Finding a replacement for an undocumented transformer

[gkmaia]

I am looking for some help to replace a transformer. I've attached a diagram of the circuit and a photo of it.

Service BOM refers to it as Transformer S1673305. I contacted Leader, they kindly sent me the service manuals but are struggling to do more about the transformer.

ps.: I have asked a similar question on the beginners section and was advised to post a question on the repair section so I could get more specific help in finding a suitable replacement. If that is not ok please let me know.

[Ian.M]

Comes from: https://www.eevblog.com/forum/beginners/flyback-transformer-testing-and-replacement/

[gkmaia]

Ian, do you want to me to remove this post? Is it ok to be here?

[bsfeechannel]

What's wrong with the transformer?

[chris_leyson]

From the picture you posted it looks like the secondary winding has burnt out, that's the winding on the two bobbin sections adjacent to C27. Buzz out the transformer with a meter between pins 7 and 9, it might have gone open, check diode D6 and C27 as well, one of them might have gone short circuit and that may be why the secondary winding may have gone open.

From your picture it looks like that secondary winding has seen its last days, all black and gnarly. If that is the case then you will have to rewind it carefully taking off the burnt out turns and counting them as well as making a note of the winding direction. Good luck with that because it looks like the core is epoxied, those white blobs. If you can get them off then warm everything up enough to get the cores apart without breaking them then you might have a shot at rewinding the secondary. The two cores look a bit longer than normal sized cores so measure them just in case you have to buy replacements. The four section bobbin will probably be unobtanium and the transformer was probably made this way because the maunufacturer had thousands of them on the shelf and it was the cheapest solution.

@TheNewLab, flyback is synonymous with buck-boost topology, old school engineers still call them flyback but it doesn't mean it's literally a flyback transformer out of a TV. Flyback is old school name for a buck-boost converter but if I said buck-boost then who would have a clue ? Well now you know. Flyback=buck-boost and they don't have to be high voltage either.

@gkmala. Did you measure 10VAC across the winding between pins 2 and 4 ? I take it that's pk-pk not RMS, you should should be more clear. If that's the case and you measured 10V pk-pk then that means the oscillator is running OK. R42 and C25 provide positive feedback and D7, C26, ZD1 detect the voltage developed across the winding between pins 5 and 1 and provide negative feedback to regulate the voltage. Also known as a blocking oscillator and they can be a pain if they don't work.

[chris_leyson]

What's wrong with the transformer?

Good question, it might be OK just doesn't look right from the photo. Best to buzz it out with a meter. If the secondary is open then you've got to reverse engineer it and rebuild it, otherwise if the transformer is good then it might be a problem on the secondary side.

[gkmaia]

On the other post we identified the Q10 was defective. I am waiting for a new Q10 to arrive.

Till my signal generator arrives I cant generate a sine to test this transformer.

Best I could do so far was to test for coil continuity and resistance, but without a datasheet and with the knowledge I have I cant really do much more.

I know it is a hard thing to ask without the datasheet and only looking at the circuit. But I would like help to identify the key elements and specs that make this transformer so I can order a replacement in case I need.

Also and way more importantly I want learn from you guys on what to do when needing to replace a transformer like this one in a situation where we do not have a datasheet, we don't know what is the exact output but we have a circuit that gives some clues.

[edpalmer42]

Instead of rebuilding the transformer, why not replace the entire function?  All I see is an isolated 5V to probably 12V converter which is then used as a DC source for IC4 and IC5.  Pull out the transformer and everything to the right of it and replace it with an off-the-shelf little brick.

Ed

[Ian.M]

That would be an easy option except for the need for an enable pin.   However Q9 could be used to switch power to the DC-DC converter if its base drive circuit was modified to remove the feedback and increase the base current so it fully saturates.

Some research and testing needs to be done to establish the DC-DC output voltage required (from the regulators output voltages, and minimum headroom required + a bit of margin) and ouptut current.  I'd probably patch in a floating bench supply and switch it with a 5V relay to mimic the enable function.  However *BEFORE* trying that its essential to confirm that the power domain to the left of T1 on the schematic is not connected to live mains or other dangerously high voltages.

In view of the hassle involved with complete replacement if you cant get a spare, and as it looks like the burnt winding(s) have relatively few turns, I think I'd personally attempt a repair by rewinding the burnt winding(s) in situ on the core.  You need to carefully strip the burnt winding(s) noting direction and number of turns, exact connections and wire gauge, and for convenience the total length of wire.  Its then a matter of cleaning up the bobbin as best as you can and rewinding it with new magnet wire of comparable or slightly larger gauge by threading enough turns through the core apertures half a turn at a time, and carefully laying the new winding in place, if necessary holding it with a wedge cut from a pencil eraser in the aperture you aren't currently threading for the current half turn.   Ideally you need a smooth work surface larger than the length of the wire so you don't have to bend it excessively to make it easier to avoid kinks that would damage the magnet wire's varnish insulation.  Sit on the floor in a large empty room if you have to.

