Samsung SyncMaster 940T blows inverter fuse

[bonney]

Re: Samsung SyncMaster 940T blows inverter fuse

If you have working one swap screens over

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

I wish we could find a full fledged datasheet for the BIT3195. Or even a source for the IC itself. I could swear those drive waveforms are wrong, but I DON'T have enough experience to be sure. I'll try to do a little digging for more information.

I did some looking around at work, and I found a functional 940T. It has leaky 820uF caps, but it works. When I get time, I'll have a look at the inverter IC's outputs to see what they do. Also Newark shipped my backordered 820uF 25v FK caps today, so hopefully I'll have them sometime next week.

Could the FETs could be damaged/leaky due to the bad caps from before? Worth a couple bucks to replace a set and see? If one hangs on (shutoff lag) while the other is coming on, wouldn't that cause this problem? As they heat, the problem gets worse and they are distorting the drive into the transformers. By time you feel 80C on the casing of them, it's way hotter than that on the substrate.

I'd spend a couple of bucks to try new ones if I could find a reasonable replacement part. Do you have any suggestions in that regard?

Since I have a working unit now, hopefully I won't have to replace the entire board. I'll watch that auction though, in case we aren't able to figure out what's wrong with the ones I have.

[Quasar]

Re: Samsung SyncMaster 940T blows inverter fuse

Could the FETs could be damaged/leaky due to the bad caps from before? Worth a couple bucks to replace a set and see? If one hangs on (shutoff lag) while the other is coming on, wouldn't that cause this problem? As they heat, the problem gets worse and they are distorting the drive into the transformers. By time you feel 80C on the casing of them, it's way hotter than that on the substrate.

Q

EDIT - Or spend $20 and use it to figure yours 4 out:
http://cgi.ebay.com/ws/eBayISAPI.dll...m=120626478733

[PlainBill]

Re: Samsung SyncMaster 940T blows inverter fuse

I wish we could find a full fledged datasheet for the BIT3195. Or even a source for the IC itself. I could swear those drive waveforms are wrong, but I DON'T have enough experience to be sure. I'll try to do a little digging for more information.

PlainBill

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

AH!!! Just had a thought. What is the frequency?

New board, different caps, same behavior. I used Panasonic FK 2200uF 25v 12.5x30 instead of the 3 CapXon GL 820uF 25v 10x20 caps Samsung used, and a Panasonic FK 2200uF 10v 12.5x20 instead of 2 CapXon KM 1000uF 10v 10x16. This one works for a little bit longer before going out of control than the last board did.

Scope pictures attached. I measured the drains of the FETs at one of the transformers, the output of pins 8 and 9, and the inverter side of the fuse. The ripple was different on both sides of the fuse, so I took my multimeter out and used a clip wire to bridge the fuse for this test. I'm not sure that the scope was able to get a good reading there though.

Short version is that the transformer gets a sine-ish (looks sawtooth-ish) wave at about 96KHz. The FETs are driven by square waves at 48KHz from controller pin 8, and 96KHz from pin 9. There are a couple of spikes, possibly from where the transistors switch on and off. The scope thinks that ripple on the 13v input is 177KHz, though it was having trouble making up its mind.

Just because I was curious, I used a compressed gas duster to cool the FETs, and this did stop the current increase at around 2.06A. Looks like the problems may be caused by the FETs getting hot (now to figure out why...).

Here's a brief description on the way that part of the circuit works.

Thanks, that was really helpful! I think I understand what's going on a lot better now.

[PlainBill]

Re: Samsung SyncMaster 940T blows inverter fuse

Here's a brief description on the way that part of the circuit works. You have a total of four FETs in two packages. Each package contains one P-channel and one N-Channel FET. The two P-channels are in parallel, the two N-channels are in parallel. The drains of all the FETs are tied to one end of the primary winding of the transformers. The source of the P-Channel FETs are tied to +14V, The source of the N-Channel FETs are tied to ground. The FETs are in the off state when the gate is at the same voltage as the source.

