Author Topic: Clicking Switchmode Supply for TV & Identifying SMD Switchmode Chip (Read 7274 times)

BAM5 · « **on:** September 15, 2015, 01:10:32 pm »

Hi friends! And not friends! Just, people in general!

I got a free tv off craig's list (broken) and figured I'd give it a crack at fixing it, I've fixed a few power supplies before, however, this time, it isn't the caps that are the culprit! Let me share with you my observations and see if we can figure this thing out!

Symptoms:
Clicking noise originating from the transformer at about .75 Hz (Less than one click per second)
No voltage on low side of the board, however there are tiny pings of voltage every time the transformer clicks.

Tests:
No caps seemed bad so removed and tested them all with Fluke to be sure. Noticed a weird thing with the large ac filter cap where the voltage would slowly go back up after discharging. Perhaps just a property of larger caps?

Various probing with DMM:
Filter cap oscillates between 158 & 163VDC
Ground to what appears to be sm (switchmode) chip Vsupply varies between 10 and 16 volts. Has the same frequency as the clicking, however it's shifted in that the voltage drop to 10 (from ~16) happens a quarter of a second after a click.
Optoisolator from low to high side that controls an input on the sm chip gets a burst of pulse frequency the same as that of the clicking. I assume this is just because it the low side isn't getting enough voltage and the voltage signals are pulsing back through the isolator and is working as intended.
Transformer driving transistor has the same pulse frequency as the clicking (because it is driving the transformer which is causing the clicking)

Possible conclusions:
Switchmode chip is bad
transformer is shorting, causing the filter cap to deplete of enough energy to cause the sm chip to cut out.
... Some other component is bad...

If the conclusion is that the sm chip is bad, how can I identify it? It's a tiny 6 pin smd chip with no significant markings only "JP 35", or "JP 15".

Thanks!

bktemp · « **Reply #1 on:** September 15, 2015, 01:18:55 pm »

Clicking sounds -> Check all the rectifier diodes on the secondary side. Either those diodes are shorted or something is drawing too much current.
There are some other possible failures, but a shorted output is the most common one.

BAM5 · « **Reply #2 on:** September 15, 2015, 02:57:19 pm »

All diodes are correctly oriented, I also desoldered and tested them and they all tested good.

Could the short possibly be in the transformer?

bktemp · « **Reply #3 on:** September 15, 2015, 03:02:10 pm »

A short inside the transformer is possible, but rather rare. A shorted transformer winding often destroys the switching transistor because the current limit can not switch it off fast enough.

Did you try to disconnect some loads on the secondary side (if possible)?
Did you check the small capacitor on the SMPS controller supply?

BAM5 · « **Reply #4 on:** September 15, 2015, 03:21:59 pm »

The board is completely removed from the TV so there is no load on the secondary side.

C104 is reading at 24nF when removed from board.

Would it be possible to completely disconnect the secondary side of the coil by desoldering it and see if it works that way to see if there is a short on the secondary? Or would that be bad for the circuitry?

bktemp · « **Reply #5 on:** September 15, 2015, 03:37:34 pm »

Never run a flyback SMPS with the feedback loop disconnected! If there is no feedback the control circuit continues to pump energy into the transformer until something breaks!
Look for a 3 pin device next to the optocoupler. This is a TL431 for controlling the feedback loop.
If you have a scope look at all outputs to see if the voltage starts to rise or if they are stuck at almost zero volts.

There is probably another capacitor somewhere on the SMPS controller supply. Typically 10-100uF.

BAM5 · « **Reply #6 on:** September 15, 2015, 04:05:24 pm »

Yeah that's what I figured

But I also figured it's not running the transformer anyway :S Thus why I asked.
Yep I see it. I don't have a scope yet. Is the DS1054Z still the ahead of the market for entry level scopes? In the mean time I'll attempt to get the same information with my fluke.

Oh, yeah, there is an electrolytic cap, but I checked it along with all the other electrolytics on the board. Is it safe to say that if it holds capacitance from the voltage from my DMM that it will also hold the higher voltages it receives from the board?

