There may be members here that have done that, but not many posts about it. This forum is by and large a forum for plugging AIO inverters into server rack batteries. Again, nothing wrong with that, as that is the intent. If you want inverter plugging in knowledge, this is the place. If you want inverter building knowledge, Backshed is where you go.
Warpverter
- Thread starter Nobodybusiness
- Start date
I think we've got a good blend of both. Of course, with 90k members, you can't expect a 50/50 split...
Again, I'm NOT expecting that at all. The reactions to the factual statement of the nature of this forum is puzzling.
Me:"This maternity clinic mostly delivers babies."
Others: "Obviously you missed the time we gave CPR to a heart attack victim. It's a good mix."
I didn't mean it literally. But yes, your clinic analogy is correct - there are dozens of us! So sometimes we tend to point it out to draw attention to it. Hence the reactions.
RCinFLA
Solar Wizard
- Joined
- Jun 21, 2020
- Messages
- 3,565
Warpspeed
Solar Wizard
The mosfet versus IGBT argument for inverters is interesting......
They both have advantages and disadvantages. For surge overload capability, IGBTs win hands down.
The reason for that is that mosfets when turned on act like low value resistors.
If you double the current, the voltage drop also doubles, and the power dissipated goes up four times.
A times ten surge current, and power dissipated goes up 100 times, and your mosfet instantly explodes.
An IGBT is different, in that the rated voltage drop at full max current might be a couple of volts, but it does not go up much higher than that at much higher current. So power dissipation tends to be very roughly proportional to current. A ten times surge overload, power might go up 15 times (as a guess) not to 100 times.
Diodes act in a rather similar way, you lose about 0.6v in the forward direction, but even with a massive current surge, its not likely to get up to even 1 volt.
IGBTs also have lower conduction losses at higher voltages. You will find IGBTs fitted in all high dc voltage grid tie inverters.
Mosfets are vastly better at lower voltages, and is what you will find in typical low voltage inverters.
Things like modern welders, and plasma cutters, use IGBTs as well.
I decided to run IGBTs in my Warpverter as its the logical choice for 100v operation. Very large old and slow IGBTs rated at hundreds of amps are not expensive, especially the Chinese clones, and the big IGBT power blocks are very compact and easy to bolt onto busbars and heavy lugs. Overload surge capacity of these big IGBTs is crazy high.
The 200 amp IGBTs I am using are rated to carry 1,000 amps of fault current for one full half mains cycle.
That is why my Warpverter has no electronic overcurrent protection.
Just a standard thermal/magnetic circuit breaker on the ac output.
Any dc current surge to trip the output breaker comes mostly from the electrolytics.
I have a two pole 63 amp dc breaker on the input, but in over six years of operation I have never tripped that.
Its always the ac breaker that opens.
Its seen quite a bit of unintended abuse over the years, and its never missed a beat in all that time.
This all works so well at 100v, I would not be surprised if similar large IGBTs would also be very successful at 48v.
While large IGBTs do drop a lot of voltage at full rated current, they are surprisingly efficient at much less than full rated current.
As domestic inverters generally run at low average power, with only occasional high peaks, I believe suitably oversized IGBTs might be brilliant in a 48v Warpverter.
In fact Alston over at The Back Shed is putting together a 48v Warpverter using large IGBTs right now.
Its now in the very final stages of completion, and I can hardly wait to see the results.
A very short thread with lots of pictures:
https://www.thebackshed.com/forum/ViewTopic.php?FID=4&TID=15183
Alston is using Klaus's control board, which I would highly recommend.
Klaus's board is identical to mine in every respect, but he has added the Hall effect feed forward load compensation modification.
My inverter has excellent line regulation, but load droops about ten volts at 5Kw, something that is not really a problem here.
The Hall sensor load compensation feature can be tweaked to have both excellent line and load regulation.
It can even be adjusted so that the load voltage increases slightly with increasing load, if you wished to do that, and its guaranteed 100% stable.
They both have advantages and disadvantages. For surge overload capability, IGBTs win hands down.