At rest, the gates of the P-Channel FETs are at 14V, the gates of the N-Channel FETs are at 0V. The primary of the transformers are floating somewhere between 14V and 0V. When the CCFLs are turned on the inverter controller will pull the gates of the P-Channel FETs down, they will pull the drains to 14V. Current will flow through the primary and the voltage at the other end of the primary will slowly rise due to the two caps (one will be charged, the other will be discharged). A few microseconds later the inverter controller turns off the P-Channel FETs. And a few microseconds later it turns on the N-Channel FETs, and again turns them off after a few microseconds. While they are on current flows in the opposite direction through the primary of the transformer.

The caps on the primary are there for two reasons. They help to form the output into a sinusoidal waveform; and they prevent any flow of DC current through the primary.

The inverter controller controls the timing of the gates so N-Channel and P-Channel FETs are not on at the same time. HOWEVER, if for some reason they are on at the same time, they will overheat and blow the fuse.

PlainBill

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

Well I may have a brain fart here, but I was wondering if this test can be applied to J305/J310 as well - that is, have both J305 and J310 replaced by an ammeter. If the current in the inverter increases, shouldn't the current through J305/J310 increase as well (or perhaps decrease)? If yes, then maybe something is wrong with the FETs, the driving circuit, or the transformer(s). If no, then we need to find what component is causing the current to increase. Parasitic noise/oscillations perhaps? It is indeed very weird that your multimeter is somehow interfering with the circuit when it shouldn't.

I don't quite understand the way that this circuit works. The transformers have 6 pins on the low-voltage side, with one set of 3 pins connected to the drains of both FETs and the other set attached to +13v through a capacitor and attached to ground via another capacitor:

+13v ---- | | --- (transformers) --- | | ---- GND

If the caps (SMD) form a voltage divider, they only pass AC according to my understanding. The one thing I thought would be bad here is AC - that will be only the ripple in the +13v from the power supply. Is that what's going on here?

AH!!! Just had a thought. What is the frequency?

The idea I had was if the cap on the inverter controller that sets the frequency went bad, the output frequency of the drive signal would go from say 30 - 40 Khz to well over 100 Khz. This would have all sorts of BAD effects, and maybe even explain the problem.

PlainBill

IIRC, the freqency of both the drains of the FETs and the ripple in the power supply are around 94KHz. I can re-measure this tonight.

I worked some on tracing out the circuits around the FETs and the inverter IC and found something I thought was interesting. One of the channels on the FETs has source attached to +13v, and the other has source attached to GND. Corresponding gates on the FETs are both driven by the same pin on the inverter IC, with one pin controlling the N-channels and one pin controlling the P-channels (these are pins 8 and 9, not sure which is which). If I understand this correctly, any phase shift or frequency variation in the gate drive signal might cause +13v and GND to be connected to the transformer at the same time (since all the drains are connected together, and all the drains are connected to half of the pins on both transformers). Not sure if that matters though, as the transistors should prevent a direct source-drain-drain-source short of +13v to GND, right?

Also I worked out that most of the pins 1-7 are connected to GND via one or two SMD components (usually caps, though a few are resistors). Those are probably the control components for the inverter IC's behavior. I can try to measure them in-circuit as my multimeter does have a capacitance function (range is small pF - small uF I think), but getting those little things off the board will be challenging if I have to remove them to measure them. I also have no idea what they're supposed to be as I don't have a parts list or a datasheet for the inverter IC. I did find a dual inverter control IC from BIT that has 4 extra pins, but the values of components it uses to adjust its behavior may be similar to this one.

[PlainBill]

Re: Samsung SyncMaster 940T blows inverter fuse

I checked the low voltage side of the inverter transformers with my oscilloscope and didn't see any noticeable variation in the RMS voltage or frequency on either transformer while current through the inverter was increasing. I also took a look at the return wires from the CCFLs, and they appear to be "or" circuits using the return voltages from the CCFLs attached to each transformer. They're connected together through a couple of resistors later, then that goes back to the control IC through a couple of other surface-mount components. I measured the voltage at the end of R345 after everything comes together, and the voltage and frequency measured there didn't change over time as current through the inverter increased (picture attached). Any other thoughts on what I could check?