You *MUST* keep track of the number of turns accurately both when unwinding and rewinding.  If you have a manual tally counter, use it!  If not, key [.][5][+] on a basic calculator and keep pressing [=] to record each half turn as you wind/unwind it.   If you take a break, for <DEITY>'s sake write the tally down in case of auto-power-off!

Disassembly to make it easier to rewind is *NOT* a viable option. It looks like the centrer leg of the core has been glued into the bobbin, and if you heat it up enough to soften the epoxy enough to extract the core, the heat will almost certainly damage the bobbin and the un-burnt winding(s).  Uneven heating will probably crack the core and ruin it.  Excess mechanical force and hard tools *WILL* crack the core.  Also if it has been gapped using glass microbeads in the epoxy, it may not have suitable saturation characteristics on reassembly as you'll loose the gap when you clean the old glue off.

[edpalmer42]

I was thinking of the same problem (enable lead) and the same solution.  But that begs the question:  Do you even need the enable lead?  Under what conditions would you turn off power to some of the chips?  What are the implications of just leaving it energized?

The circuit looks quite odd to me.  Two linear regulators for two voltages is fine, but why would you put 10 ohm resistors in series with the outputs?? Kind of kills the idea of 'regulator' doesn't it?

Gkmaia, what is this thing?

Ed

[bsfeechannel]

Best I could do so far was to test for coil continuity and resistance, but without a datasheet and with the knowledge I have I cant really do much more.

If all the windings are OK regarding continuity, another simple test you can do is to identify the primary and connect your meter (set to volts DC) to the other windings. Then you connect that 1.5V battery of the other thread only momentarily, no more than 1 second, to the primary and watch if your meter indicates a momentary voltage different from zero. If that happens, this can be a good indication that your transformer doesn't have a shorted turn.

It is also a good idea if you could post more hi-res pics of your transformer from all angles so we could help you with the visual inspection. Smelling is also a good troubleshooting tool: check it for any smell of burnt stuff. May sound silly, but I see Dave do that in almost every episode of his.

[Ian.M]

With the transformer out of circuit, drive the largest unburnt winding from a signal generator with a 100KHz sinewave (for ferrite core switching transformers. For laminated iron core transformers try 1KHz), monitoring the voltage aross it with a scope, and turn it up as high as you can without distortion or 1V pk-pk whichever comes first.  Scope the other windings and note if there is a significant change if you thread an extra single turn winding through the core round the bobbin and short it.   If not its already got a shorted turn.   

However if the winding is burnt its not fit to use anyway even if it currently tests OK as it could fail shorted at any time.  Heck, I can see bright copper hilights where the insulation has melted and run off the top turn where it goes round the corner of the bobbin.

You can also use the test above to detect shorted turns even if you only have one winding by monitoring the voltage across it as you drive it as long as the signal generator has 50R or 600R output impedance (if its much lower add a series resistor), so if you are going for a rewind its worth checking it after you strip the burnt winding(s) so you can be reasonably certain the remaining winding(s) are OK.

[gkmaia]

This is the power block of the DMM board of my oscilloscope. The DMM board connects to the main board. So the main board feeds power to the DMM thought this circuit.

By the way I am amazed by how many answers I got so far. And all people trying to help. This is a very friendly forum.

[gkmaia]

I just ordered a signal generator. It should arrive in one week time. Also, the missing bits for the motherboard must arrive in one week time.

I will then be able to run a proper test with the signal generator. I will take some more photos today as well.

[edpalmer42]

This is the power block of the DMM board of my oscilloscope.

Brand and model?  That info should always be in the first message.    Don't force me to come over there and beat it out of you! 

The DMM board connects to the main board. So the main board feeds power to the DMM thought this circuit.

So if this circuit was permanently enabled, the DMM would always be turned on?  That might be okay.  Note that this application will require high voltage isolation between the two sides of the transformer.  If you decide to use a DC-DC brick, make sure it has that feature.  If you have to rewind the transformer, make sure you don't compromise that isolation.

By the way I am amazed by how many answers I got so far. And all people trying to help. This is a very friendly forum.

We love puzzles!  Some people like sudoku or crosswords, we prefer resurrecting dead circuits!   

Ed

[ogden]

This is the power block of the DMM board of my oscilloscope. The DMM board connects to the main board. So the main board feeds power to the DMM thought this circuit.