Richard Head · « **Reply #7 on:** September 15, 2015, 04:32:31 pm »

Provide the unit with an external isolated 16V DC supply from a bench PSU or similar. Don't worry about connecting mains yet. Then check the gate drive of the MOSFET. It should be a square wave with almost 100% duty cycle. This indicates that the chip is working. If you have a variac connect it to the AC input and gradually increase the variac while watching the DC output voltage. It should come up with less than 50VAC input. Gradually increase the AC input voltage until the output voltage reaches the correct value.
If all is still ok then increase the AC input voltage to 230VAC (or 110V if applicable).
Place a load on the output and confirm that the output voltage remains in regulation.
Remove the 16V DC input from the external PSU. and see if the unit continues to operate. It probably won't. That indicates some component in the feedback winding is probably faulty like the electrolytic cap that's across the chip supply pin and GND.
Good luck and be careful around mains.

BAM5 · « **Reply #8 on:** September 15, 2015, 05:24:45 pm »

Hey Richard! Thanks for the reply!
I used 2 9 volts to create a voltage source as my PSU is not with me. When I checked the gate on the mosfet I found that there didn't seem to be any voltage, let alone pwm on it. (My fluke has a duty cycle setting.) Funny thing happened when I hooked it up to the contacts for the voltage with the mains plugged in... It stopped clicking! But it also never sent any power over to the secondary. This would lead me to believe that the switchmode controller is faulty. Is this correct?

Richard Head · « **Reply #9 on:** September 16, 2015, 06:00:47 am »

It's possible but the fact that the transformer is clicking every second or so suggests that some of the chip functions are operating correctly. You really do require a scope and PSU at least to investigate the problem. There may be a short on the DC output. It's worthwhile checking that with a meter.

T3sl4co1l · « **Reply #10 on:** September 16, 2015, 06:20:46 am »

The chip is doing precisely what it's supposed to be doing:

During startup, there is initially no supply voltage to the chip.
Current is supplied from a chain of resistors from the rectified mains. This charges the smaller filter cap.
When supply is high enough (usually 14-18V), the chip wakes up and starts driving the transistor. As it goes, it's drawing more current (usually 1-20mA) from its supply, which is more than the charge current, so the supply starts pulling down.
If the chip isn't able to supply itself, it shuts down as the voltage drops out, usually around 8-12V.
If the chip turns off, supply current goes low again, and the filter cap resumes charging.

This is what you're hearing. The transformer clicks audibly (due to physics of the core material, or loose windings), then it turns off for a while before it goes again.

When you power the chip externally, it keeps operating, so you don't hear clicking. It's just running flat out, all the time.

In normal operation, the chip's supply droops down during the first powerup cycle, while delivering enough power to get all the voltages (coming from the transformer) up to operating level. Which includes the auxiliary output on the mains side, so that the chip can be self-powered. If all the outputs come up to normal voltages, then the aux side will as well, and the chip stays running. If one or more are kept low (due to excessive capacitance, or a heavy load, or a short circuit), it will give up, wait, and try again, maybe succeeding after a few tries (each time, ratcheting up all the voltages, until it can keep itself operating).

So your problem is absolutely a classic brown-out or short symptom. One or more of the output voltages has failed, somehow. It could be a shorted transformer (as mentioned: likely to take out the switching transistor as well -- but not necessarily guaranteed), shorted diodes, capacitors, loose wires, failed resistors (I suppose it's possible that a load resistor has failed with a low resistance), who knows. You will find a low resistance somewhere; it might not be obvious (a normal transformer still measures ~0 ohms at DC), or it might be something dumb and tricky (a capacitor that's failed in such a way that it acts like a zener diode, only presenting a short at higher voltages??), but it's something. Keep looking.

Tim

BAM5 · « **Reply #11 on:** September 16, 2015, 06:19:14 pm »

Thanks Tim! Your insight as to how a smps operates is very helpful. However I see a slight inconsistency with what's happening and what you said. You said that when I power the chip externally that it keeps operating and running flat out all the time. The thing is, the power mosfet appears to be completely open and there is no current traveling through the coil when I do this. Is that what you are describing?

