BIG transformers

[theoldwizard1]

So I am researching the cost of build an inverter.  The cost of bare bones inverter boards, on eBay, capable of several kW of power are ridiculously cheap (<$100).  The trick is purchasing/winding a transformer that will take an approximate 16VAC @ 60Hz output and transforming it into 120VAC.

I am willing to wind my own, but I can not find a source for large toroid cores or even silicon steel E-I cores.  Anyone ?

Are custom wound transformer (maybe 2-3 kVA) with in the price range of a DIYer.

[MiDi]

Think there is very low demand for this, nobody, except the manufacturers go to design and build such big transformers and test and get certified them.
Why not buy one that fits?

[theoldwizard1]

Think there is very low demand for this, nobody, except the manufacturers go to design and build such big transformers and test and get certified them.
Why not buy one that fits?

There is nothing "off the shelf" that would work and I am afraid to even ask for a quote on a custom made one !

[MiDi]

There is no 120V to 16V awailable?
Only to make sure: you simply swap input and output, transformers work both ways...

[ahbushnell]

So I am researching the cost of build an inverter.  The cost of bare bones inverter boards, on eBay, capable of several kW of power are ridiculously cheap (<$100).  The trick is purchasing/winding a transformer that will take an approximate 16VAC @ 60Hz output and transforming it into 120VAC.

I am willing to wind my own, but I can not find a source for large toroid cores or even silicon steel E-I cores.  Anyone ?

Are custom wound transformer (maybe 2-3 kVA) with in the price range of a DIYer.

Why would you run an inverter at 16 volts at 2 -3 kva? That's 190A. 

[edpalmer42]

So you're going to take a low DC voltage and turn it into mains voltage?  The perfect source for such a transformer is an old UPS.  You're doing the same thing they do.  Dead UPSs are available cheap or free if you're lucky.  If you look for a 2U rackmount unit, you have a good chance of getting a toroidal transformer if that's what you really need.  In fact, it might be practical to rebuild the UPS into an inverter with better cooling so that it can run at full power continuously.  A UPS typically isn't designed to do that.

Ed

[T3sl4co1l]

Toroidy.pl?  Wind-your-own option, basically.

Transformers are perfectly symmetrical, so if you can find a 120:16V transformer, just run it backwards and you're done.

You may find it's, well, lighter or more compact, certainly; most likely better, but probably not easier, to do it with multiple DC-DC stages (wired in parallel, phase interleaved for lower ripple), to make a high voltage rail, then inverter-ize that for AC output.  Here's an example:



Left: 12 to 170V converter (about 1A max output) and 12:12 DC-DC (for isolation); middle board: PWM controller; right: mains generator (controller, including current limiting, drivers and inverter); breadboard: filter chokes; other bits of wire and outlet: AC output going to the lamp illuminating the scene, actually (~50W load).

A 170VA transformer is sized about the same as the whole assembly, just for this poorly built prototype.  The real downside is, unless you're doing this for the academic exercise (as I did), you're better off buying one.  This is a $40 value, at best.  And it doesn't even have a shiny extruded aluminum housing!

Given that 2kVA inverters range from laughably shitty to actually-good, eh, you might not find one that's actually 2kVA and a reasonable price (say circa $300?), but if you need one, it's a far, far better deal than anything you can make (sadly!).

Tim

[theoldwizard1]

First, I HAVE been around the block a time or 2 and I do know that transformers work in both directions.  (The "old" in my name isn't there for no reasons !)

The 2-3kVA rating is for a 24VDC to 120VAC inverter that is capable of 2000W continuous.

The 16VAC come from a peak-to-peak voltage of about 20V.  (RMS voltage = PP voltage / √ ̅2)

[sibeen]

The 16VAC come from a peak-to-peak voltage of about 20V.  (RMS voltage = PP voltage / √ ̅2)

Cough You may want to look that over once more Cough

[theoldwizard1]

The 16VAC come from a peak-to-peak voltage of about 20V.  (RMS voltage = PP voltage / √ ̅2)

Cough You may want to look that over once more Cough

OOPPSS !!   √ ̅2 is 1.414  NOT 1.212 !  So more like 14VAC !

