I recently decided to try to upcycle a microwave oven transformer into an output transformer with roughly a 2500:24 ohm impedance ratio. I chose a turns ratio of 1500:150 merely because those are round numbers. I didn't want to make this any more complicated than strictly necessary since I've never tried it before. It took a fair amount of time and patience, but I have both. Microwave iron is so readily available that I figure that someone else may have a need to know how this turned out for me. I challenge anyone to try the same instead of sending yet another dead microwave oven to the landfill. I am not claiming that this experimental transformer is "just as good" as any factory made transformer. That is entirely subjective in my opinion. It does sound great to my unqualified ear though.
I put together a test setup consisting of a 12AT7 gain stage coupled to a 6AS7G dual power triode for the output stage, given that it has such low output impedance. I take it that the 6AS7G has much lower "technical" Z than 2500 ohms, but I got far and away the best results with what I have here. The transformer that I compared this one to is an Edcor GSXE10-2.5k - 16ohm since that's the closest match that I have on hand. I intend to try this with smaller iron in the future for a headphone amplifier.
The microwave iron is freakishly huge for that purpose, but it's still in the experimental stage and it was easier to wind, given it's large size. The microwave iron that I have is already in an E-I configuration, and the "E" and "I" were bead welded together. I cut the welds with a grinder and pryed the "I" off, exposing the factory windings. I removed all three of those, along with the shunts. On one piece of iron, I ground the middle leg of the "E" down by a couple of mm to increase the air gap, but didn't notice any improvement over leaving it alone. I 3d-printed a bobbin for the new iron out of Nylon, and put it on a hand-cranked jig to wind it.
On my first attempt, I wound the primary and the secondary both in one large go each, laying the secondary entirely on top of the primary. I used 28awg magnet wire for both, since I have a large amount of it that's been salvaged from shaded pole electric motors, and isolated the windings from each other using Kapton tape. The results of that one were mixed. I wasn't overly careful with getting the turns very neat and even, and the result was a square wave output that had a considerable amount of rolloff. That is, the "corners" of the waveform were rounded on all frequencies. Not sure of "rolloff" is the correct term for that, but there it is.
On the second attempt, I carefully wound four layers of the primary, then one layer of the secondary, then another four of the primary, etc until finished. I then spliced the segments together. I again isolated the primary and secondary at all transitions with Kapton tape. I fed a square wave into the amp at several frequencies and documented the output on a scope. I don't yet know how to calculate THD, so I apologize for not including that data. I observed near-identical performance at low frequencies. The edges of the waveform were somewhat rounded off on the DIY transformer at high frequencies though. Not so much that I found it unpleasant to the ear. I really like it in fact, and am excited that it worked anywhere near as well as it did. The trebel is nice and clear, but it's not as sharp and piercing here as I've seen with some other setups.
So anyway, there it is. The iron on the DIY trafo is a little janky looking, but one would be surprised what a little sandpaper and spray paint can do to remedy that. There's certainly room for improvement overall, but it's workable and I think that's pretty awesome.
I put together a test setup consisting of a 12AT7 gain stage coupled to a 6AS7G dual power triode for the output stage, given that it has such low output impedance. I take it that the 6AS7G has much lower "technical" Z than 2500 ohms, but I got far and away the best results with what I have here. The transformer that I compared this one to is an Edcor GSXE10-2.5k - 16ohm since that's the closest match that I have on hand. I intend to try this with smaller iron in the future for a headphone amplifier.
The microwave iron is freakishly huge for that purpose, but it's still in the experimental stage and it was easier to wind, given it's large size. The microwave iron that I have is already in an E-I configuration, and the "E" and "I" were bead welded together. I cut the welds with a grinder and pryed the "I" off, exposing the factory windings. I removed all three of those, along with the shunts. On one piece of iron, I ground the middle leg of the "E" down by a couple of mm to increase the air gap, but didn't notice any improvement over leaving it alone. I 3d-printed a bobbin for the new iron out of Nylon, and put it on a hand-cranked jig to wind it.
On my first attempt, I wound the primary and the secondary both in one large go each, laying the secondary entirely on top of the primary. I used 28awg magnet wire for both, since I have a large amount of it that's been salvaged from shaded pole electric motors, and isolated the windings from each other using Kapton tape. The results of that one were mixed. I wasn't overly careful with getting the turns very neat and even, and the result was a square wave output that had a considerable amount of rolloff. That is, the "corners" of the waveform were rounded on all frequencies. Not sure of "rolloff" is the correct term for that, but there it is.