I tested the power FET and it appears to be working properly.

Now what you said about there being a short in one of the outputs, this makes me very tempted to desolder a winding or two to figure out which leg of this board has an issue. There are 6 output leads on the low side of the transformer that are connected to 4 distinct traces (2 leads of the transformer attach to the same trace in 2 spots). I figure if I disconnect some of these leads and plug it in for a short time and it doesn't click I should be able to figure out which of the traces the leads are shorting on. I'd like to do this, but am curious if there is any other way as I have been warned against this.

bktemp · « **Reply #12 on:** September 16, 2015, 08:24:17 pm »

Maybe this SMPS controller senses the input voltage and has a undervoltage lockout. Many modern controllers also have an open loop detection: If there is no feedback signal after a fixed time period, the controller shuts down.
The method Tim described works great for the classic 3842 based SMPS, but fails on more modern ones. They can be really difficult to troubleshoot. Without a DSO it is almost impossible, because you need to be able to capture the signals in the short time inverval when the controller starts up until it shuts down.

I do not recommend disconnecting secondary side diodes without knowing where the feedback is connected to. If there is no load and no feedback the voltage on all other outputs rises until something breaks.

BAM5 · « **Reply #13 on:** September 16, 2015, 08:45:08 pm »

Alright, well I guess I'm buying a Oscilloscope then. Is the DS1054Z still the recommended scope for entry level?

T3sl4co1l · « **Reply #14 on:** September 17, 2015, 05:26:45 am »

Yeah, the switching pulse width while operating might be very short already, and the duty cycle of the slow pulsing is also very short. If the transformer isn't shorted, then the controller will still be developing full current during that pulse, but it won't be measured on the primary side, but the faulted secondary.

Though you should still be measuring some small current from the primary side, too..

When it's ticking away, there should be a very small voltage on the faulted channel (it might be mV), and perhaps a measurable voltage drop as you follow traces from the transformer to the fault. The other channels should have some voltage, perhaps 10s of mV or 1V even, or there may be faults there too.

This isn't a very good justification for a scope, but if you were simply looking for an excuse... go for it.

Or if you want to save some money, anything in the $100-200 range from eBay that's working should be okay. Tek 465 and etc. is pretty old and likely to have various geriatric problems, but anything 2000 series should be pretty usable. HP 546xx series is good early DSO stock, easy to use. Or a number of less prominent but excellent brands like Fluke, Philips, Hitachi, etc. If you're going to go with a newer model, I'd probably suggest avoiding any of the slightly cheaper off brands, or the earlier models (the early Rigol 1000 series were worse than the TDS1000s, and that's saying something), and yeah, 1054Z looks like a great deal.

Tim

BAM5 · « **Reply #15 on:** September 17, 2015, 07:31:01 pm »

I really was just looking for an excuse

Wanted one for a few years now. Just ordered one from TEquipent with the discount. I'll probably get it next week, so before then I'll rewatch the how to blow up your oscilloscope vid and then fool around with this board. I'll update when I do. Hopefully I'll have fixed it. Er, well, diagnosed it at least. If not I'll come crying back to here

Monkeh · « **Reply #16 on:** September 17, 2015, 08:34:34 pm »

I'm in the process of repairing a monitor with the exact same symptoms (clicking, no life). In this case I've found a shorted diode and FET driving the backlight transformer. Easily located with a multimeter. Should be back to life with whatever the hell FETs I found which are more-or-less similar to the originals tomorrow. And the (schottky) diode, which I happen to have in stock.

BAM5 · « **Reply #17 on:** September 17, 2015, 10:41:15 pm »

Hey Monkey, That's interesting, might I ask the method you used to find out what was wrong? Did you just take parts out you figured would be problematic and test them? The board I'm working on has a good bit of smd stuff so testing every component would be a huge hassle for me. Also, backlight transformer you say? Is the backlight CCFL? There's an LED Driver chip on this board OB3362HP.

