SMPS for tube pedals

[printer2]

Finally got back to it. Frequency 72.5 kHz, 300V 79 mA. Flattened out a couple of pennies (make sure they are copper kiddies) and did some fancy soldering as well as painted them and the original heatsink flat black, still bare copper in the picture.

I did not get a lot of time on it but the heatsinks felt a little warm to the touch.

Very cool.

I'll try to match something with your numbers.   I'm assuming oscilloscope waveform is the output ripple at the first cap?   (As it sure looks like the diode current passing through the output cap ESR.)

It would be good to get the PWM output from the chip.  It's not possible to extract the MOSFET on-time from the diode current waveform.  The diode current waveform is when the switch is off but it is shorter than the total off time.   So we don't know the on-time or the total off-time.

Almost forgot I found this,
https://dalmura.com.au/static/YH11068A.pdf

It quotes 75kHz operating frequency, which is in agreement with you measurement.    I need to look at the datasheet for that chip.  I was assuming it's constant off time and variable off time, which is how the MC34063 works, but it could be different.

Nonetheless I might be able to narrow the times based on the 8V to 32V input and 40V to 390V output.   Your  72.5 kHz, 300V 79 mA (and 12V in?) would then pin things down in the middle of range.

I did not vary the input voltage (12V) or the load on the module. I did step it up from 100V in 50V steps. At 100V it had a time of 59 kHz and from 150V to 300V it was 72 kHz. I need to make a proper jig so I can clip in my probe, as it was I was only able to clip onto the copper heatsink on the output cap side (only one cap, the other is on the input).

I don't know when I can get back to this, I kind of lost three weeks of my life due to health issues and have a big backload of stuff that needs to be done. But sometimes (heck, most of the time) I get fixated on something and waste time on something rather than more important stuff so I may get to it in a timely matter yet.

[Rob Strand]

I did not vary the input voltage (12V) or the load on the module. I did step it up from 100V in 50V steps. At 100V it had a time of 59 kHz and from 150V to 300V it was 72 kHz. I need to make a proper jig so I can clip in my probe, as it was I was only able to clip onto the copper heatsink on the output cap side (only one cap, the other is on the input).

That might be enough to fill in the gaps.   The key thing being the variable frequency.   I'm assuming a resistive load of 300/79mA = 3.8k, so at 100V the load is 100/3.9k = 26mA.

And yes,   it's often easy to do some quick poking around but there comes a point where you have to set-up everything like a scientific experiment - and that wasn't the idea!

I don't know when I can get back to this, I kind of lost three weeks of my life due to health issues and have a big backload of stuff that needs to be done. But sometimes (heck, most of the time) I get fixated on something and waste time on something rather than more important stuff so I may get to it in a timely matter yet.

I hear you there ...   I've be out for 9 months (and no, I wasn't pregnant ).

[printer2]

At 300V as before.

[Rob Strand]

OK cool, I'll give it another shot.

[printer2]

OK cool, I'll give it another shot.

C1 might be 3 nF, R6 1k. I did a rough calibration of my meter's capacitance when it went out but I am ordering some 1% caps. I could not find my camera tripods so I was poking around with my right hand, had my near sight glasses on with a pair of cheaters over them (eyes getting too old for surface mount), SLR in my left hand trying to keep in focus and push the darn button. It was an interesting evening.

[Rob Strand]

C1 might be 3 nF, R6 1k. I did a rough calibration of my meter's capacitance when it went out but I am ordering some 1% caps. I could not find my camera tripods so I was poking around with my right hand, had my near sight glasses on with a pair of cheaters over them (eyes getting too old for surface mount), SLR in my left hand trying to keep in focus and push the darn button. It was an interesting evening

Is that the clock resistor and cap?  There does seem to be a problem on with the clock capacitance.  I'm seeing C8 on the schematic you posted earlier (no idea if the PCB has designators at all, or that the schematic matches them.)

https://www.diystompboxes.com/smfforum/index.php?topic=130852.msg1284850#msg1284850

For my rough simulation I stripped everything back.   I force the switching frequencies to match what you measured.  I don't even have the chip in the simulation.   As drawn, the schematic has no voltage clamps or snubbers so I didn't add any.   I do get quite a bit of ringing at some points in the circuit.

