Possible bad transformer.

[Th3_uN1Qu3]

Got a little stereo in for repair today. It's a pretty nice one, it has DVD playback with Dolby Digital, a digital TV tuner, and a built-in 7" screen.

Trouble: No power. Owner said he replaced the controller IC (FAN7554) and the mosfet because they were visibly cracked. But still no go.

I found a bunch of bad parts including the current sense resistor, the opto, a SMD 1N4148, some resistors, and i eventually ended up blowing both the FAN7554 and the FET. Oops. A quick look around showed that the FAN7554 isn't easy to come by, so i adapted a UC3842 (they are pretty much the same anyway, the only thing i had to do was to lift pin 2 from the PCB and wire it to primary ground, the other pins line up), and put in another FET for testing.

I got voltages to come up on the secondary but only in bursts, and not up to spec. The supply didn't even come into regulation. I looked for secondary shorts: clean. Removed all rails but the regulated 5v one: still doing the same thing. I have a 60W bulb in the primary, it lights up, goes out, lights up, goes out, in sync with the Vcc on the UC3842. If i ensure a steady supply to the 3842 the bulb lights up fully and remains lit.

Watching the scope, i noticed that as soon as the duty cycle goes past 10% or so, bam, bulb lights up, power goes out, rinse and repeat. I think the transformer saturates. It gets a bit warm too... I pulled the transformer and it shows very low inductances on all windings: The primary is 27uH, the aux is 9uH, and secondary windings are so low inductance that they can't even be measured with the LC function of my DMM. I've never seen such a thing before - the primary should have been a couple mH at the very least. This is a flyback transformer and likely has an air gap, but still, to have only 27uH primary inductance at 100kHz switching frequency seems questionable at the very least. It doesn't appear shorted tho - could the core have demagnetized?

Your guess is as good as mine.

[Th3_uN1Qu3]

Re: Possible bad transformer.

Okay, boiled the thing, core came open easily. It's an ERL-28L btw, with a ~2mm gap. The demagnetized core theory fails btw, because i had another one of the same, but without a gap. I tried the other core and i got 47uH on the primary... still awfully low. When i tried this gapped core in the other transformer tho, i got 450uH. Closer to what i should have been seeing. So the core is good.

I think i have shorted turns in the primary. Since i boiled it in water i'll let it dry until tomorrow, and i'll try the other core to see if it makes a difference, then try a 5vsb transformer to make sure the controller and FET behave good. If the 5vsb transformer works then i'll call the guy who brought it in and ask if he's willing to pay me to get the transformer rewound, then we'll see.

[Th3_uN1Qu3]

Re: Possible bad transformer.

Confirmed bad primary. Same thing with the other core. I wired a 5vsb transformer to the supply, and bam, 5v is up and running and within tight regulation, auxiliary runs okay too at 21v. So the rest of the supply is OK, the transformer is bad.

FYI, the numbers on the 5vsb transformer i've wired in are 13uH for the aux, and 1.3mH for the primary. Quite a big difference from 9uH/27uH don'tcha think?

[Rulycat]

Re: Possible bad transformer.

Sounds good. I thought you might have felt since no-one replied no-one was interested.

Do you rewind cores yourself? What's the air gap for? There's still a lot I need to know.

[Th3_uN1Qu3]

Re: Possible bad transformer.

Sounds good. I thought you might have felt since no-one replied no-one was interested.

On this forum, i don't think so. When nobody replies to my threads, i take it that maybe it's a new situation for you as well, and you don't want to say anything stupid.

Do you rewind cores yourself? What's the air gap for? There's still a lot I need to know.

Yes, i rewind them myself, i've also designed and wound several from scratch for the supplies that i built. SMPS transformers aren't difficult to wind by hand, unless you need crazy voltages or want to use really small cores, the number of turns never exceeds 50-60 on any winding.

The air gap is only used in flyback transformers. A flyback transformer isn't a transformer per se, as the voltage transformation doesn't follow the volts/turn ratio as in normal transformers, instead it works as an inductor with multiple windings, and what is kept constant is the number of amps/turn. The flyback is useful because it needs relatively few turns to make high voltages, and also it does not need an output inductor (since the flyback transformer IS an inductor), hence its popularity for cheap, multiple output supplies.