AH!!! Just had a thought. What is the frequency?

The idea I had was if the cap on the inverter controller that sets the frequency went bad, the output frequency of the drive signal would go from say 30 - 40 Khz to well over 100 Khz. This would have all sorts of BAD effects, and maybe even explain the problem.

PlainBill

[momaka]

Re: Samsung SyncMaster 940T blows inverter fuse

Another thing to try is lifting J304 and inserting an ammeter.

Well I may have a brain fart here, but I was wondering if this test can be applied to J305/J310 as well - that is, have both J305 and J310 replaced by an ammeter. If the current in the inverter increases, shouldn't the current through J305/J310 increase as well (or perhaps decrease)? If yes, then maybe something is wrong with the FETs, the driving circuit, or the transformer(s). If no, then we need to find what component is causing the current to increase. Parasitic noise/oscillations perhaps? It is indeed very weird that your multimeter is somehow interfering with the circuit when it shouldn't.

As far as caps, it looks like C345 is the Vcc filter cap for the PWM controller of the inverter. I doubt this can mess up its operation, but I'm not expert in logic circuits so I could be wrong.
C107, C113, and C123 are part of the power supply, but I think you pretty much have the problem isolated to the inverter now (since you obtained the same results with the bench PSU) so it's likely they are not the problem either.
The blue caps near the CCFL connectors (C341, C342, C343, C344) are high-voltage ceramic type, most likely 2 kV or 3 kV. They rarely fail, and if they did, you would have noticed as they often burn up. You can change them if you want, but I don't think that would fix the problem.

One last thing - in other monitors, normally I see a cap or two near the driving FETs in the inverter, but this one doesn't have any. Maybe this is why you're seeing bad noise on the 13v rail. In which case, I wonder if this can be cured if a cap is placed between 13v and ground near J304 and J305. Of course, this suggestion is completely experimental.

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

Do you have any other monitors with the same lamp connectors? Those should be helpful to you.

Yes and no. I have 5 of these Samsung 940T monitors all with the same problem. I have a 19" wide LG that needs a recap (I'm pretty sure that'll fix it, I'm waiting for parts), a 17" that has a failing CCFL (it's going red), and an old BenQ 17" that I'm pretty sure just needs new CCFLs, but I haven't gotten around to ordering them yet (not sure on the connectors on that monitor). Most of this stuff doesn't actually belong to me, so I can't keep working units around for parts and testing; once something is fixed, it's re-assembled and returned.

What are you suggesting that I try? I still don't understand what swapping CCFLs around is going to accomplish. I have tried different sets of CCFLs with this IP board and did not see any difference in behavior. I don't want to risk damaging every lamp I have in case whatever is happening is over-volting the lamps, so I don't want to try all my parts on this IP board. I've tested the CCFLs with a spare inverter and they all work fine.

[seanc]

Re: Samsung SyncMaster 940T blows inverter fuse

Do you have any other monitors with the same lamp connectors? Those should be helpful to you.

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

I took a complete set of CCFLs from a cracked screen. I use that as my guidance to test whether the 2sec to balck is due to inverter or bad CCFLs.

I also kept the p/s, inverter board and control board for testing. I was tired of spenidng time/material replacing all the bad caps on the p/s only to find out the CCFL(s) is bad.

Now, I can test either part and start to rule out things.

So, once I opened up a monitor, I connect both set of lamps to my test jig. I power it up with a VGA signal. I let it run for about 5 min. If the CCFLs stay on for the whole time, then the lamps are likely OK. I can then check the p/s of the defective monitor for bad caps.

If the lights cannot stay on with my jig, then I swap in my test lights to check which of the 4 lights is defective. i do find some design requires the whole set to be swapped in, not sure why.