[printer2]

C1 might be 3 nF, R6 1k. I did a rough calibration of my meter's capacitance when it went out but I am ordering some 1% caps. I could not find my camera tripods so I was poking around with my right hand, had my near sight glasses on with a pair of cheaters over them (eyes getting too old for surface mount), SLR in my left hand trying to keep in focus and push the darn button. It was an interesting evening

Is that the clock resistor and cap?  There does seem to be a problem on with the clock capacitance.  I'm seeing C8 on the schematic you posted earlier (no idea if the PCB has designators at all, or that the schematic matches them.)

https://www.diystompboxes.com/smfforum/index.php?topic=130852.msg1284850#msg1284850

For my rough simulation I stripped everything back.   I force the switching frequencies to match what you measured.  I don't even have the chip in the simulation.   As drawn, the schematic has no voltage clamps or snubbers so I didn't add any.   I do get quite a bit of ringing at some points in the circuit.

Oops, wrong schematic of the same board. R25, C10. I got the 1k but might as well ignore the 3 nF value I got(schematic value 0.1 uF). I just spun a trimmer so the display on my meter matched the reading on my meter but with only one point of calibration I would not bet any money on it. Some 1% caps on order to do it properly.



On the ringing, I grabbed this picture earlier on yesterday but if you ask me where I could not be sure now. I think it might have been at 100V out without a load. Thought it might be due to no load, my spaghetti wiring, who knows so I forgot about it.



Since I got those pictures I decided to live dangerously and bumped up the voltage to 390V and 73 mA, 29W. Before I had a chance to scope anything, 'pop'. I felt the copper fins after and they did not seem overly warm. Puts me out of commission until I get my order in. At least now I can clean up the corner of my desk

[printer2]

OK, just a small update. Got my parts in, decided to go with a 6A fast diode that should be able to pass more current before heating up too much. The diodes I picked are just big enough to straddle the copper pads on the board and I thought end of story. But after getting them I saw they are not as tall as the original part and soldering on a copper heatsink to each side is problematic. I thought these could be a low cost somewhat easy way for a beginner to get into high voltage tube amps but I am starting to think maybe not. The jury is not out yet, still going to try blowing up more parts. The original part is on the right.



But I think with my little penny trick I should be able to feed a SE 6V6 without it blowing. On the bench is one thing, in an amp another, and we are not quite there yet. I made a little aluminum enclosure around the module so it could be outside the chassis and have a supply of fresh air.



Here is the original on the right, the 6A surface mount one with an attempt to solder on the copper pennies after some reworking, the left a 15 A part but the pads dictate how it is soldered. Which sucks as the bolt to mount a heatsink is right over the transformer pins. Tried to figure out how to cool it and in the end I decided to CA glue a piece of aluminum on it.





Painted the bare heatsink then decided to test it. Ran it at 350V and 80 mA for 28W. I stuck a thermister on the original heatsink and it got to 50 C with the rate of increase not really changing too much so I did not think it would stabilize to soon. The aluminum plate on the diode was also approximately the same temperature as the other one. Since I only have two of these diodes I decided to stop it and put a larger heatsink on the mosfet. Cranked it up again and the temperatures stabilized at 46 C.



It seemed happy enough now with the room temperature of 25 C. I then took the temperature of the transformer. The displays only go up to 70 C and the transformer was above that. This has me thinking that might be why the two heatsinks were roughly the same temperature. I planning on mounting the module upside down, already did above, might not be a good idea with the transformer heat rising. I am going to try the module with the stock part and my pennies for a 6V6 (actually 12V6) in single ended as a small Champ. If it works I may just leave it but for a higher power amp I will be going to the BYC15-1200PQ (TO220).

[Rob Strand]

I decided to stop it and put a larger heatsink on the mosfet. Cranked it up again and the temperatures stabilized at 46 C.

A bigger heatsink will be far more effective than painting etc.

