Bizarre transformer like item. Possible RF filter?

[BrandonGlatz]

Trying to troubleshoot the power supply on a very expensive projector that I was given due to it not working. I wouldn't mind having another projector and my last projector repair attempt failed.

Here is a link to an imgur album with pics of this component: http://imgur.com/a/eF5zz

I have no idea what this thing does. At first I thought it was a transformer and that it went open, but when I desoldered it I found that the pin wasn't connected at all. My guess is that it has to do with RF, but I've never seen this before. I dont /think/ it could be for feedback for the switching fets, but I don't know for sure.

Any help would be appreciated.

Here is an album with pictures of the board: http://imgur.com/a/4fz58
Sorry the back of the board is so dirty, I ran out of alcohol.


Here is a quick schematic (simplified) of what is going on:

Thanks!

[SeanB]

Choke with a pair of shields on the outside to reduce EMI. There should be continuity between the 2 pins not connected to the shields, and they should be isolated from the shields.

[mikeselectricstuff]

Inductor for power factor correction circuit - this is basically a boost converter that converts whatever voltage the AC line is at any instant throughout the cycle to the 340-ish volts DC across the main reservoir cap.

[BrandonGlatz]

Oh how sneaky, I did not recognize that as a boost converter due to the odd form of the inductor...

So I am assuming that the secondary is just shielding, right?



So I was correct in thinking that this part of the circuit is faulty.  On the output of that it is only outputting 160V. Nowhere near what it should be (it says on the board that the output should be 300+.


I know how boost converters work, but only basically. I know it generates a magnetic field in the inductor, then switches the output so that it is mains+collapsing magnetic field output.


What could cause it to output lower than it should. Could a bad fet cause it?

[mikeselectricstuff]

So I was correct in thinking that this part of the circuit is faulty.  On the output of that it is only outputting 160V. Nowhere near what it should be (it says on the board that the output should be 300+.

What could cause it to output lower than it should. Could a bad fet cause it?

Assuming  you are running from a 110v mains supply, the 160VDC will be just the voltage via the diode  (not shown on your schem but it will be there), so the boost cct is not running at all. First thing to look for is the startup circuit of  the PFC controller - there will probably be a high value (a few hundred K) resistor from the rectified mains to the control circuit, and a low voltage electrolytic cap across the control cct power rail - replace this electrolytic (don't bother testing, just replace it), and check the value of the resistor is correct as they can fail high.

[BrandonGlatz]

The only cap I can find across the voltage lines is the one circled in red here:

It is a small blue one like the one next to the connector above the large cap. Not sure on the value or anything, it just says
"6H564K:
2L     J"

Not sure how to read that (if it even means anything)


As to the resistor, I dont really see any that could be what you are talking about...

Here is a schematic as best as I can get it. It is a really odd circuit and was hard to fit in there.

[BrandonGlatz]

hmm looking at it some more. I just noticed a small trace going off to the side. It comes from between the two 100(ohm,K?) resistors on the gates. It runs over to a small 8 pin chip.

This chip is an IR 1150IS

This is what I believe to be the datasheet for it: http://pdf1.alldatasheet.net/datasheet-pdf/view/143395/IRF/IR1150ISTR.html


It goes through a resistor with "200" written on it and a capacitor into pin 8 on it (gate on the datasheet)

So it looks like this chip is responsible for triggering the gate of the FETs below the inductor. I wonder if it could have failed. I dont have an oscilloscope so is there another way to test it?

[TerraHertz]

Brandon, just to clear up a misconception that caused you to get a bit confused with that 'transformer', any conductive path that is entirely outside the ferrite magnetic loop, is not a transformer winding, it's a shield.

The copper foils wrapped around that component act as shorted turns, absorbing stray magnetic (and electrostatic) fields that might otherwise radiate from the coil. You'll notice they are connected together, via one pin.

I'd expect that pin on the PCB to actually be connected to ground. Or at least the local 'zero volts' reference.

Also on your schematic, the FETs symbols are surely wrong. They show N-channel FETs, and with the g,s,d pins mislabeled. They are much more likely to be MOSFETs.  Also one is more likely to be a single or dual diode.

Tracing circuits to derive a schematic, starting with an unknown PCB, is a very iterative process. It's better to do it with pencil and paper, and don't be afraid to redraw sections as you start to understand circuit function.
It also helps to use a fine-tipped felt pen to mark traces and component nodes on the PCB as 'done', once you've included them in the schematic. Also mark the components with numbers you assign, if the PCB overlay has no component designators. Saves a lot of confusion.

Another tip: Have a strong light that you can shine through the PCB. That helps register the side you're looking at with whatever is on the other side.