A normal transformer transfers energy when the transistor is turned ON. A flyback transformer, being an inductor, stores energy when the transistor is turned ON, and transfers it to the secondary when the transistor turns OFF. This energy has to be stored somewhere. This is the purpose of the air gap - the energy isn't stored in the magnetic material, but in the air gap. All inductors that work with DC bias have an air gap - for example the toroid cores commonly used in PC PSU outputs, have non-magnetic particles distributed between the powdered iron particles, creating what's called a "distributed air gap".

This guy's not answering his phone, i'll take a break and call him again tomorrow.

Either way, i franken-wired another power supply, a battery and a few resistors to get the necessary voltages, to know if the device itself survived the major failure in the SMPS. This thing has a crapload of voltages. Apart from the usual 5v and 12v, it also has -23v (it uses a VFD), and another unknown voltage (not marked anywhere) that i eventually figured out to be 9v. With just 5v, 12v and -23v, the VFD wouldn't light up. When i found that the filaments weren't getting power, i found a circuit going to yet another winding of the transformer. 3v? No luck. 5v? Nothing. A 100 ohm resistor from the 12v battery got me 6.4v, and this produced a dim text on the VFD. Attaching a second one in parallel and the VFD got pretty bright, voltage now reads 8.15v. Menu works fine. At least i know it's alive. So i think it should be 9v.

And there's yet another voltage for driving the Class D audio amplifier (TPA3123, pretty nice chip), but i didn't need that one to get the thing up and running.

[Rulycat]

Re: Possible bad transformer.

Thanks for that write up there Unique. I'm ordering that power supplies A to Z book tomorrow. Going to be a fair bit of work though, because I'm not at university yet, and there's no electronics course at the college I'm at. The problem is there's so much cause-and-effect whereby something affects another thing which affects another thing which then affects the first thing some more, my brain gets frazzled easily.

[Th3_uN1Qu3]

Re: Possible bad transformer.

The one i started with is "Switching Power Supply Design, Third Edition." I found that easy to understand for the most part. It starts with linear regulators, then moves on to simple SMPS topologies like the buck and the boost, and then goes to explain all kinds of SMPS known to man. The one you mentioned has very good explanations on the more complicated aspects of SMPS design, such as control loop compensation, but it has A LOT of math which can sometimes be avoided.

[Rulycat]

Re: Possible bad transformer.

Sorry to begin derailing the thread but can you recommend any books on general electronics I should read first? I've read Electronics for Dummies but it's far too basic, such as "This is what a capacitor looks like" etc. Thanks Unique and I'll be sure to start posting more informed replies as time goes on.

[Khron]

Re: Possible bad transformer.

Rulycat, i can assure you of one thing: it's (nearly?) pointless to count on formal education, for useful or practical information on electronics (for instance). Hell, between the basics that dad taught me, and the stuff i've learned from here and the DiyAudio forums, school (as in high-school (electronics & telecommunications) and university (automation engineering)) have taught me nearly jack about what i perceived as useful and practically-applicable knowledge in this field

Rod Elliott also has more than a few really well-written articles on a great number of electronics- and audio-related topics ( http://www.sound.westhost.com/articles.htm )

[Th3_uN1Qu3]

Re: Possible bad transformer.

Okay, progress. I believe i have identified the fault that blew the power supply in the first place.

I thought i'd test the LCD screen as well. I started connecting the supplies one by one, and when i wired the battery for 12v... smoke. One of the LCD connectors overheated and the wires started melting. I disconnected the 12v rail and proceeded to start the system up with just the others. I could saw something changing on the screen when i powered up the 5v, so the panel is likely okay. I'm betting on a failed inverter. Time to take it apart.

[b700029]

Re: Possible bad transformer.

Yes, i rewind them myself, i've also designed and wound several from scratch for the supplies that i built. SMPS transformers aren't difficult to wind by hand, unless you need crazy voltages or want to use really small cores, the number of turns never exceeds 50-60 on any winding.

That sounds really low. What are the typical number of turns for e.g. a standard ATX supply with 1 primary and 2 center-tapped secondaries for 12v and 5v?

[Th3_uN1Qu3]

Re: Possible bad transformer.

In a standard halfbridge ATX cheapie you have 30 to 40 turns for primary (most have 33 or 35), 3 turns for 5v, and another 4 in series with those (for a total of 7 turns) for 12v.

Coming back to our stereo here, i took the LCD apart to find THIS (see pic). Yes folks, that is a tantalum capacitor that has blown and gone short. That little bugger killed the whole supply. Had the power supply been better engineered, this disaster could have easily been avoided. But more on that later.