It seemed happy enough now with the room temperature of 25 C. I then took the temperature of the transformer. The displays only go up to 70 C and the transformer was above that. This has me thinking that might be why the two heatsinks were roughly the same temperature. I planning on mounting the module upside down, already did above, might not be a good idea with the transformer heat rising.

Yes the hot transformer doesn't help.   To me having the diode under the transformer knee-caps any attempt to remove heat from both.   I like the TO-220 version but there some issues.  The region between the PCB and the aluminum plate is going to be very ineffective for removing heat from the transformer or the diode.   One side of the plate is essentially unused halving doubling the thermal resistance.  Pushing the diode down onto the PCB doesn't really help either.  If you use a finned TO-220 heatsink it will maintain or improve the area but it will also have a smaller foot print and allow for better cooling of the TO-220 and the bottom of the PCB. Even one with shallow fins it could help things along.    On idea is to mount the TO-220 diode to the main heatsink.  Another is to cut some tracks and move it to the another edge of the PCB away from the Tx but then you will need another heatsink and have to mount the heatsink somehow.   A kind of in between version would have the heatink going upto the edge/corner of the PCB where there is no Tx, then leave the area under the Tx open.  That would mean putting the TO-220 on the edge if the heatsink which would be OK provide the heatsink material is thick enough.   A larger gap under and around the PCB also helps.

My apologies for not getting back to the simulation.  It's on my list of things to do.   I've been trying to finish some stuff off.

[printer2]

I decided to stop it and put a larger heatsink on the mosfet. Cranked it up again and the temperatures stabilized at 46 C.

A bigger heatsink will be far more effective than painting etc.

It seemed happy enough now with the room temperature of 25 C. I then took the temperature of the transformer. The displays only go up to 70 C and the transformer was above that. This has me thinking that might be why the two heatsinks were roughly the same temperature. I planning on mounting the module upside down, already did above, might not be a good idea with the transformer heat rising.

Yes the hot transformer doesn't help.   To me having the diode under the transformer knee-caps any attempt to remove heat from both.   I like the TO-220 version but there some issues.  The region between the PCB and the aluminum plate is going to be very ineffective for removing heat from the transformer or the diode.   One side of the plate is essentially unused halving the thermal resistance.  Pushing the diode down onto the PCB doesn't really help either.  If you use a finned TO-220 heatsink it will maintain or improve the area but it will also have a smaller foot print and allow for better cooling of the TO-220 and the bottom of the PCB. Even one with shallow fins it could help things along.    On idea is to mount the TO-220 diode to the main heatsink.  Another is to cut some tracks and move it to the another edge of the PCB away from the Tx but then you will need another heatsink and have to mount the heatsink somehow.   A kind of in between version would have the heatink going upto the edge/corner of the PCB where there is no Tx, then leave the area under the Tx open.  That would mean putting the TO-220 on the edge if the heatsink which would be OK provide the heatsink material is thick enough.   A larger gap under and around the PCB also helps.

My apologies for not getting back to the simulation.  It's on my list of things to do.   I've been trying to finish some stuff off.

No worries about the sim, think any fast diode will work if they remain cool. Until I checked the transformer temperature we thought it was the diode heating itself to be the problem. Or it could have been and now the power is high enough for the transformer to be the next issue. Earlier when I was thinking of the diodes I wondered if using two and the winding to produce the negative voltage (for the ones that give a +/- supply) to halve the power going through a single diode. Would flip the second diode and capacitor around and combine the two through a pair of resistors to hopefully balance out the current through the two.



Now wondering if using the second winding would decrease the transformer losses. Just depends if the heat is more from copper losses or from the core. Let's see, P = I² x R, drop R in half and you dissipate less power heating up the place. Looking at the schematic, seems like there might be another winding for the primary. Another place to reduce copper loss? Would they be significant?

This sure is becoming more involved than I originally anticipated. Yes the T0-220 can be mounted on a common heatsink but the existing mosfet would have to be replaced with one having longer leads to extend over the edge of the circuit board. Or use the replaced heatsink in the same place as it was but under the board and bolt the diode to it. If I could find bolts long enough and the same diameter as the ones used to hold the original heatsink to the board (some modules do not even have the heatsink mounted and just hanging off the mosfet) I could sandwich the two heatsinks together.