On the second attempt, I carefully wound four layers of the primary, then one layer of the secondary, then another four of the primary, etc until finished. I then spliced the segments together. I again isolated the primary and secondary at all transitions with Kapton tape. I fed a square wave into the amp at several frequencies and documented the output on a scope. I don't yet know how to calculate THD, so I apologize for not including that data. I observed near-identical performance at low frequencies. The edges of the waveform were somewhat rounded off on the DIY transformer at high frequencies though. Not so much that I found it unpleasant to the ear. I really like it in fact, and am excited that it worked anywhere near as well as it did. The trebel is nice and clear, but it's not as sharp and piercing here as I've seen with some other setups.
So anyway, there it is. The iron on the DIY trafo is a little janky looking, but one would be surprised what a little sandpaper and spray paint can do to remedy that. There's certainly room for improvement overall, but it's workable and I think that's pretty awesome.
Attachments
Last edited:
It looks like laminations but it looks like they are welded together. Aren't laminations supposed to be separated from each other with an oxide layer?
Last edited:
Hi Jason
I to have been interested in recycling microwave oven transformers. I had trouble accurately separating the "I's" from the "E's" so I made a jig. The idea is to place the MWO transformer on the jig and then with the angle grinder not running adjust the wing nuts such that the cutting disk just kisses the transformer at both ends of the proposed cut, the grinder was started the wing nuts turned < 1/4 turn and a fine cut made, the wing nuts adjusted the same amount again and another fine cut made, this was repeated until the desired depth of cut is obtained.
It was found the weld depth of penetration is a fraction of a mm, except for the ends of the weld where the weld folds around the end and the penetration is much deeper.
My first transformer was cut much too deep before it came apart, the idea is to notch the ends of the weld before attempting to remove the "I's" .
The I's are often welded to a back plate that is easily removed.
Fine cuts ensure no heating of the laminations.
I cut a few transformers apart and have not yet made any transformers or chokes with them.
A dry lubricant is used on the wooden rails for smooth operation.
I to have been interested in recycling microwave oven transformers. I had trouble accurately separating the "I's" from the "E's" so I made a jig. The idea is to place the MWO transformer on the jig and then with the angle grinder not running adjust the wing nuts such that the cutting disk just kisses the transformer at both ends of the proposed cut, the grinder was started the wing nuts turned < 1/4 turn and a fine cut made, the wing nuts adjusted the same amount again and another fine cut made, this was repeated until the desired depth of cut is obtained.
It was found the weld depth of penetration is a fraction of a mm, except for the ends of the weld where the weld folds around the end and the penetration is much deeper.
My first transformer was cut much too deep before it came apart, the idea is to notch the ends of the weld before attempting to remove the "I's" .
The I's are often welded to a back plate that is easily removed.
Fine cuts ensure no heating of the laminations.
I cut a few transformers apart and have not yet made any transformers or chokes with them.
A dry lubricant is used on the wooden rails for smooth operation.
They are laminated and are welded on the outside of the E's, the weld has a very small depth of penetration, if there was a corresponding weld on the inside of the "E" or "I" it would constitute a shorted turn and that would be very bad, my guess is the transformer would be useless. One could model it in a magnetic simulation program maybe something like QuickField. When I get back to this project I will run some BH curves and wind some test transformers; chokes seem to work fine as I tested with a coil recovered from a MWO, very easy to adjust the gap for DC. BTW the stack can easily be halved to make two out of one.
Thanks for all the replies, guys.
The iron on these is laminated, and it is indeed shallowly welded in several places. I only cut the two welds that were holding the "I" on though. I cut all welds on two other transformers so that I could separate all of the laminations though. I intend to try making power transformers with those at some point in the future, and I'll need to interleave the laminations on those. They're also coated in some kind of hard wax/varnish which can be burned off if desired. So they're not technically metal-on-metal when placed against each other, at least with the trafos that I have on hand.
Cutting these with a grinder is indeed tricky. One false move and you'll cut a gash where you don't want it. I have cut several others with a hacksaw though. It's prettier, but it's some decent work to undertake. I like the idea of a jig for the grinder though. I may have to look into that for when I'm ready to make one for real.
I think that splitting the stack to make two is an awesome idea. I'm absolutely going to try that. Doing this has made me really curious how these would handle higher power. Say, 25 watts+. That's when factory OPTs really start getting expensive.
The iron on these is laminated, and it is indeed shallowly welded in several places. I only cut the two welds that were holding the "I" on though. I cut all welds on two other transformers so that I could separate all of the laminations though. I intend to try making power transformers with those at some point in the future, and I'll need to interleave the laminations on those. They're also coated in some kind of hard wax/varnish which can be burned off if desired. So they're not technically metal-on-metal when placed against each other, at least with the trafos that I have on hand.
