 0.6 x (1 + Rf1 / Rf2)

You may arrive at a voltage which makes sense as a Vdd for one or more
of the chips. Other regulators may use a 1.25V internal reference
instead of 0.6V.


AIUI, 1.2V is required for the CPU core, 3.3V for I/O (analogue), and
2.5V for the DDR-SDRAM interface:

http://www.renesas.com/media/company_info/news_and_events/press_releases/2005/0926/20050926e.pdf

According to the technical brief, "the R8A66950BG requires only the
addition of a single 16-bit-width 256-Mbit DDR333 in order to
implement various kinds of processing supporting digital TV
broadcasting".

It seems to me that there should be a discrete SDRAM chip which will
allow you to confirm whether your SMPS is a 2.5V regulator. Otherwise
I'd measure any feedback resistors for the other two (?) regulators
and determine their output voltages that way.

- Franc Zabkar

That's why I was asking here. You have given me two ideas which bear
investigating. Yes, there is a single Samsung RAM chip. That does
provide an avenue to pursue. Frankly, I had not considered the
possibility of determining if this was really a adjustable output
regulator. I was fixated on it being a fixed output device and never
made the leap.

Thank you for two good ideas.

PlainBill

 

[PlainBill]

Here's a TSOT23-5 with the marking A17: http://www.micrel.com/_PDF/mic7300.pdf

It's possible the "d" isn't even supposed to be a letter at all?

If this helps you, I'd appreciate it if you took a look at my head-
scratcher:
http://groups.google.com/group/sci....72/f46dfde4d7ae8aaa?lnk=raot#f46dfde4d7ae8aaa

-JD

There is a very good chance that the last character is not a 'd'. It
was partially destroyed when the chip shorted, and could be part of a
manufacturer's logo, part of a 4 or an 8, etc. Datasheet databases
have excellent search engines and can find matches for partial part
numbers. I was unable to find any part number containing 'A17' which
was available in a TSO23-5 package. This led to the conclusion that
it was not a partial part number, but rather a 'smd part marking'.

I am also trying to get access to a microscope. That might give me
more information.

I'd already looked at that thread. Frankly, you are already beyond
what I would consider trying. And I'm notorious for NEVER giving up
on a repair until I've irretriveably broken it - and sometimes not
even then.

PlainBill

 

[PlainBill]

PlainBill Inscribed thus:


Not much point putting pictures on a site where you have to register to
see them !

Agh!!! Sorry about that.

PlainBill

 

[PlainBill]

Perhaps a clue as to the marking code, "A17d", may lie on page 3 of
this datasheet:
http://www.micro-bridge.com/data/Axelite/AX3503.pdf

The "7" could designate the year, 2007, and the "d" may indicate the
week, 30, in which case the "A1" would be the important part of the
code.

Is this consistent with the date codes on the other chips?



If the input voltage is shorted to the output pin, then I'd be worried
about damage to the downstream parts.


FWIW, the following parts lists have the same R8A66950BG chip (p/n
0IMCR02243A). You can actually buy it for $8 !!!

LG Electronics/Zenith DU-27FB34C:
http://finder.mcmb2b.com/lg.php?model=100159
http://bestbuy.partsearch.com/Model/LG+Electronics+Incorporated/LG+Electronics/Zenith/DU27FB34C.aspx

LG Electronics/Zenith H20J54DT:
http://finder.mcmb2b.com/lg.php?model=100183
http://bestbuy.partsearch.com/Model/LG+Electronics+Incorporated/LG+Electronics/Zenith/H20J54DT.aspx

LG Electronics/Zenith H27H49S:
http://bestbuy.partsearch.com/Model/LG+Electronics+Incorporated/LG+Electronics/Zenith/H27H49S.aspx

Unfortunately none of the "power management" ICs look like yours. :-(

Maybe you could track down a compatible chassis by searching for TVs
that use the same major parts. Can you identify the panel?

Can you find another TV brand/model that uses the same remote codes as
yours? You could take your remote to a store and try it on the display
sets, or search the http://www.oneforall.com/ web site.

FWIW, the part number for your remote appears to be
"HS-Y3719-BLK-320".