Oh wait, I am looking at a board with the original heatsink on it and the one I pulled. Could cut another section of the cpu heatsink I used for the bigger replacement heatsink and do the same thing. It would be convenient using the original one though, already has the hole tapped and ready to mount the TO-220 diode. Just depends if the original one is up to the task. Was the bigger heatsink just helping to get rid of the heat allowing me to draw more current? I know a normal person would just put a fan on a stock module and be done with it. I dislike fans as they can fail, replace them on speed drives at work from time to time.

edit: just a quick note, flipped the diode and capacitor around on a dual voltage out board. Weird response, almost half the voltages as compared to when it is configured as a negative supply..

[Rob Strand]

Now wondering if using the second winding would decrease the transformer losses. Just depends if the heat is more from copper losses or from the core. Let's see, P = I² x R, drop R in half and you dissipate less power heating up the place. Looking at the schematic, seems like there might be another winding for the primary. Another place to reduce copper loss? Would they be significant?

Yes it's a good idea.  It could make a small improvement.  Not sure if it will work due to technicalities - see below.  Windings in switching supplies are tricky because of the proximity effect and losses due to the gap in the core.  However in this case the unused winding is already present.  Unsed copper is bad, so you might as well put it to use.

This sure is becoming more involved than I originally anticipated. Yes the T0-220 can be mounted on a common heatsink but the existing mosfet would have to be replaced with one having longer leads to extend over the edge of the circuit board. Or use the replaced heatsink in the same place as it was but under the board and bolt the diode to it. If I could find bolts long enough and the same diameter as the ones used to hold the original heatsink to the board (some modules do not even have the heatsink mounted and just hanging off the mosfet) I could sandwich the two heatsinks together.

I'm not really surprised.  The unit has one or more minor design faults which come from design decisions early on - basically there is no margin at maximum load.  It's always hard to patch fix these types of problems from the outside other than operate at outputs less than rated load  You can do *something* and you can try your best but the final result might not be what you would end-up with if you could start from scratch.

Was the bigger heatsink just helping to get rid of the heat allowing me to draw more current? I know a normal person would just put a fan on a stock module and be done with it. I dislike fans as they can fail, replace them on speed drives at work from time to time.

More than likely.  Fans help a lot as they improve virtually all aspects of heat dissipation, that includes the transformer.  It's just a pain having a fan.

edit: just a quick note, flipped the diode and capacitor around on a dual voltage out board. Weird response, almost half the voltages as compared to when it is configured as a negative supply..

You can't flip just the diode.  On a flyback converter the output from each winding has to happen on the same part of a switch cycle.  There's no option for a push-pull type arrangement.  For the negative winding, you have to flip the winding and make sure the diode is around the right way.



[edit: for the second case you could leave the diode with it's original direction in the negative lead.  That means you can use the negative diode mounting.


]

The biggest unknown is if the winding on the negative supply matches that on the positive winding.  If not it might not actually contribute any current.  If the negative winding was multifilar wound with the positive winding then you can parallel the winding.  If you use an extra diode it might help with some mismatch, not much.    A third option is to create a completely separate positive rail then combine the outputs with resistors.

Yet a fourth option, which requires *much* more care, would be to put the windings in series.  That means you can utilize the second winding without worrying about the windings sharing current.  However it's significant change and could fry something.

Another unknown is the pin numbers on the transformer that's going to require some checking/measurements.  You can go off that schematic but is it correct and is 5,6 a pair or 5,7 the pair?  Then for the negative winding why do we see pins 8,9 in parallel?