Monkeh · « **Reply #18 on:** September 17, 2015, 10:53:58 pm »

Quote from: BAM5 on September 17, 2015, 10:41:15 pm

Hey Monkey, That's interesting, might I ask the method you used to find out what was wrong? Did you just take parts out you figured would be problematic and test them? The board I'm working on has a good bit of smd stuff so testing every component would be a huge hassle for me. Also, backlight transformer you say? Is the backlight CCFL? There's an LED Driver chip on this board OB3362HP.

Educated guess and examination of the board.

Taking out the links which connected different sections of the supply helped and was quick.

Yes, in my case the backlight was CCFL. Your issue may be different. Break the board into functional blocks, isolate them (removal of components, links, or even trace cutting), and probe.

AF6LJ · « **Reply #19 on:** September 17, 2015, 11:19:30 pm »

Worked on these before;
You have a short or very low resistance on one the outputs from said power supply.

BAM5 · « **Reply #20 on:** September 18, 2015, 01:07:49 am »

Quote from: Monkeh on September 17, 2015, 10:53:58 pm

Break the board into functional blocks, isolate them (removal of components, links, or even trace cutting), and probe.

I keep wanting to do that, but I keep getting warned against it

Although now that you mention it there are quite a few jumper wires and smd jumpers that I can remove but still have the smps feedback and what looks like a minimum load resistor still included in the circuit, so actually, I'll check that out!

Going on a weekend trip tomorrow though so can't do it over the next few days. Thanks for the info though, I appreciate it!

Quote from: AF6LJ on September 17, 2015, 11:19:30 pm

Worked on these before;
You have a short or very low resistance on one the outputs from said power supply.

Thanks AF6LJ, that's been the general consensus so far

BAM5 · « **Reply #21 on:** September 30, 2015, 02:22:38 am »

All right, after getting my scope (Love this thing btw <3) And building an isolation transformer out of a couple old UPS units I found a bad diode in the mains voltage rectifier. Went to the local electronics shop and payed $1.50 for a new one... Didn't fix it. But hey, now there's a nice steady charge on the filter cap.

After some more probing around I believe I've found the culprit. A small SMD component which I assume is a zener diode. It's in a sot23 package and has a label on the silk screen that says ZD102. The diode is between the power rail and the ground for the SMPS controller so I assume it's a regulator for the rail. Thing is, its breakdown voltage is either ~0.5V or ~0.9V according to my 87V's diode test function Neither of which seem like the correct voltage since the filter cap for that rail is rated for 50V, and the aux winding on the transformer is sure to output significantly more than 1V.

So my question to you fellows is, how can I find a replacement for this component. It doesn't appear to have any markings on it so I have no way to discern what specs I need for a replacement...

Update: Managed to get some alligator clips on that tiny diode. Hooked it up with a 12VAC transformer and my scope and saw the full waveform, no flat spots like I'd expect with a zener. Just a perfect sinusoidal wave. So Yeah, back to the question, how do I figure out what characteristics I need for the replacement diode?

dave_k · « **Reply #22 on:** September 30, 2015, 05:26:11 am »

Have you been able to identify the part number of the SMPS controller IC? Having a loon on the data sheet (if available) might yield a "typical application" schematic.

tautech · « **Reply #23 on:** September 30, 2015, 07:17:47 am »

Quote from: dave_k on September 30, 2015, 05:26:11 am

Have you been able to identify the part number of the SMPS controller IC? Having a look on the data sheet (if available) might yield a "typical application" schematic.

+++++1

So often the SMPS IC datasheets typical application schematic is exactly or very similar to what's found in real life. It should always be the first thing you check for.

BAM5 · « **Reply #24 on:** September 30, 2015, 12:42:48 pm »

I thought of that, but it is also difficult to identify the smps controller as it too is a smd component in a sot23-6 package. I can barely make out JP 35 or JP 15 on the smps controller. Tried Google but couldn't find anything.

Anyone have some good resources for identifying components?


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Clicking Switchmode Supply for TV & Identifying SMD Switchmode Chip (Read 7274 times)

Share me