If your TV has an FCC ID, then you could possibly obtain schematics
here, or at least determine who actually made it:
https://gullfoss2.fcc.gov/oetcf/eas/reports/GenericSearch.cfm

- Franc Zabkar

Franc,

Thanks again for so much help. The problem with identifying a part by
it's top markings is that unless you can determine the manufacturer it
can be pretty tedious. (I know I'm preaching to the choir here). Note
that AXElite's AX3503 has a top marking that starts with 'G3'. Yes,
it is possible that 'A1' is the significant part of the top marking
and 7d represents the 30th week of 2007.

Much of my recent effort has been spent perusing the data sheets of
various voltage regulatrors available in the SOT23-5 package, looking
for top markings beginning with the latter A. So far this has not
yielded any results.

Yes, I am very much aware that the shorted power regulator could have
blown the chip it was powering. I'm willing to take a chance on
buying a $2.00 part (or two or three of them) just to see if that will
solve the problem. I'm a little less certain about trying to replace
the R8A66950BG. SMD fuses and resistors don't present a challange.
I've replaced a few 16 and 24 pin chips. Ball grid? Well, maybe.

The FCC ID is another avenue I had never considered. Unfortunately, I
can't see an FCC ID on it!!!

Thanks for the help.

PlainBill

 

[Franc Zabkar]

You have given me two ideas which bear
investigating. Yes, there is a single Samsung RAM chip. That does
provide an avenue to pursue. Frankly, I had not considered the
possibility of determining if this was really a adjustable output
regulator. I was fixated on it being a fixed output device and never
made the leap.

One other thing I've done when tracking down manufacturers is to look
for similar user manuals. Sometimes an OEM will take the easy way out
and regurgitate the original Chinese docs. A search for distinctive
phrases or typos can sometimes yield results. Adding "filetypedf" to
Google's search box can narrow down your results.

I've also tracked down the manufacturer of my supermarket DVD recorder
via markings and labels on the PCB artwork, and I've even managed to
find a service manual from yet another source.

Yet another trick may be to hook up the TV's VGA input to a Windows
PC. Windows PnP will then interrogate the monitor/TV. Search the
registry for EDID information, or use a utility to display this
information directly.

Here's an EDID utility for Windows 95/98/Me:
http://www.entechtaiwan.com/files/ddctest.exe

Another more sophisticated program is PowerStrip:
http://www.entechtaiwan.com/util/ps.shtm

The EDID bytes may include text strings identifying the manufacturer
or the OEM or the model.

- Franc Zabkar

 

[Franc Zabkar]

Yet another trick may be to hook up the TV's VGA input to a Windows
PC. Windows PnP will then interrogate the monitor/TV. Search the
registry for EDID information, or use a utility to display this
information directly.

The monitor does not need to be powered up for this to work. This is
because the PC's graphics card provides the +5V power for the
monitor's EDID EEPROM.

- Franc Zabkar

 

[PlainBill]

Can you apply the voltage formula in the APW7101 datasheet using the
measured values of your feedback resistors?

Vout = 0.6 x (1 + Rf1 / Rf2)

You may arrive at a voltage which makes sense as a Vdd for one or more
of the chips. Other regulators may use a 1.25V internal reference
instead of 0.6V.

I finally got back to this briefly yesterday and found that rf1 apears
to equal rf2 (each 300K) before being interrupted. I'm going to hone
my technique before drawing conclusions; there seems to be a
transistor (ID H1A) in the circuit.

I haven't had a chance to trace the output further.

Fox International got back to me; no service manual is available.
When time permits I'll be checking if U11 or the entire daughter board
is available.

PlainBill

 

[Franc Zabkar]

On Wed, 15 Apr 2009 11:50:09 +1000, Franc Zabkar


I finally got back to this briefly yesterday and found that rf1 apears
to equal rf2 (each 300K) before being interrupted. I'm going to hone
my technique before drawing conclusions; there seems to be a
transistor (ID H1A) in the circuit.

1.2V (= 0.6 x 2) does make sense as a CPU core voltage, and 2.5V (=
1.25 x 2) could supply the SDRAM.

- Franc Zabkar

 

[PlainBill]

1.2V (= 0.6 x 2) does make sense as a CPU core voltage, and 2.5V (=
1.25 x 2) could supply the SDRAM.

- Franc Zabkar

Yeah, I caught the irony of that. I'm hoping to spring some more time
free to probe the source of the 2.5V for the DDRam chip. Working on
both sides of a multilayer board is a bitch. It would be easier if I
could figure out which of a couple of dozen boxes holds my 30 ga
insulated wire.