The loss in the transformer is due to the copper loss and also the loss in the ferrite.  Usually both are equally significant.  You can can reduce ferrite losses by reducing the switch frequency (actually increase the on time).    However for the same power output if you reduce the switch frequency you need more winding current and that will increase the copper loss.  Similarly you you can increase the switch frequency to trade the other way.   We don't know which one is worse for this design.  An easy way to do a test is to measure the efficiency.    Then make a small change to the frequency up/down/or both and see which way improves the efficiency.  Then perhaps try a bit more in the same direction.  There's a lot of checks required to be confident this doesn't cause problems.  The idea here is if the supply is more efficient it will generate less heat.  However,  if you want to reduce the loss in a specific part that might result in poorer efficiency - it's like the other components have to suffer more to prop-up a weak one.

[printer2]

edit: just a quick note, flipped the diode and capacitor around on a dual voltage out board. Weird response, almost half the voltages as compared to when it is configured as a negative supply..

You can't flip just the diode.  On a flyback converter the output from each winding has to happen on the same part of a switch cycle.  There's no option for a push-pull type arrangement.  For the negative winding, you have to flip the winding and make sure the diode is around the right way.



[edit: for the second case you could leave the diode with it's original direction in the negative lead.  That means you can use the negative diode mounting.


]

The biggest unknown is if the winding on the negative supply matches that on the positive winding.  If not it might not actually contribute any current.  If the negative winding was multifilar wound with the positive winding then you can parallel the winding.  If you use an extra diode it might help with some mismatch, not much.    A third option is to create a completely separate positive rail then combine the outputs with resistors.

I am not as good as I used to be, realized I had to move the winding around and flip the diode and cap. Then went to do other stuff and came back, flipping the cap and diode with the transformer winding forgotten. I probably wander around more looking for stuff I put down, part of the reason I am trying to get some projects done and shelve the rest and concentrate on health and wellbeing.

I marked up the board to show what needs to be done, not the easiest as the transformer needs to be pulled and the ground plane around pin 5 needs to be cut away to provide enough insulation. Pin 8 and 9 are common, a cut is needed between them, pin 9 removed and pin 8 to pin 7 as a ground point. Pin 5 can go into the pin 9 hole. Then you combine the two capacitor positive terminals together with a pair of resistors going to the amplifier circuit. I think. One other thing, I noticed the single output transformers have the windings below the surface of the transformer core and the bipolar board has the windings protruding above the surface of the core. Makes sense, why pay for more going into a part than need be if it will not be used. Would it be worth going through the trouble? I will have to mull it over, maybe a winter project, this has already taken more time than I had planned on. Good to know about the shortcomings though.

[printer2]

And I said I would be putting this away today, hah. Popped the transformer off the board.





I took some measurements, not sure how accurate my meter is reading mH. Used a voltage divider to determine the resistances, the output side matches my meter readings on ohms. Just for kicks I fed each side of the secondary with my signal generator with a sine wave and read the voltages off my scope. from 70 kHz to 150 kHz the step up ratio roughly 16:1

[Rob Strand]

I took some measurements, not sure how accurate my meter is reading mH. Used a voltage divider to determine the resistances, the output side matches my meter readings on ohms. Just for kicks I fed each side of the secondary with my signal generator with a sine wave and read the voltages off my scope. from 70 kHz to 150 kHz the step up ratio roughly 16:1

Very cool.   The 2.1mH/2.2mH don't look off the mark.  In my previous simplified simulation I think the current started to dropped off at 500uH.

The 2.75R vs 2.17R secondary resistances means the windings aren't multifilar wound.  The 2.75R is wound over the 2.15R.   The 2.24mH inductance vs 2.12mH inductance are just annoyingly different that it's unclear if the voltages  (turns) on the two secondary winding are the same.

The 0.11mH primary might be 0.011mh = 11uH.   I had 17uH to ensure enough energy was built up in the core for a 48W output.
If we take your voltage ratio of 16, then 2.1mH should translate to 2.1mH/16^2 = 8.2uH.

To me the transformer core looks like an EFD25 or EFD30 (tell-tale sign is the winding doesn't extend past the the outer ferrite wall).   I suppose you could measure the dimensions to confirm.