It was 47uF/20v. Replaced temporarily with a random 100uF 25v lytic, will search for a better replacement the following days. Everything's working fine now, all that's left is to rewind that transformer.

[ben7]

Re: Possible bad transformer.

In a standard halfbridge ATX cheapie you have 30 to 40 turns for primary (most have 33 or 35), 3 turns for 5v, and another 4 in series with those (for a total of 7 turns) for 12v.

Coming back to our stereo here, i took the LCD apart to find THIS (see pic). Yes folks, that is a tantalum capacitor that has blown and gone short. That little bugger killed the whole supply. Had the power supply been better engineered, this disaster could have easily been avoided. But more on that later.

It was 47uF/20v. Replaced temporarily with a random 100uF 25v lytic, will search for a better replacement the following days. Everything's working fine now, all that's left is to rewind that transformer.

Wow, crazy little tantalum bugger!

Amazing that it made the psu transformer get fried too!

Time for modding!

BTW, you should design a super good and reliable computer smps/psu and sell kits for it xD

[Th3_uN1Qu3]

Re: Possible bad transformer.

Got the OK from the owner regarding the price so i'll be rewinding the transformer.

Here's some more pics from last night. The USB connector on the front had its plastic tab broken so it was very touchy, i replaced that as well. I pulled one from an old laptop motherboard, and the one i pulled was sitting on the board, while the one on the stereo was mounted upright. A little intervention from the latest addition to my tool set fixed that.

Also are a few pics of the whole thing in operation and the paraphernalia required - it isn't my fault this thing uses half a dozen different voltages! I found a suitable replacement for the tant on the LCD too, on the same laptop board i pulled the USB from.

[Th3_uN1Qu3]

Re: Possible bad transformer.

Unwound the transformer, and it confirms my findings. Tantalum cap on 12v rail went short when no one was watching. 12v rail had a single 0.3mm wire - too thin to trigger the primary overcurrent protection. Insulation melted and it shorted with the 5v rail. Switching transistor keeps pumping more current into the primary in an attempt to keep the 5v regulated, auxiliary voltage rises, and the whole thing goes boom.

No visible damage to the primary tho, but judging by the inductance it was definitely shorted as well. The way it was wound was atrocious - each half primary has two layers (it's just 40 turns in total, 20 on the inside, 20 on the ouside, but for some reason they only wound on the middle of the bobbin and left a lot of space on the top and bottom), and there was no insulating tape between the two layers of each half primary. That's a sure recipe for disaster. Worse, the melted windings went thru the tape and were less than 1mm apart from shorting to the primary. Imagine the destruction and the safety risk that would have been.

Now let's mention what could have been done to avoid this disaster. If a zener would have been used on the aux winding, as soon as the aux rose beyond limits it would have shorted, brought the controller Vcc down, and a lot of the damage would have been avoided. But simpler than that: A plain, simple FUSE on the 12v rail would have made this repair a breeze. Tantalum shorts, fuse blows, 12v goes out, everything else keeps working, easy repair. But no. They were too inept to do that, and their idiocy resulted in this catastrophic failure. I will be winding the 12v with two strands of wire instead of just one, and also fusing all rails and adding a zener to the PWM controller supply.

Sorry, no pics, i was too busy noting down the starts and ends and number of turns on each winding. At least they used 0.3mm wire throughout, with several in parallel when higher current was required, so it's gonna be cheap to rewind.

Now let's talk a bit about the hardware used in this thing.

Main processor is a Cheertek CT219RZ-LF, which i didn't find any data on. It's in a rectangular TQFP package either 100 or 144 pins, can't be bothered to count them and they aren't numbered on the PCB.
Next to it is a CT901A-LF which could be the DSP/Dolby Digital decoder (thanks Chris for the suggestion), again no data on it.
In the same area there's a Winbond W9864G6GH-6 which is a 8Mbyte SDRAM chip, and a Eon EN25F80-100HCP 1Mbyte Flash memory, and a AT5669H which is the motor controller for the DVD drive.
There's also a bunch of vregs, some digital logic chips and stuff like that.