I'm pretty sure my issue is not related to the lamps. I separately tested the CCFLs and they're fine. Even if the lamps were bad, the inverter control IC should shut the inverter off if something is drawing too much current. The problem I have is that the inverter draws enough current over the course of about a 30 seconds to blow the inverter fuse at 3A. As I previously mentioned, if I put multimeters in series with the 13v to the FETs, this behavior is delayed until 5 minutes after power is switched on; the inverter uses less than 2A during this time. The fact that adding something in series with the 13v to the FETs that doesn't consume any meaningful amount of current changes the behavior would make me skeptical of CCFLs causing problems anyway, especially since the behavior is different with one meter in vs. both, and the behavior change follows a particular meter, not a transformer/FET/lamp. Anything is possible though - I did try switching CCFLs to different connectors on the IP board and that didn't change anything. I don't have a known working monitor of this model to use for comparison, sadly.

[lucky13]

Re: Samsung SyncMaster 940T blows inverter fuse

I took a complete set of CCFLs from a cracked screen. I use that as my guidance to test whether the 2sec to balck is due to inverter or bad CCFLs.

I also kept the p/s, inverter board and control board for testing. I was tired of spenidng time/material replacing all the bad caps on the p/s only to find out the CCFL(s) is bad.

Now, I can test either part and start to rule out things.

So, once I opened up a monitor, I connect both set of lamps to my test jig. I power it up with a VGA signal. I let it run for about 5 min. If the CCFLs stay on for the whole time, then the lamps are likely OK. I can then check the p/s of the defective monitor for bad caps.

If the lights cannot stay on with my jig, then I swap in my test lights to check which of the 4 lights is defective. i do find some design requires the whole set to be swapped in, not sure why.

[alexanna]

Re: Samsung SyncMaster 940T blows inverter fuse

Here's my best guess at what's going on. All I know for sure is that current into the inverter at the fuse on the 13v line increases until the fuse blows. Given that, something that uses 13v is drawing a lot of power. From what I can tell, the only things that the 13v goes to are the FETs via J304 and J307, and the transformers via C231 (not sure where else it goes from there). This suggests that either there's something wrong with the FETs as they do get hot, or there's something wrong somewhere through the transformers. Since the inverter IC's OVP is clearly working (disconnecting a lamp causes the inverter to shut off after a second), then it seems likely that either the OCP isn't working or the thresholds for shutting the inverter off aren't getting set correctly. It's also possible, as you suggested, that there's some other path for current to take to ground (causing the inverter IC to change the switching behavior of the FETs to compensate, causing them to draw more current, etc.). If that's true, then it's likely somewhere on the HV side of the transformers, else the hissing sound likely wouldn't happen (I don't know what else would make that sound); it's also not a dead short, else the lamps likely wouldn't light properly. Since adding multimeters to the 13v input to the FETs changes the behavior, I'm not sure there's another path to ground showing up somewhere (or do you have a theory that includes an extra path to ground that also explains the change in behavior when multimeters are introduced?). I'm wondering if the remaining ripple has an impact on this, though there shouldn't have been any while I was using my adjustable power supply (I wasn't watching on the oscilloscope though, so I don't know if there was any ripple). Also there might still be something up on the 5v line, though nobody seems particularly interested in investigating that yet.

Given that I have 5 of these things now that behave the same way, either it's a design defect, a manufacturing defect in all 5 (unlikely), or the same component/components in all 5 have failed (cheap parts/bad design/normal wear). Hopefully someone will correct me if I missed something in there.

I'm also curious how the ESR of the secondary filter caps impacts things. I used Nichicon HE on this one, which should be a good replacement for the CapXon KM that Samsung used. I also have Rubycon MCZ and Nichicon HZ that I could use here, but I'm not sure what difference that would make.



AFAIK the ring test is constantly running since the reading changes as I touch different pins. Since I don't have the top cover for mine, I didn't bother wiring up the test button so it's on all the time (I just unclip the power when I'm done with it). I have the Dick Smith kit, though I think most of these are based on the same design by Bob Parker if you want to search online for a schematic or other info.