If you look at p15 or, IR AN1024
https://www.infineon.com/dgdl/an-1024.pdf?fileId=5546d462533600a401535591115e0f6d
Table of ratings
EFD25   20 to 30W
EFD30   30 to 50W

However, one of TDK's docs gives an EFD25 a 55W flyback rating at 100kHz,
https://www.tdk-electronics.tdk.com/inf/85/ds/b82802a.pdf

The devil is in the details on this stuff which might explain the higher TDK rating.  Nonetheless t looks like the EFD25/EFD30 size is in the ball-park of the 48W rating.

I haven't tried to match up the winding resistances and inductances.  It's a bit tricky as we don't know the division of the winding area between the primary and secondary.   It looks like the primary is a single wire but not a *lot* thicker than the secondary.  Also the size of the core EFD25 vs EFD30 is going to add to the uncertainties.

[R.G.]

I'm guessing that the inductances you measured were the open-circuit inductances of the windings. If so, did you happen to measure the leakage inductances? I usually found that the energy stored in leakage inductance was as troublesome as switching losses.

[printer2]

I took some measurements, not sure how accurate my meter is reading mH. Used a voltage divider to determine the resistances, the output side matches my meter readings on ohms. Just for kicks I fed each side of the secondary with my signal generator with a sine wave and read the voltages off my scope. from 70 kHz to 150 kHz the step up ratio roughly 16:1

Very cool.  The 2.1mH/2.2mH don't look off the mark.  In my previous simplified simulation I think the current started to dropped off at 500uH.

The 2.75R vs 2.17R secondary resistances means the windings aren't multifilar wound.  The 2.75R is wound over the 2.15R.  The 2.24mH inductance vs 2.12mH inductance are just annoyingly different that it's unclear if the voltages  (turns) on the two secondary winding are the same.

The 0.11mH primary might be 0.011mh = 11uH.  I had 17uH to ensure enough energy was built up in the core for a 48W output.
If we take your voltage ratio of 16, then 2.1mH should translate to 2.1mH/16^2 = 8.2uH.

I just checked again, the meter gave me two different readings on the primary. On the 20 mH setting it reads 0.11 mH and on the 2 mH setting 0.012 mH, on the 200 mH setting it was reading 0.2 mH so I thought the 0.11 mH reading would be the better one to report. Also the primary has two windings in parallel.

To me the transformer core looks like an EFD25 or EFD30 (tell-tale sign is the winding doesn't extend past the the outer ferrite wall).  I suppose you could measure the dimensions to confirm.

25.5 x 25.0 x 9mm, like these easy questions.

If you look at p15 or, IR AN1024
https://www.infineon.com/dgdl/an-1024.pdf?fileId=5546d462533600a401535591115e0f6d
Table of ratings
EFD25    20 to 30W
EFD30    30 to 50W

However, one of TDK's docs gives an EFD25 a 55W flyback rating at 100kHz,
https://www.tdk-electronics.tdk.com/inf/85/ds/b82802a.pdf

The devil is in the details on this stuff which might explain the higher TDK rating.  Nonetheless t looks like the EFD25/EFD30 size is in the ball-park of the 48W rating.

I haven't tried to match up the winding resistances and inductances.  It's a bit tricky as we don't know the division of the winding area between the primary and secondary.  It looks like the primary is a single wire but not a *lot* thicker than the secondary.  Also the size of the core EFD25 vs EFD30 is going to add to the uncertainties.

[printer2]

I'm guessing that the inductances you measured were the open-circuit inductances of the windings. If so, did you happen to measure the leakage inductances? I usually found that the energy stored in leakage inductance was as troublesome as switching losses.

Do not know how I missed your post, I'll see about it tomorrow.

One step forward, two step back kind of day. I rewired the winding so they are in parallel. Twice, second time the charm, you know the above board and below board being backwards? Not the first time, I drilled the chassis holes for mounting the other module on the wrong side, going to look more Marshall than Fender. But I got it sorted out and did a quick check and it is making the proper voltage, tomorrow I will put it under load. Oh shoot, I soldered in the transformer already? How about a single supply transformer for the leakage inductance?

[Rob Strand]

I just checked again, the meter gave me two different readings on the primary. On the 20 mH setting it reads 0.11 mH and on the 2 mH setting 0.012 mH, on the 200 mH setting it was reading 0.2 mH so I thought the 0.11 mH reading would be the better one to report. Also the primary has two windings in parallel.