PlainBill

 

[PlainBill]

1.2V (= 0.6 x 2) does make sense as a CPU core voltage, and 2.5V (=
1.25 x 2) could supply the SDRAM.

- Franc Zabkar

I decided to give Fox Int'l another try and received both good and bad
news. Contrary to my earlier information, the service manual is
available. Unfortunately, it is not in stock. They list three parts
for this set - the main board, the tuner board, and the 'crystal'
(Liquid crystal display?) . The tuner board (TL ATSC LCD board?) is
listed, and costs only $22. But 'it's out of stock for 2 to 12 weeks'

PlainBill

 

[Franc Zabkar]

I decided to give Fox Int'l another try and received both good and bad
news. Contrary to my earlier information, the service manual is
available. Unfortunately, it is not in stock. They list three parts
for this set - the main board, the tuner board, and the 'crystal'
(Liquid crystal display?) . The tuner board (TL ATSC LCD board?) is
listed, and costs only $22. But 'it's out of stock for 2 to 12 weeks'

PlainBill

Hmmm. Fox are the service agents for Memorex, yet neither they nor
Memorex have a service manual for a TV that was sold on the US market
possibly within the last few years. A service manual would most likely
be provided in digital form as a PDF file, so why the delay? Surely
it's just a matter of emailing you a copy, or burning a CD-R?

If you intend to sell 10,000 TV sets, say, to a particular market,
then surely you will prepare yourself with a service manual and other
support materials. What goes on inside these companies? Are their
directors really just a bunch of ignorant, incompetent sales droids?

- Franc Zabkar

 

[PlainBill]

Hmmm. Fox are the service agents for Memorex, yet neither they nor
Memorex have a service manual for a TV that was sold on the US market
possibly within the last few years. A service manual would most likely
be provided in digital form as a PDF file, so why the delay? Surely
it's just a matter of emailing you a copy, or burning a CD-R?

If you intend to sell 10,000 TV sets, say, to a particular market,
then surely you will prepare yourself with a service manual and other
support materials. What goes on inside these companies? Are their
directors really just a bunch of ignorant, incompetent sales droids?

- Franc Zabkar

I've done business with Fox for decades, and have always had
satisfactory service. Still, I prefer dealing locally. at least you
can see the item before buying it.

There does seem to be multiple levels of incompetence here. Not
having service information and parts available, not identifying the
availability to a customer, not having the parts in stock.

That brings something else to mind. At one time a manufacturer had to
keep parts available for a certain length of time after production
ceased - 5 years IIRC. Is that no longer the case?

PlainBill

 

[Franc Zabkar]

At one time a manufacturer had to
keep parts available for a certain length of time after production
ceased - 5 years IIRC. Is that no longer the case?

This requirement is easily circumvented by offering spare parts as
modules, and then pricing them so that repairs become uneconomical.

- Franc Zabkar

 

[PlainBill]

This requirement is easily circumvented by offering spare parts as
modules, and then pricing them so that repairs become uneconomical.

- Franc Zabkar

Well, I'm tentatively reporting success. Focusing on the
K4D551638H-LC50 RAM, I was able to determine it got power from U8 a
UTC1117. THAT turned out to be a 3.3V regulator. WTF??? The
K4D551638H-LC50 requires 2.6 volts! The I discovered that YOUWANG
ELECTRONICS makes variants with other output voltages. There are
three UTC1117s on the board, the one powering the K4D551638H-LC50 has
B25 on it (this is not the date code per the data sheet). The others
have B33 (near the Flash memory) and B18 (near the LGDT3303). I
haven't been able to find a data sheet for the LGDT3303, but 3.3 volts
appears to be right for the flash memory. And all of the UTC1117's
were from YOUWANG ELECTRONICS.

The next problem was locating a replacement for the dead chip. My
perfered parts supplier is DigiKey. While they list a large number of
regulators in the SOT23-5 package, they only stock those from National
Semiconductor. And none those have the correct pinout. Bummer!!!

Fortunately, Mouser Electronics came to my rescue. They stock only
one adjustable bucking regulator in the SOT23-5 package. It's the
Exar SP6669AEK-L, and it has the correct pinout. The best news they
will sell in quantities of 1.

Now to make up an order (I need caps, transistors, and a fuse for a
dead LCD monitor). Then a weeks wait for the pars to come in.
Finally, solder in the regulator and give it the smoke test.

PlainBill