Old guy with CRS disease (Can't remember SH!T)!

[sibeen]

I'd try one more time.

The RMS is √ ̅2 of the peak, not the peak-peak, so 7.071 would be close enough for field work

[T3sl4co1l]

First, I HAVE been around the block a time or 2 and I do know that transformers work in both directions.  (The "old" in my name isn't there for no reasons !)

Just checking. We have lots of newbies and oldbies here alike, good to know

Tim

[mariush]

You should be able to easily find 110v -> 15v AC  or 120v -> 15v AC ....  or 110v AC -> 18v AC (6.11 ration)  / 120v AC -> 18v AC  (6.66 ratio)

If you reverse it and put 16v AC on the 18v winding, it should still work, but instead of getting 120v, maybe you'll get  only 100v AC  ( with same ratios above, 98v ac or 106v ac )... and pretty much everything on the market should still tolerate it. 

[ahbushnell]

I'd try one more time.

The RMS is √ ̅2 of the peak, not the peak-peak, so 7.071 would be close enough for field work

So now the inverters needs to produce 3000/7=428 A RMS. 

[CJay]

An old APC UPS would do exactly what you need but if you really want to roll your own, I'd think about using low voltage DC to power a high frequency switcher to convert it up to higher voltage DC and then convert to AC

[theoldwizard1]

I'd try one more time.

The RMS is √ ̅2 of the peak, not the peak-peak, so 7.071 would be close enough for field work

I don't like to dispute anyone on line, ESPECIALLY when I can not find 100% proof, but ..

"The RMS is √ ̅2 of the peak, not the peak-peak ..."  Yes, that is the "classic definition, BUT it assumes that GROUND is in the center of the sine wave.  In this case, ground is at the bottom of the wave.

[T3sl4co1l]

Well, good luck transforming an offset sinewave.

Also to be perfectly specific, note that's sine wave only, and other ratios for other waveforms (of note, triangle is sqrt(3), square is 1, rectangle wave is sqrt(D)*sqrt(1-D), etc.).

Tim

[theoldwizard1]

Well, good luck transforming an offset sinewave.

A "generated" sine wave (electronically or mechanically) does not know "ground" until it is connected to it.  It "floats".  You are just used to the fact that US mains have neutral and ground "bonded" typically at the first disconnect after the meter.

[T3sl4co1l]

Hmm?

I meant more to the point that, if it's going through a transformer, the average value (across the transformer) is either zero, or the transformer is a charred wreck of molten copper.  Transformer don't care about ground.

Tim

[Gregg]

Look up buck boost transformers like this one https://www.ebay.com/itm/NEW-GE-5KVA-transformer-1-one-single-phase-120v-240v-16v-32v-DELTA-Buck-Boost/282795565837

[sibeen]

I'd try one more time.

The RMS is √ ̅2 of the peak, not the peak-peak, so 7.071 would be close enough for field work

I don't like to dispute anyone on line, ESPECIALLY when I can not find 100% proof, but ..

"The RMS is √ ̅2 of the peak, not the peak-peak ..."  Yes, that is the "classic definition, BUT it assumes that GROUND is in the center of the sine wave.  In this case, ground is at the bottom of the wave.

I think I'll stick with the classic definition Really, I also think you need to look at how an inverter will produce an output AC waveform. I'm assuming here that you are looking at a standard H bridge, PWM controlled inverter. If you had a waveform where the sinewave was offset in such a manner that the bottom peak was the ground of the system then I'd call that wave DC because it certainly ain't AC, and as Tim pointed out putting something like that through a transformer is going to ensure a world of hurt.