Alternatively, what I do to reverse engineer PCBs that are a bit complex, is take high-res photos of both sides of the board, and have them on-screen in photoshop while I trace the circuit. Combine them as semi-transparent layers, so you can see both sides of the board overlaid together.

That way I can add component designators and 'done' colour dots and lines to 'notes' layers on the images. Still draw the schematics by hand though, since sections will need to be redrawn over and over as functionality emerges. And by 'redraw', I mean start on a fresh sheet of paper, so you still have the earlier messy circuit to refer to.

Personally I don't see any reason to suffer the irritations of schematic editors unless one intends to produce a new PCB layout via the schematic -> netlist -> layout -> verify netlist cycle.

[BrandonGlatz]

Brandon, just to clear up a misconception that caused you to get a bit confused with that 'transformer', any conductive path that is entirely outside the ferrite magnetic loop, is not a transformer winding, it's a shield.

The copper foils wrapped around that component act as shorted turns, absorbing stray magnetic (and electrostatic) fields that might otherwise radiate from the coil. You'll notice they are connected together, via one pin.

Yep, I know that now from one of the earlier posts.

I'd expect that pin on the PCB to actually be connected to ground. Or at least the local 'zero volts' reference.

As far as I can tell, it does.

Also on your schematic, the FETs symbols are surely wrong. They show N-channel FETs, and with the g,s,d pins mislabeled. They are much more likely to be MOSFETs.  Also one is more likely to be a single or dual diode.

I have a very hard time remembering small details and names such as that. Please forgive me for the mislabeling. You have no idea how frustrating it is to me.

Tracing circuits to derive a schematic, starting with an unknown PCB, is a very iterative process. It's better to do it with pencil and paper, and don't be afraid to redraw sections as you start to understand circuit function.
It also helps to use a fine-tipped felt pen to mark traces and component nodes on the PCB as 'done', once you've included them in the schematic. Also mark the components with numbers you assign, if the PCB overlay has no component designators. Saves a lot of confusion.

I have absolutely terrible handwriting and would have a really hard time reading what I wrote. Also I would end up redrawing it endlessly as I move stuff around. It is easiest for me to have two or three schematic editors open and use those because I can move things around without having to redo the whole thing.

Another tip: Have a strong light that you can shine through the PCB. That helps register the side you're looking at with whatever is on the other side.

This one is too thick to do that with, but I can very easily follow a trace as it goes between both sides by rotating it about that point.

Alternatively, what I do to reverse engineer PCBs that are a bit complex, is take high-res photos of both sides of the board, and have them on-screen in photoshop while I trace the circuit. Combine them as semi-transparent layers, so you can see both sides of the board overlaid together.

That way I can add component designators and 'done' colour dots and lines to 'notes' layers on the images. Still draw the schematics by hand though, since sections will need to be redrawn over and over as functionality emerges. And by 'redraw', I mean start on a fresh sheet of paper, so you still have the earlier messy circuit to refer to.

Good idea, I might do that in the future!

Personally I don't see any reason to suffer the irritations of schematic editors unless one intends to produce a new PCB layout via the schematic -> netlist -> layout -> verify netlist cycle.

It is much easier for me to do it that way than to write it down.

[TerraHertz]

Also on your schematic, the FETs symbols are surely wrong. They show N-channel FETs, and with the g,s,d pins mislabeled. They are much more likely to be MOSFETs.  Also one is more likely to be a single or dual diode.

I have a very hard time remembering small details and names such as that. Please forgive me for the mislabeling. You have no idea how frustrating it is to me.

I can imagine. Have you considered making up some reminder cards, with the symbols, labels and functional descriptions of components you find hard to remember? I do this sometimes, such as for pinouts. Or use pages from catalogs and data sheets.

I have absolutely terrible handwriting and would have a really hard time reading what I wrote. Also I would end up redrawing it endlessly as I move stuff around. It is easiest for me to have two or three schematic editors open and use those because I can move things around without having to redo the whole thing.

If you have a schematic editor that can drag around, rotate, etc whole circuit sections while retaining all interconnections, please tell me what it's called.
The handwriting - that's really unfortunate. Is it not possible to improve that? Even for printed capitals?

Another tip: Have a strong light that you can shine through the PCB. That helps register the side you're looking at with whatever is on the other side.

This one is too thick to do that with, but I can very easily follow a trace as it goes between both sides by rotating it about that point.

Looking at your photo of the board, I bet it isn't too thick, you just didn't try holding it up to a strong light yet. It's standard FR4, which is quite translucent. And it has no internal copper planes. Try it and see for yourself.

[BrandonGlatz]

I can imagine. Have you considered making up some reminder cards, with the symbols, labels and functional descriptions of components you find hard to remember? I do this sometimes, such as for pinouts. Or use pages from catalogs and data sheets.