The reason for that is that mosfets when turned on act like low value resistors.
If you double the current, the voltage drop also doubles, and the power dissipated goes up four times.
A times ten surge current, and power dissipated goes up 100 times, and your mosfet instantly explodes.
An IGBT is different, in that the rated voltage drop at full max current might be a couple of volts, but it does not go up much higher than that at much higher current. So power dissipation tends to be very roughly proportional to current. A ten times surge overload, power might go up 15 times (as a guess) not to 100 times.
Diodes act in a rather similar way, you lose about 0.6v in the forward direction, but even with a massive current surge, its not likely to get up to even 1 volt.
IGBTs also have lower conduction losses at higher voltages. You will find IGBTs fitted in all high dc voltage grid tie inverters.
Mosfets are vastly better at lower voltages, and is what you will find in typical low voltage inverters.
Things like modern welders, and plasma cutters, use IGBTs as well.
I decided to run IGBTs in my Warpverter as its the logical choice for 100v operation. Very large old and slow IGBTs rated at hundreds of amps are not expensive, especially the Chinese clones, and the big IGBT power blocks are very compact and easy to bolt onto busbars and heavy lugs. Overload surge capacity of these big IGBTs is crazy high.
The 200 amp IGBTs I am using are rated to carry 1,000 amps of fault current for one full half mains cycle.
That is why my Warpverter has no electronic overcurrent protection.
Just a standard thermal/magnetic circuit breaker on the ac output.
Any dc current surge to trip the output breaker comes mostly from the electrolytics.
I have a two pole 63 amp dc breaker on the input, but in over six years of operation I have never tripped that.
Its always the ac breaker that opens.
Its seen quite a bit of unintended abuse over the years, and its never missed a beat in all that time.
This all works so well at 100v, I would not be surprised if similar large IGBTs would also be very successful at 48v.
While large IGBTs do drop a lot of voltage at full rated current, they are surprisingly efficient at much less than full rated current.
As domestic inverters generally run at low average power, with only occasional high peaks, I believe suitably oversized IGBTs might be brilliant in a 48v Warpverter.
In fact Alston over at The Back Shed is putting together a 48v Warpverter using large IGBTs right now.
Its now in the very final stages of completion, and I can hardly wait to see the results.
A very short thread with lots of pictures:
https://www.thebackshed.com/forum/ViewTopic.php?FID=4&TID=15183
Alston is using Klaus's control board, which I would highly recommend.
Klaus's board is identical to mine in every respect, but he has added the Hall effect feed forward load compensation modification.
My inverter has excellent line regulation, but load droops about ten volts at 5Kw, something that is not really a problem here.
The Hall sensor load compensation feature can be tweaked to have both excellent line and load regulation.
It can even be adjusted so that the load voltage increases slightly with increasing load, if you wished to do that, and its guaranteed 100% stable.
Last edited:
The point I’m making is that although the back shed is cool, if you are looking for advanced technical knowledge regarding inverter design and construction, you will have better results here.
Sure 99% of the posts are regarding the basics, and this place also helps out there.
It’s more about the sheer numbers of members, and the number of technically capable minds that are present.
You can try for yourself if you like - ask an engineering or design question on both sites and see how you go ?
You might be right regarding numbers. I've read a great deal of threads and there are many smart people here no doubt. However, many questions end up getting the "You'll shoot your eye out, kid!" treatment. One thread I read had a guy asking about high voltage DC DIY charge controller design.
Forum: The consensus? Don't do it, high voltage DC is dangerous. Yeah, but I'm an electrician used to dealing with HVDC, I just need some guidance on the types of switch mode converters used. No, don't do it, HVDC is dangerous. Yeah, but I'll be OK. No, I'm not going to give advice like that...
Backshed: How about a resonant forward converter? Maybe a full bridge design with a transformer for isolation? Be sure to use 600 volt MOSFETS...
Lots of times engineers have no practical experience. Almost everyone on Backshed has wound a transformer. Nearly everyone here would likely say just buy one.