Next up we have a Princeton PT2314 4-channel input audio processor, a couple more logic ICs, and an Afa AF9013AS DVB-T demodulator, in charge of the TV tuner.
VFD display driver is again by Princeton, a PT6311.
Audio amp is a class D, 25W/ch TI TPA3123D2 which is a pretty amazing chip - the supporting caps and inductors take like 10 times the space of the IC itself!
LCD controller is a MStar MST720A-LF which is a component/s-video input chip designed specifically for car LCD applications. An analog input LCD, just as i suspected. That explains the interference i'm getting on the screen. It's gotten better since i put a proper tantalum instead of the lytic i initially tried in place of the blown tant, but the other tantalums may need replacement as well.

[tom66]

Re: Possible bad transformer.

It's kind of incredible how bodged together these multi-function devices are. For example, why have both a VFD, and an LCD? Why not simply use the LCD and embed an OSD for control? It's just like that DVD player I repaired which used a massive transformer for the audio and then a switching supply for the rest, why they didn't just buck down the 30V from the transfomer I don't know.

Interesting failure mode. Many CE certifications require the power supply to remain safe when rail shorts occur so I wouldn't be surprised if that was the China Export edition of the CE certification.

Also one thing I've noticed with flybacks is the capacitor ripple current is often substantially high due to them operating discontinuously - is there a way to make a continuous flyback?

[Th3_uN1Qu3]

Re: Possible bad transformer.

It's kind of incredible how bodged together these multi-function devices are. For example, why have both a VFD, and an LCD? Why not simply use the LCD and embed an OSD for control?

It *does* have an OSD. Actually, the VFD is pretty useless - the source selection for example, only appears on the LCD! The VFD just sits there showing "MENU" at that time. It only shows progress in the current track/video, channel and frequency for TV and FM (possibly RDS as well, but the FM tuner is screwed into the case, so i disconnected the cable and left it there, will test once it's all back together again), whether a disc is inserted in the DVD drive, "NO USB" if you attempt to select USB without a device inserted... and that's about it. Everything else shows on the LCD screen only.

Many CE certifications require the power supply to remain safe when rail shorts occur so I wouldn't be surprised if that was the China Export edition of the CE certification.

You can bet on it. This is the worst power supply failure i have seen so far.

is there a way to make a continuous flyback?

Yes. But it complicates control loop compensation, hence why they are generally avoided. Recently there have been special ICs developed that work at the transition between continuous and discontinuous at all times, thus getting the best of both worlds. But y'know... they cost money.

[ben7]

Re: Possible bad transformer.

LOL! "Cheertek" !!!

We should start a thread about dangerous chinese products

And its amazing how many people buy these super cheap things, and yet still think that they should last a long time!

[Th3_uN1Qu3]

Re: Possible bad transformer.

LOL! "Cheertek" !!!

Apparently they bought Winbond's flat panel division in 2006 (never knew Winbond made displays), but also they don't apparently, you know, uh, exist. At least i can be "cheer"ful that it's still alive. Hmm. They appear to have been bought out by Novatek in 2007. That explains why i couldn't find much about them or their products.

And its amazing how many people buy these super cheap things, and yet still think that they should last a long time!

Don't think it was exactly cheap. Also, the menu is in French, so i assume this has been bought or imported from France. It likely cost a pretty penny.

Tbh, if i saw this in a thrift store, i too would think it's just a cheap gimmick, but actually, i'm pretty amazed at the engineering that went into this thing (apart from the stupid PSU of course). This is all on one huge board, but everything's divided nicely into sections, clearly marked on the silkscreen. The board is all double sided, and has plated thru-hole. Too bad it doesn't give any clue to who actually made it.

As for the power supply, it ain't THAT bad actually. Apart from the obvious omissions which brought its demise, there are pi filters on all outputs, and most caps are Aishi, with a few by Cheng. Aishi aren't bad, i'd compare them with the likes of Samxon. It's certainly better than having all different noname brands scattered throughout the whole board.

[momaka]

Re: Possible bad transformer.

On this forum, i don't think so. When nobody replies to my threads, i take it that maybe it's a new situation for you as well, and you don't want to say anything stupid.

That would be people like me .
I do often bookmark and refer back to your threads for knowledge, though .

It's certainly better than having all different noname brands scattered throughout the whole board.

True.
Someone brought a car DVD player at my old work last week and I took the task of opening it. My friends there were wondering why I was giggling the whole time. Well, if I had told you the cap brands I saw in it, you'd smile a little as well. I remember two of the brands... one was "Wincap", the other "Maxcap". I'm still chuckling a bit when I say those names. Helps if you say it in a funny voice with the right intonation too .