Well you're asking questions I cannot answer. You have used a ring tester to test the inverter coils.
My mind keeps going back to the arcing in the inverter coil, If there was arcing I would suggest trying another set of coils from one of your other monitors and see if anything changes .There are a lot of reasons for a inverter fuse to fail, And hopefully one of the other monitors failed in a different way.
the only thing you have to loose is a little bit of time
Al.

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

My recommendation is a 'shotgun' approach. If you haven't already done so, replace every capacitor in the 14V supply with good quality low ESR types such as Panasonic FM or FC, or equivalent Nichicon or United Chemicon types. Also replace the electrolytic caps in the inverter area with as low an ESR cap as you can find. Then power it up and observe the current draw. If it starts to rise, replace both dual FETs. My first suspicion is that too high an ESR is allowing oscillations on the power supply line, resulting in the FETs not switching fast enough. The second possibility is that one of the FETs is failing to shut off properly.

PlainBill

Here are all the electrolytic caps I could find on the board, along with my proposed replacements (existing --> replacement):

• C105: CapXon KM 100uF 450v --> (not replacing)
• C107: CapXon KM 47uF 50v -->Rubycon YXF 47uF 50v
• C109: CapXon KM 1000uF 10v --> Nichicon HE 1000uF 10v
• C110: CapXon KM 1000uF 10v --> Nichicon HE 1000uF 10v
• C111: CapXon GL 820uF 25v --> Nichicon HE 680uF 25v
• C112: CapXon GL 820uF 25v --> Nichicon HE 680uF 25v
• C113: CapXon KM 330uF 10v --> Rubycon ZLH 330uF 10v
• C123: CapXon KM 10uF 50v --> Panasonic FC 10uF 50v
• C301: CapXon GL 820uF 25v --> Nichicon HE 680uF 25v
• C345: CapXon ST 33uF 25v --> Panasonic 33uF 50v

Possible alternatives for C111/112/301:

• Panasonic FK 2200uF 25v (replaces all 3)
• Panasonic FK 820uF 25v (backordered)
• Rubycon ZLH 680uF 25v

Possible alternatives for C109/110:

• Nichicon HZ 1000uF 10v
• Rubycon MCZ 1000uF 10v
• Panasonic FC 2200uF 10v (replaces both)

Possible alternatives for C113:

• Rubycon ZL 330uF 25v

Please let me know if these sound reasonable. I'm also not sure if any of the surface-mount capacitors around the inverter IC are electrolytic, and I don't know what to do with the blue caps.

For the FETs, AF4502C and AM4502C are almost exactly the same. I can't find a part that matches or exceeds all parameters, but the closest I've come so far is Vishay Si4563DY.

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

This is a question about poast#58.
And this a question I have.
Something is really bugging me on this monitor. The inverter circuit is able to detect an open ccfl and is able to shut off. The test I suggested to use one bulb and look at the brightness did not work as I suspected it would. But in the real world it is possible to have three bad ccfls and MOAST monitors should exhibit a sec to black issue before they self-destruct .I have not done the math and probably would get it wrong if I did, but if arcing did occur in the ccfl coil that cannot be good [insulation breakdown] I really suspect voltage finding another path to ground, if that is the cause or effect of the excessive amperage I am unsure. I also realize a scope has been used and a pattern observed, but if arcing is observed with sight wouldn't that trump a scope reading?

Here's my best guess at what's going on. All I know for sure is that current into the inverter at the fuse on the 13v line increases until the fuse blows. Given that, something that uses 13v is drawing a lot of power. From what I can tell, the only things that the 13v goes to are the FETs via J304 and J307, and the transformers via C231 (not sure where else it goes from there). This suggests that either there's something wrong with the FETs as they do get hot, or there's something wrong somewhere through the transformers. Since the inverter IC's OVP is clearly working (disconnecting a lamp causes the inverter to shut off after a second), then it seems likely that either the OCP isn't working or the thresholds for shutting the inverter off aren't getting set correctly. It's also possible, as you suggested, that there's some other path for current to take to ground (causing the inverter IC to change the switching behavior of the FETs to compensate, causing them to draw more current, etc.). If that's true, then it's likely somewhere on the HV side of the transformers, else the hissing sound likely wouldn't happen (I don't know what else would make that sound); it's also not a dead short, else the lamps likely wouldn't light properly. Since adding multimeters to the 13v input to the FETs changes the behavior, I'm not sure there's another path to ground showing up somewhere (or do you have a theory that includes an extra path to ground that also explains the change in behavior when multimeters are introduced?). I'm wondering if the remaining ripple has an impact on this, though there shouldn't have been any while I was using my adjustable power supply (I wasn't watching on the oscilloscope though, so I don't know if there was any ripple). Also there might still be something up on the 5v line, though nobody seems particularly interested in investigating that yet.