That seems like an unnecessary pissing contest. The world is big enough for multiple forums catering to different levels and domains of DIY. For example, I wouldn't look for hands-on info on EV inverter design and builds here or backshed. OpenInverter is a much better place. Be glad Warp decided to come here often enough to share his knowledge. It was sad to see him leave backshed because apparently some idiot pissed him off.
Warpspeed
Solar Wizard
I find the biggest problem here is the sheer volume of posts.
Its far too easy to miss something potentially really interesting.
Its like wading through a thick fog looking for shiny nuggets.
Its far too easy to miss something potentially really interesting.
Its like wading through a thick fog looking for shiny nuggets.
stienman
Mostly Harmless
- Joined
- Jan 6, 2021
- Messages
- 477
Out of curiosity, since old, used 3 phase transformers are cheap in the US (ie, https://www.ebay.com/itm/334855546358 shows a $400 45kVA three phase) can a 3 transformer warpverter be made using a 3 phase transformer form? I suppose a 4th transformer could be added for the tiny one if THD was important, but if I wanted a large warpverter it seems like a relatively cheap way to go. These transformers commonly come in 3, 6, 8, 15, 30, 45, and 75 kVA sizes.
Of course it would need to be re-wound, like every other warpverter, though I'm guessing if the transformer has taps for 240 and 120 on the 480 side and a 480VDC source was used then it might be able to be used as-is.
Of course it would need to be re-wound, like every other warpverter, though I'm guessing if the transformer has taps for 240 and 120 on the 480 side and a 480VDC source was used then it might be able to be used as-is.
Hedges
I See Electromagnetic Fields!
- Joined
- Mar 28, 2020
- Messages
- 20,964
Most 3-phase transformers are 3 sets of primary/secondary windings on an "E" core. A few are 3 separate transformers.
I think you'll do better with multiple single-phase transformers. No rewinding needed according to my analysis:
The most common have 240/480V primaries and 120/240V secondaries. That gives you four ratios (well, three; 240:120 and 480:240 are same ratio, different current handling). Consider 96V fed into 1:1, 2:1 4:1; that gives 96V + 48V + 24V = 168V, the peak of 120Vrms. I think they'll work for 240V output as well, with 192VDC input. That should be enough below the 340Vpeak of a 240VAC winding to not saturate (I think, different waveform from what I applied.)
Three like this one, or better three of different kVA ratings:
If you use 3-phase there will be at least some interaction among the coils on different legs of the "E". I don't understand it well, but I ran into issues when feeding imbalanced 3-phase into an E-core 3-phase transformer. (I was using 3x Sunny Islands, one passing through 125Vrms from grid, the other two driving the other phases at 120V. Too much current was being drawn. It worked better with "Neutral" center of "Y" unconnected from source. I did connect it to neutral of Sunny TriPower and got OK results.)
Some have a number of voltage taps. I would try to avoid needing to rewind. These were originally wound on bobbins, then the E core inserted.
I think I rewired HV primary as Wye instead of Delta,
connected L1, L2, L3 (but not N) of 120/208Y source to HV tap 3 "230V"
Connect N to new center of Wye and L1, L2, L3 to HV tap 2 "460V".
That boosted 120V to 240V, for 240/416Y, which I fed to Sunny TriPower.
I originally fed 120V into HV tap 4 "200V" so HV tap 2 "460V" produced 276V for 277/480Y, but that caused TriPower to trip off after connecting. Maybe voltage rose too high.
I think you'll do better with multiple single-phase transformers. No rewinding needed according to my analysis:
The most common have 240/480V primaries and 120/240V secondaries. That gives you four ratios (well, three; 240:120 and 480:240 are same ratio, different current handling). Consider 96V fed into 1:1, 2:1 4:1; that gives 96V + 48V + 24V = 168V, the peak of 120Vrms. I think they'll work for 240V output as well, with 192VDC input. That should be enough below the 340Vpeak of a 240VAC winding to not saturate (I think, different waveform from what I applied.)