Hmm no but that is a good idea. Although flash cards have never helped me with stuff in the past...

If you have a schematic editor that can drag around, rotate, etc whole circuit sections while retaining all
interconnections, please tell me what it's called.

I use a program called expresssch. I design and build the electrical systems for robots and it is very handy for documentation.
http://
www.expresspcb.com/expresspcbhtm/Free_schematic_software.htm

The handwriting - that's really unfortunate. Is it not possible to improve that? Even for printed capitals?

I have been trying for years. It seems to run in my family

Looking at your photo of the board, I bet it isn't too thick, you just didn't try holding it up to a strong light yet. It's standard FR4, which is quite translucent. And it has no internal copper planes. Try it and see for yourself.

I'll have to dig out the halogen then. I'll let you know after work.

[codeboy2k]

Tracing circuits to derive a schematic, starting with an unknown PCB, is a very iterative process. It's better to do it with pencil and paper, and don't be afraid to redraw sections as you start to understand circuit function.

+1 for pencil and paper... it's much faster, just draw ugly circuits and fix it up later.

It also helps to use a fine-tipped felt pen to mark traces and component nodes on the PCB as 'done', once you've included them in the schematic. Also mark the components with numbers you assign, if the PCB overlay has no component designators. Saves a lot of confusion.

Another tip: Have a strong light that you can shine through the PCB. That helps register the side you're looking at with whatever is on the other side.

Alternatively, what I do to reverse engineer PCBs that are a bit complex, is take high-res photos of both sides of the board, and have them on-screen in photoshop while I trace the circuit. Combine them as semi-transparent layers, so you can see both sides of the board overlaid together.

That way I can add component designators and 'done' colour dots and lines to 'notes' layers on the images. Still draw the schematics by hand though, since sections will need to be redrawn over and over as functionality emerges. And by 'redraw', I mean start on a fresh sheet of paper, so you still have the earlier messy circuit to refer to.

Excellent points.  I also use the hi-res photograph technique and view onscreen, and mark it up onscreen.  I don't mark up the original board, ever.
I also apply some color corrections and image processing to the pictures, it makes a much higher contrast between the traces and the overlaying silkscreen and can make everything stand out so well.

[TerraHertz]

Hmm no but that is a good idea. Although flash cards have never helped me with stuff in the past...

Who said anything about 'flash' cards? It's not an exam, just stick them up in view where you work.

www.expresspcb.com - I'll have a look.  I wonder if eevblog has had a thread comparing the different schematic/pcb software available these days?

re shining a light through the pcb:

I'll have to dig out the halogen then. I'll let you know after work.

You're still expecting it to be too hard. But really even a simple LED torch will do. See pics.
Another thing this really helps with, is figuring our where tracks go under DIP ICs. With the board backlit, you can look in under the IC and see quite a lot of the traces on the component side. Without backlighting, can't see sh*t, and typically have to remove the component.
Btw, this works with phenolic PCBs too. (The brown, brittle ones, aka cheap consumer junk.)

Edit:

+1 for pencil and paper... it's much faster, just draw ugly circuits and fix it up later.

Yep! No one is ever going see anything but your final version. So who cares what intermediates look like.
Also I find it's best to use loose sheets of A4 photocopy paper on a clipboard. So I can lay several sheets out like tiles if I need to see them all at once while trying to unify the circuit fragments I drew. Plus it's cheap.

I also use the hi-res photograph technique and view onscreen, and mark it up onscreen.  I don't mark up the original board, ever.

Depends how quick and nasty you want to do it. Also, if you are desoldering some components, and the board needs to work afterwards, you're going to be giving it an alcohol scrub to get the flux off anyway. That removes the marker pen too.

I also apply some color corrections and image processing to the pictures, it makes a much higher contrast between the traces and the overlaying silkscreen and can make everything stand out so well.

Me too. With carefully taken photos, scaling the images to match up, and some colour/contrast enhancement, it's amazing how close the result can look to  something like an original service manual board overlay.

[BrandonGlatz]

Hey everyone, I apologize for the lack of replies. I was too busy with work and school to work on my projects.


I have determined that the issue is the buck converter is not triggering. That is as far as I can figure out. I'm not sure if it is the mosfets or the small 8 pin chip their gates are connected to. I looked them up on digikey in hopes of ordering them to see if that fixed them, but they are out of stock.

Here are the digikey links for them:

http://www.digikey.com/product-search/en/discrete-semiconductor-products/fets-single/1376381?k=k3797
http://www.digikey.com/scripts/dksearch/dksus.dll?vendor=0&keywords=1150is


Where should I go from here? I dont have an Oscope to check it out any further