It is unlikely that of 5V->3V isolation transformer is fried. Before you decide that transformer is bad, you shall measure dc resistance of it's windings first. I do not see any need for signal generator here. You shall do  "live" measurements of voltages and waveforms per service manual, especially looking for 10V AC on primary and proper function of DMMP signal. Other option would be just to test DMM power supply alone with 5VDC lab supply

[gkmaia]

Hahaha! 

The service manual is attached.

Page 051, DMM PCB. Top right corner. T1.

DMMV constant +5v DC

DMMP high (5vDC) when DMM is off. Low (0v) when DMM is on.

DMMP could also be a pulse. But as it is my scope I cannot test it. Will need another scope for that.

Q10 was faulty. I am waiting it to arrive from Digikey in a week or so.

The story is. I bought it from a retired gentleman. The main pcb had a pretty bad electrolytic leak and the DMM had a battery leak and he used it for a while like this. The scope still works just fine. But the coils on the main board were humming really loud. So I am replacing all coils, all caps, diodes, regulators on the power block.

The transformer may never be faulty, but myself on the process of testing it may have damaged it. Sadly.

On basic resistance test the coil 2-4 has 3ohms, 1-5 has 5ohms and 7-9 has 10ohms. Also there is not short between the coils. So it may still be just fine. But not a 100% sure.

The BOM refers to it as S1673305. No voltage. Schematics have no voltage as well.

So I am hoping to have some good news when the new component arrive.

[ogden]

You already posted picture with DMM power supply. It's enough.

The transformer may never be faulty, but myself on the process of testing it may have damaged it. Sadly.

Oh, now I see photos in other thread with 5VDC, 0.9A(!) DC(!) supply and handheld DMM in DC(!) voltage mode. Ouch! Loads of facepalms. Now you know, right?

On basic resistance test the coil 2-4 has 3ohms, 1-5 has 5ohms and 7-9 has 10ohms. Also there is not short between the coils. So it may still be just fine. But not a 100% sure.

It could be still fine. Did it smell and "caramelize" ?

I agree that any small signal transistor is fine in place of Q10. You could try to test that power supply w/o Q10 which will mean "always enabled".

[chris_leyson]

Hi Gkmaia, forgot to mention C26 is the wrong way around in the schematic, the voltage at the junction of D7 and ZD1 should be negative as that together with ZD1 does the voltage regulation by sensing the primary side flyback voltage which is also proportional to the secondary side flyback voltage because the transformer windings are coupled. Check to see which way around C26 is really wired on the PCB, it's unlikely that they carried the mistake on the schematic through to production as they would have picked this up in the development stages.

[gkmaia]

Check to see which way around C26 is really wired on the PCB, it's unlikely that they carried the mistake on the schematic through to production as they would have picked this up in the development stages.

Yes, it is wired on the PCB as per the schematic. C24 & 26 sharing the ground.

[bsfeechannel]

And this may well be the cause of the failure of Q10. C24 is part of the AGC (Automatic Gain Control) loop of the oscillator. Installed the wrong way, it may have forced Q10 to work outside its ideal operating point and caused premature failure. Just a guess.

[chris_leyson]

Hmm, that's strange. The voltage on the base of Q9 will probably be a little under +1V when Q9 is on and maybe -1V or -2V when the oscillator switches Q9 off. The way the voltage regulation works is as follows, the feedback winding and D7 generate a negative voltage across C26 and when the voltage across C26 reaches ZD1 Zener voltage then ZD1 starts to conduct and recuces the drive Q9. According to Renesas datasheet Zener voltage for HZ4BZLL is 3.9V nominal. I think C26 is the wrong way around and maybe that's why it failed.;

[gkmaia]

Will that may have caused one specific leg of Q10 to fail?

And this may well be the cause of the failure of Q10. C24 is part of the AGC (Automatic Gain Control) loop of the oscillator. Installed the wrong way, it may have forced Q10 to work outside its ideal operating point and caused premature failure. Just a guess.

The one that failed was B-C which flows from R41 which flows from DMMV.

Q10 base to emitter was fine.

Would it be more likely a surge DMMV considering the supply block of the MB was affected by an electrolytic leak?

Q10 should take 50v/150ma. Q9 50v/5a.

If not from the capacitor you mention. From where else and what could have caused the failure of Q10?

[james_s]

It's not impossible, but small transformers like this very rarely fail. It would be one of the last parts I'd suspect on that board.

[chris_leyson]

I could be wrong about the Zener, it might be there for reverse Vbe protection and not for regulation. In which case there would be a small positive voltage across C26 acting as a bias supply, let's say 1V for sake of argument and D7 acts as a regulator by reducing the bias voltage slightly. C26 is most probably the right way around and the Zenner has nothing to do with voltage regulation, it's just there for reverse Vbe protection. Sorry, brain fart