[Gregg]

[/quote]
I think I'll stick with the classic definition Really, I also think you need to look at how an inverter will produce an output AC waveform. I'm assuming here that you are looking at a standard H bridge, PWM controlled inverter. If you had a waveform where the sinewave was offset in such a manner that the bottom peak was the ground of the system then I'd call that wave DC because it certainly ain't AC, and as Tim pointed out putting something like that through a transformer is going to ensure a world of hurt.
[/quote]
The transformer primary input could care less where earth or other ground is referenced; it is just a coil of wire wound around a core, as is the transformer secondary.  Plus the secondary can be totally isolated from the primary which means one could attach the ground anywhere appropriate for the desired output.  In other words, an input sine wave or modified sine wave input 0 to 16 volts won't make a difference to the transformer from the same wave input plus and minus 8 volts.   

[sibeen]

I think I'll stick with the classic definition Really, I also think you need to look at how an inverter will produce an output AC waveform. I'm assuming here that you are looking at a standard H bridge, PWM controlled inverter. If you had a waveform where the sinewave was offset in such a manner that the bottom peak was the ground of the system then I'd call that wave DC because it certainly ain't AC, and as Tim pointed out putting something like that through a transformer is going to ensure a world of hurt.
[/quote]
The transformer primary input could care less where earth or other ground is referenced; it is just a coil of wire wound around a core, as is the transformer secondary.  Plus the secondary can be totally isolated from the primary which means one could attach the ground anywhere appropriate for the desired output.  In other words, an input sine wave or modified sine wave input 0 to 16 volts won't make a difference to the transformer from the same wave input plus and minus 8 volts.   
[/quote]


Sure, all true, a transformer doesn't care about a reference, swap the active and neutral around on a transformer, no biggie; ground one of the primary legs which is what is done in a TNS-C system, the transformer will merrily operate. But in these cases the hot wire compared to the cold wire goes up positive 10 volts and then it goes down to negative 10 volts. The overall voltage across the transformer is zero.

That's not what the OP is suggesting. He's got a waveform where the hot to the cold goes up to positive 20 volts and then down to zero volts. The average is a 10 volt offset. Put that through a transformer and you'll generate a shedload of heat and not much else - until the smoke gets out.

[theoldwizard1]

That's not what the OP is suggesting. He's got a waveform where the hot to the cold goes up to positive 20 volts and then down to zero volts. The average is a 10 volt offset. Put that through a transformer and you'll generate a shedload of heat and not much else - until the smoke gets out.

I think you need to do your home work !  This is not somewhat off design.  It is a well proven design that has been around for MANY years, used heavily in the RV/Marine/trucking industry.

What is new is, how inexpensive the implementation has gotten to be.  Ebay - 3kW inverter board -$80

Basically, add a housing, cooling and an output transformer and you are there.  The whole design revolves around the EG Micro EGS8010 chip  Datasheets

[sibeen]

That's not what the OP is suggesting. He's got a waveform where the hot to the cold goes up to positive 20 volts and then down to zero volts. The average is a 10 volt offset. Put that through a transformer and you'll generate a shedload of heat and not much else - until the smoke gets out.

I think you need to do your home work !  This is not somewhat off design.  It is a well proven design that has been around for MANY years, used heavily in the RV/Marine/trucking industry.

What is new is, how inexpensive the implementation has gotten to be.  Ebay - 3kW inverter board -$80

Basically, add a housing, cooling and an output transformer and you are there.  The whole design revolves around the EG Micro EGS8010 chip  Datasheets

WAIT....WHAT???

The unit you have linked to is a 24VDC (or 48/60) to 220VAC inverter. The output of that unit is 220VAC. This means it will have a peak voltage of 311 volts above and below the neutral or reference line, or 622 volts peak to peak. You don't need a transformer unless you want to change the 220 volts into a 110 volt system. If you want to do that you'd be far better off just buying an inverter that will output the 120 volts.

I really don't understand where your 16 volts from the OP comes into it?

[theoldwizard1]

The unit you have linked to is a 24VDC (or 48/60) to 220VAC inverter. The output of that unit is 220VAC.