Three like this one, or better three of different kVA ratings:
If you use 3-phase there will be at least some interaction among the coils on different legs of the "E". I don't understand it well, but I ran into issues when feeding imbalanced 3-phase into an E-core 3-phase transformer. (I was using 3x Sunny Islands, one passing through 125Vrms from grid, the other two driving the other phases at 120V. Too much current was being drawn. It worked better with "Neutral" center of "Y" unconnected from source. I did connect it to neutral of Sunny TriPower and got OK results.)
Some have a number of voltage taps. I would try to avoid needing to rewind. These were originally wound on bobbins, then the E core inserted.
I think I rewired HV primary as Wye instead of Delta,
connected L1, L2, L3 (but not N) of 120/208Y source to HV tap 3 "230V"
Connect N to new center of Wye and L1, L2, L3 to HV tap 2 "460V".
That boosted 120V to 240V, for 240/416Y, which I fed to Sunny TriPower.
I originally fed 120V into HV tap 4 "200V" so HV tap 2 "460V" produced 276V for 277/480Y, but that caused TriPower to trip off after connecting. Maybe voltage rose too high.
Warpspeed
Solar Wizard
This will only work with four (or three) completely independent transformers.
If you need split phase for 110/220v as in the US, that can most easily be done by winding two secondaries (of half voltage each) on each transformer, as suggested in a much earlier post by by Hedges.
In descending order of preference:
1/ Buy brand new good quality toroidal cores of a suitable size. Relatively expensive, and a lot of work to wind.
Might be worth getting a quote.
2/ Recover the toroids from old blown up grid tie inverters, or other high power equipment.
These cores can be stacked to increase the power level.
Lowest cost solution by far, still a lot of work to wind, but highly recommended.
3/ Buy brand new grain oriented steel E and I laminations. That is what I did, it will be the most expensive for the core, but the easiest to wind .
4/ Recovered ordinary crappy steel laminations salvaged from old E and I transformers, will work, but not very well.
The problem is the low quality of the steel, the greater thickness of the laminations, and the unavoidable very small gaps between the ends of the laminations where they meet.
This will produce a lower inductance and higher eddy current loss creating a much higher idling power.
Definitely not recommended for a typical domestic inverter.
But if you already have such a transformer you could always do some tests to decide if you can live with the very high losses.
Perhaps you can, its a judgement call......
If you plan to run one large inverter to only drive one particular load, and the inverter is completely off when the load is off, high idling power is no longer a disadvantage. Something for the deep well pump guys to think about perhaps ?
This might be quite practical for running large intermittent motor loads. One inverter per motor.
Warpverter is ideal for heavy induction motor starting, very high surge power, and excellent tolerance of highly reactive loads.
If you need split phase for 110/220v as in the US, that can most easily be done by winding two secondaries (of half voltage each) on each transformer, as suggested in a much earlier post by by Hedges.
In descending order of preference:
1/ Buy brand new good quality toroidal cores of a suitable size. Relatively expensive, and a lot of work to wind.
Might be worth getting a quote.
2/ Recover the toroids from old blown up grid tie inverters, or other high power equipment.
These cores can be stacked to increase the power level.
Lowest cost solution by far, still a lot of work to wind, but highly recommended.
3/ Buy brand new grain oriented steel E and I laminations. That is what I did, it will be the most expensive for the core, but the easiest to wind .
4/ Recovered ordinary crappy steel laminations salvaged from old E and I transformers, will work, but not very well.
The problem is the low quality of the steel, the greater thickness of the laminations, and the unavoidable very small gaps between the ends of the laminations where they meet.
This will produce a lower inductance and higher eddy current loss creating a much higher idling power.
Definitely not recommended for a typical domestic inverter.
But if you already have such a transformer you could always do some tests to decide if you can live with the very high losses.
Perhaps you can, its a judgement call......
If you plan to run one large inverter to only drive one particular load, and the inverter is completely off when the load is off, high idling power is no longer a disadvantage. Something for the deep well pump guys to think about perhaps ?
This might be quite practical for running large intermittent motor loads. One inverter per motor.
Warpverter is ideal for heavy induction motor starting, very high surge power, and excellent tolerance of highly reactive loads.