The meter is having some issues measuring small inductances.  I'd probably trust the reading on the lowest range (2mH) of 0.012mH.   Once you start measuring value below 1/10 full scale you expect some loss in accuracy but at 1/100 full scale it's unlikely many of the digits are significant.  2mH/100 = 0.02mH so even the 2mH range is pushing your luck a bit.

25.5 x 25.0 x 9mm, like these easy questions.

Ok cool thanks.  I'll work off that.

I did some rough checks.    I did simulation with full power and 390V output running at 12V and at the minimum of 10V.   At 10V the inductor current was peaking at 12A or so, and the timing with an 8.2uH primary only just makes it.

The primary inductance needs to be around 8.2uH in order to get enough power going through at full load.   I'm trusting the estimate of the primary inductance based on the secondary inductances you measured and the 1:16 voltage ratio, which is around 8.6uH.

Then based on the core being EFD25 gapped core and with an assumed AL factor of 160nH/t^2, the largest standard gap, I juggled the inductance measurements and estimated that at a maximum flux density of 300mT (for ferrite) the inductor would handle 15A.   In short all those numbers look pretty consistent.   The 12A peak of the simulation vs 15A maximum current look fine.  The core is probably operating at flux density of 300mT * 12/15 = 240mT.
https://www.tdk-electronics.tdk.com/inf/80/db/fer/efd_25_13_9.pdf

The next step was to match the DC resistances and the turns on the transformer.    When I did this it looks like the transformer isn't wound to fill the winding window.   However when you look at the yellow tape the winding looks pretty full.  I even tried to guess the wire thicknesses from the pic and I got a similar result.   So not I'm not sure what's going on.

Based on the transformer rating tables I posted earlier and we might expect an EFD25 to be a little pushed at 48W and that might explain why the transformer gets hot.

The last thing I tried was to workout how hot we expect the transformer to get.   This requires a lot of work.  I did some simple calculations and from what I can see the transformer shouldn't get that hot.   I need to check over things.   When I saw the 240mT flux density I was expecting it to heat up the core, especially for 75kHz.

Do not know how I missed your post, I'll see about it tomorrow.

It's tricky to measure accurately and it's a small value so your meter might have issues.   I've got some estimates in my simulation.   The TDK document I posted earlier has table with some leakage values for the EFD transformer (although these need to be translated for different turns).

[printer2]

Found a mistake in the wiring of the secondary, fixed the image above. Ran the module at 325V, 73mA, 24W for an hour and the stock heatsink was 43 C, the core was 46 C and the stock diode with my two penny heatsink soldered on it was 32.5 C.  Then I turned the pot to increase the voltage to 250V and the voltage dropped to 314V. Under no load it would get up to 390V, not under load. Decided to let it cool down and try it, still no luck. I did check the voltage of the 12V supply then, down to 11.8V, never checked it before with the other module which is a shame. Not sure what the fault could be, it seems to work fine at 325V.

I measured the transformer after I did the fix with the coil going to the correct pin and got 0.006 mH on the primary, leakage 0.001 uH (the lowest reading on the meter, I was unsure of the reading on this scale before in case the zero value was out). On the secondary I got 2.6 mH and with the primary shorted, 1.95 mH. Just for giggles I turned it up to 360V and it sat there for a moment and then dropped down to 314V. This is the only dual voltage board I have might have to order another.

Running it at 275V now, getting a lot hotter now, 60 C or more. I'll look at it again tomorrow.

[printer2]

I will have to check at other supply voltages but with this one at 11.7V full load the module sags. I tried three different ones, two single and the dual voltage one I modified and they all sagged with a higher load. Just eyeballing two meters I get 28-30W as it starts to droop and it goes down to 20-23W. I did the tests with them cold and ran them only as long as I needed to to get my readings so the heating is taken out of the equation.

Just did a quick test with a 16.2V supply. Set for 390V it did not sag with 104 mA out, 41W. Cool. Next is to test the heating but that will take some time and I am behind with regular life right now.