Given that I have 5 of these things now that behave the same way, either it's a design defect, a manufacturing defect in all 5 (unlikely), or the same component/components in all 5 have failed (cheap parts/bad design/normal wear). Hopefully someone will correct me if I missed something in there.

I'm also curious how the ESR of the secondary filter caps impacts things. I used Nichicon HE on this one, which should be a good replacement for the CapXon KM that Samsung used. I also have Rubycon MCZ and Nichicon HZ that I could use here, but I'm not sure what difference that would make.

Also when a ring tester is used is it a constant test [always running until you turn it off or a momentary test?]and how much of load is placed on the coil when the test is running?
Al.

AFAIK the ring test is constantly running since the reading changes as I touch different pins. Since I don't have the top cover for mine, I didn't bother wiring up the test button so it's on all the time (I just unclip the power when I'm done with it). I have the Dick Smith kit, though I think most of these are based on the same design by Bob Parker if you want to search online for a schematic or other info.

[alexanna]

Re: Samsung SyncMaster 940T blows inverter fuse

This is a question about poast#58.
And this a question I have.
Something is really bugging me on this monitor. The inverter circuit is able to detect an open ccfl and is able to shut off. The test I suggested to use one bulb and look at the brightness did not work as I suspected it would. But in the real world it is possible to have three bad ccfls and MOAST monitors should exhibit a sec to black issue before they self-destruct .I have not done the math and probably would get it wrong if I did, but if arcing did occur in the ccfl coil that cannot be good [insulation breakdown] I really suspect voltage finding another path to ground, if that is the cause or effect of the excessive amperage I am unsure. I also realize a scope has been used and a pattern observed, but if arcing is observed with sight wouldn't that trump a scope reading?
Also when a ring tester is used is it a constant test [always running until you turn it off or a momentary test?]and how much of load is placed on the coil when the test is running?
Al.

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

This suggests it's similar to BIT3105 and a Fairchild FAN7311

Note: Company name is PIS Corp.

I had looked at the BIT3105, but it has a different number of pins (20 vs. 16 on the BIT3195G) so they must have reworked their PCB. There are a few variants of the FAN7311, so I'll have to pick through them to see if any have 16 pins that appear to line up with what I see on these boards. I'm slightly curious if attempting to match up pin function to the circuit might be worthwhile. Some manufacturers have sample circuits in their datasheets and I might be able to use one of those to figure out what's what here.

[stevenps]

Re: Samsung SyncMaster 940T blows inverter fuse

Here's the manufacturer but they aren't so big on data sheets.
Probably have to email and hope they read English.
http://www.bitek.com.tw/e-company.htm

Not a bad idea. I might give that a try sometime this weekend.

I picked up yet another one of these monitors today, but this time someone had already attempted a repair and failed (from the sound of things, he only attempted cap replacement and didn't do anything else). The board isn't damaged badly enough that I can't salvage it, and he said that it had the usual bad power supply symptoms. If cap replacement is enough to make this one work, I may have a functional monitor to help diagnose the ones that don't work. The telltale sign will be if the inverter fuse is blown, which I can check tonight. Otherwise, I have yet one more 940T to fix.

[PCBONEZ]

Re: Samsung SyncMaster 940T blows inverter fuse

This suggests it's similar to BIT3105 and a Fairchild FAN7311

Note: Company name is PIS Corp.