In fact that unit does NOT output 220VAC.  That would be AFTER the output transformer that I have been discussing.

Read the datasheet I referenced.  That board contains an IC that is basically a sine wave generator at either 50Hz or 60Hz.  This is feed to a Class D amplifier that goes to MOSFET drivers that goes to the final drive MOSFETs.  The voltage out of the MOSFETs ranges from slightly higher than the DC zero volts to something slightly less then the DC maximum positive volts.  The gain of the amplifier is adjusted mased on a voltage divider circuit that is connected AFTER the transformer.



At this point you are just being ARGUMENTATIVE.  Do some real research, like READ THE DARN DATASHEET !

[sibeen]

You linked to a site selling a 220VAC output inverter. You also linked to the data sheet of a controller chip. The chip data sheet shows that the chip can be used different techniques for inverter operation, one of which includes a transformer option and therefore a low DC voltage can be used as the input to the output inverter switches. The other techniques show a 400 volt DC link as the driving voltage to the inverter output switches.

The inverter you linked to will be of the latter type construction. It will have an intermediate stage which takes the incoming 24VDC and converts this to a 400VDC bus, so a DC to DC converter. This 400 volts will then be fed to the output switching devices of the inverter to give the required output AC waveform. The high DC voltage is required to be at a level that is above the peak voltage of the AC waveform, which for a 220 VAC machine will be 311 volts. For a lower output voltage the designers may even use a lower DC link voltage.

So this inverter is really a DC-DC-AC style machine.

The inverter can also come with different output voltages and here is one that has a 110 or 120 VAC output.

https://www.ebay.com/itm/3000W-Pure-Sine-Wave-Power-Frequency-Inverter-Board-DC-48V-24V-AC-to-110-120V/172689320870?_trkparms=aid%3D888007%26algo%3DDISC.MBE%26ao%3D1%26asc%3D20131227121020%26meid%3D318d3f0e24de4e35b3bcedf582717058%26pid%3D100009%26rk%3D1%26rkt%3D1%26sd%3D292739535438%26itm%3D172689320870&_trksid=p2047675.c100009.m1982

[sibeen]

Actually, the 120 Volt unit that I linked to does need a transformer and it shows it on the website that I linked to, perhaps I should have read to the bottom of the page

The page actually  shows the calculations you'll need to build the transformer.  Now the OP makes a bit more sense. You're not switching 16 volts AC across the transformer when using this inverter board. When using a 24 volt battery for the first half (positive) of the sine wave you'll be switching the positive 24 volts from primary 1 to primary 2 (P1 to P2) with a PWM waveshape. The switching transistors are just switching the DC voltage across the primary coil.  In the second half (negative) of the sinewave that will be switched around and the positive 24 volt will connect to P2 and you'll be switching from P2 to P1 with a PWM waveshape.  So with a H bridge configuration you'll be switching transistors 1 and 3  for the positive section of the sinewave and then transistors 2 and 4 for the neagative half.

That first inverter you linked to is very confusing as it does not mention a transformer at all - as far as I can see.

[fourtytwo42]

Oldwizard if you go to this site http://www.thebackshed.com/forum/home.asp it is almost entirely devoted to inverters built around the egs chip. They are mostly Australian outbackers running entirely off grid, 6Kw being a typical size. If you trawl through the threads or do a search there is lots of info on transformers and of course a whole host of practical experience. Good luck

[theoldwizard1]

Oldwizard if you go to this site http://www.thebackshed.com/forum/home.asp it is almost entirely devoted to inverters built around the egs chip.

Been there, done that, asked the same question did not get a definitive source for transformers or cores.

[theoldwizard1]

You linked to a site selling a 220VAC output inverter. You also linked to the data sheet of a controller chip. The chip data sheet shows that the chip can be used different techniques for inverter operation, one of which includes a transformer option and therefore a low DC voltage can be used as the input to the output inverter switches. The other techniques show a 400 volt DC link as the driving voltage to the inverter output switches.

The inverter you linked to will be of the latter type construction.