Cutting these with a grinder is indeed tricky. One false move and you'll cut a gash where you don't want it. I have cut several others with a hacksaw though. It's prettier, but it's some decent work to undertake. I like the idea of a jig for the grinder though. I may have to look into that for when I'm ready to make one for real.
I think that splitting the stack to make two is an awesome idea. I'm absolutely going to try that. Doing this has made me really curious how these would handle higher power. Say, 25 watts+. That's when factory OPTs really start getting expensive.
True. A lot of MOTs have relatively THIN laminations compared to cheap run of the mill Chinese audio power trafos.
MOTs get a bad rep because they typically run the flux density up around one half Tesla where they make all manner of racket and the magnetization current is terribly nonlinear. Half wave rectification used in the application doesn’t help matters. If it wasn’t for the welded construction, they would shake themselves apart. Add more turns per volt on the rewind. You wont get 600 or 800 watts out of it, but who cares if it was free?
I don’t particularly like the split bobbins they typically use. The idea is to increase the isolation between primary and secondary which increases the hell out of leakage reactance. Most of them go so far as to add magnetic shunts in there so that flux is deliberately leaked - you can short the secondary and it doesn’t draw ungodly current. Obviously, you remove those. And I‘d use a standard bobbin, if you can get one to fit.
Those hand held angle grinders are hard to hold in place. Done it, but it is a bitch. At some point I may set up a bench grinder with a similar cutting wheel and try it.
I was able to get 100 turns per layer with 28awg magnet wire, so I broke the secondary into 50-turns-per-layer chunks. It worked out as: 400-50-400-50-400-50-300
@JasonWatkins
4//3 looks like a good starter providing for a reasonably ok coupling between primary-secondary windings translating into music, and the scope pictures didn't look too bad either although starting to round of earlier as seen at 5kHz shot but otoh no ringing it seems (wondering how Edcor's interleaving scheme looks like), looking forward to see and read about the end result with listening impression, good luck with the build.
4//3 looks like a good starter providing for a reasonably ok coupling between primary-secondary windings translating into music, and the scope pictures didn't look too bad either although starting to round of earlier as seen at 5kHz shot but otoh no ringing it seems (wondering how Edcor's interleaving scheme looks like), looking forward to see and read about the end result with listening impression, good luck with the build.
Did anyone manage to sucessfully separate the EI laminations of a MOT and stack them alternately and densely, for example to build a power or PP output transformer?
Best regards!
Best regards!
Two or three months ago I »disassembled« two different MOT's by removing the HV and heater windings with a hacksaw in order to build a spot welding device. The laminations are standard metric EI86 and EI96 resp., stacking height is 66 mm for each tranny.
Best regards!
Best regards!
I have cut all of the welds on a few examples to separate the laminations. The few on the edges of the iron don't like to come off without bending, but the rest separate easily. Just insert a thin chisel between them and tap lightly with a hammer to pop what's left of the weld after grinding/cutting. I have yet to actually assemble a power transformer, but it's possible to do it. Will it work though? I don't know yet. Most likely yes, I would think. I wouldn't reuse any of the original windings though.
As for dimensions, I've seen several different versions. The one that I modified has the dimensions below. I don't know what standard size that it translates to though.
As for dimensions, I've seen several different versions. The one that I modified has the dimensions below. I don't know what standard size that it translates to though.
Attachments
Last edited:
That works too, but being stuck with the original 60 Hz primary is a no-no for a tube amp. Even as a power trafo. Unless you like your amp to be heard two blocks away with the volume turned all the way down. Well, it’s really no louder than your microwave oven. Which is still louder than the fan on a Behringer iNuke.
I did it once - with three of them. Each with a 6 volt winding out of #6 THHN. 4 120A stud diodes. Never had to take the cores apart. Starts any vehicle regardless of the condition of the battery. You should hear the buzz when it’s plugged in - and again when you turn the key.
@JasonWatkins: Either you did some error in your measurements, or your laminations aren't standardized ones. Standard »lossless« EI laminations are stamped by a pair the same time or multiples of a pair. The two E parts are facing each other by their open sides, the I parts come from the winding windows. Hence, in a standard core of, say, 105 mm width (= height in your drawing), the window should be 52.5 mm deep instead of 50 mm.
Best regards!
Best regards!
It is 52mm. 52.62 to be exact. I thought you just wanted the rough size so I didn't get too precise. The only example that I have with me to measure ATM still has the original winding in it too, complicating the process.
@wg_ski: Of course one cannot just reuse a MOT's primary, as these transformers aren't designed to be used for more than just a few minutes without break (decreased turns per volt counts, increased induction). In addition, they're cooled by means of a fan. But wasn't the idea to completely rewind it and just use the core?
Best regards!
Best regards!