In fact, that board doess NOT step up the DC !  I am pretty certain they are EXACTLY the same board. (The description was written by someone that does not use English as their primary language, so you have to take it with a LARGE "grain of salt" !)

I have STUDIED the specs and example implementations of the EGS8010 IC. I have REVIEWED many different implementation of inverter of both designs.  The "step up" inverters use multiple E-I core transformers (mounted on the board) and a high frequency (100 kHz ?) switch mode power supply boost converter circuit (multiple PWM chips available for this application).  This design is only slightly more complex than the low frequency design but has the bonus of using smaller, easier to wind transformers.  For whatever reason, this design does NOT seem very "robust".  (The magic smoke seem to come out.)

While the link you posted did give some formulas for winding transformers it does NOT answer the original question I asked in this thread

I am willing to wind my own, but I can not find a source for large toroid cores or even silicon steel E-I cores.  Anyone ?

[fourtytwo42]

Been there, done that, asked the same question did not get a definitive source for transformers or cores.

Then the only suggestion I could offer is a large toroid auto-transformer with enough window left for your primary. I have used that method myself for custom grid trafo's. Also if you notice people like Oztules has stacked toroids for higher output powers. The reason I say auto-transformers is they can often be had at a lot lower price per Kw.

[ahbushnell]

I'd try one more time.

The RMS is √ ̅2 of the peak, not the peak-peak, so 7.071 would be close enough for field work

I don't like to dispute anyone on line, ESPECIALLY when I can not find 100% proof, but ..

"The RMS is √ ̅2 of the peak, not the peak-peak ..."  Yes, that is the "classic definition, BUT it assumes that GROUND is in the center of the sine wave.  In this case, ground is at the bottom of the wave.

And what happens when you connect that to the transformer?  Not much DC going through. 

[f4eru]

For big 50 Hz transformers, I'll recommend to not wire it yourself, because it's a real pain to put the tousands of turns of wire, then interlace all the sheet metal.

For a good source of powerful transformers, i'll recommend microwave oven transformers. You can cut out the HV secondary, and wire in your 2x12V primary.
You'll have to remove the core shunt though.

Why 2x12V  ? glad you asked! I'll recommend a push pull configuration.

[vk6zgo]

I'd try one more time.

The RMS is √ ̅2 of the peak, *not the peak-peak, so 7.071 would be close enough for field work

I don't like to dispute anyone on line, ESPECIALLY when I can not find 100% proof, but ..

"The RMS is √ ̅2 of the peak, not the peak-peak ..."  Yes, that is the "classic definition, BUT it assumes that GROUND is in the center of the sine wave.  In this case, ground is at the bottom of the wave.

*No, RMS is peak times the reciprocal of the square root of 2, which is 0.7071.
Hence, peak voltage of a 230 v RMS ac waveform is about 325v.

[Red Squirrel]

You can probably skip the bulky 60hz transformer by doing everything in high frequencies using SMPS transformers, then the output would basically be a smoothed PWM genreated 60hz sine wave. I'm not an expert only read up on this a tad as I would like to experiment with making my own power stuff at some point.

I was curious and it is indeed hard to find the hardware to make your own 60hz transformers.  Suppose you could get a metal fab shop to make the parts if you wanted to.  Not sure how much that would cost though, probably a lot.

[f4eru]

Yes, it would cost a lot to do prototypes of silicon steel parts.
On a former job, for proto laminations for motors, we used electro-erosion on complete lamination packets for creating a shape in the full packet. Of course, you can't alternate interlaced shapes (not needed usually for a motor), but you can do the same with a transformer, the alternating is a small optimization you could skip.
I don't recommend the electro-erosion because it'S expensive.

Also, of course, the HF (100kHz) SMPS version is much more compact, less cost, and higher efficiency, but it's a hassle to develop, and not worth it for only a few pieces compared to the 50Hz bulky transformer variant.

[jh15]

Not a good idea to piss off a wizard. You are (grey) bearding the lion.