Last edited:
Hedges
I See Electromagnetic Fields!
- Joined
- Mar 28, 2020
- Messages
- 20,964
I think the losses from E-core transformer may not really be too bad, so long as you operate well below the voltage for a given turn count that it was designed for.
25 kVA transformer, driving one of two primary windings at half of nominal voltage, it carried 45W of reactive power (I didn't determine real power.) That winding and voltage would have been capable of delivering at least 6kW full load, so reactive power a bit under 1%. I'm not sure if it would have been same driving both primary windings in parallel, which would support 12.5 kW full load, but I think so.
Not as good as the 6W no-load for a 5kW inverter mentioned above but still practical to use. The possibility of using off the shelf (or from the scrap yard) transformers without having to rewind greatly reduces labor to build.
diysolarforum.com
25 kVA transformer, driving one of two primary windings at half of nominal voltage, it carried 45W of reactive power (I didn't determine real power.) That winding and voltage would have been capable of delivering at least 6kW full load, so reactive power a bit under 1%. I'm not sure if it would have been same driving both primary windings in parallel, which would support 12.5 kW full load, but I think so.
Not as good as the 6W no-load for a 5kW inverter mentioned above but still practical to use. The possibility of using off the shelf (or from the scrap yard) transformers without having to rewind greatly reduces labor to build.
transformer excitation current - isolation transformer used as auto-transformer
So you think you can backfeed a transformer secondary, either as step-up or as auto-transformer? Turns out these things aren't so ideal and reversible as one might expect. It isn't just the greater inrush (primary is often wound outside secondary to reduce inrush current), but also idle current...
diysolarforum.com
Warpspeed
Solar Wizard
As you say, using an existing winding at a much lower voltage does exactly what we want, lower the flux density.I think the losses from E-core transformer may not really be too bad, so long as you operate well below the voltage for a given turn count that it was designed for.
25 kVA transformer, driving one of two primary windings at half of nominal voltage, it carried 45W of reactive power (I didn't determine real power.) That winding and voltage would have been capable of delivering at least 6kW full load, so reactive power a bit under 1%. I'm not sure if it would have been same driving both primary windings, which would support 12.5 kW full load, but I think so.
Not as good as the 6W no-load for a 5kW inverter mentioned above but still practical to use. The possibility of using off the shelf (or from the scrap yard) transformers without having to rewind greatly reduces labor to build.
transformer excitation current - isolation transformer used as auto-transformer
So you think you can backfeed a transformer secondary, either as step-up or as auto-transformer? Turns out these things aren't so ideal and reversible as one might expect. It isn't just the greater inrush (primary is often wound outside secondary to reduce inrush current), but also idle current...
The only down side is we are still limited by the maximum rated current of the particular winding stated on the rating plate.
So running at half the rated voltage will lower the power rating by half, but half of 25Kva is still a lot !
That should work pretty well as the largest transformer, as it runs at the fundamental of 50/60Hz, so eddy current losses should be low as well at half flux density. We still need to fit a suitable primary to that.
If the desirable secondary is the inside winding of the original transformer, we might be in luck stripping off the outer winding and replacing it with a suitable primary.
Its common practice to place the highest voltage winding on first, as it will have the thinnest wire and produce the flattest winding.
The lower voltage top winding can then use thicker wire and be more lumpy. Its always easier to wind like that than the other way around.
If you are seeing 45va reactive power when energised at 220v 60Hz, that sounds very good for what is potentially a 12Kw transformer.
It won't be any more than that, and usually a bit less when finally running in a Warpverter.
Warpspeed
Solar Wizard
One further thought on that first transformer.
If you run the original unmolested primary (230v) off a nominal 115v dc, or something close to that, a 450v rated secondary would then provide the desired 225v square wave. There appear to be enough tappings to play around with to get it close to being right.
So it might work pretty well as is !
There is a fair bit of wiggle room selecting the exact ratio of the largest transformer.
Once that has been determined, the next smallest inverter (4Kw) can be wound to provide exactly one third of the secondary voltage of the largest transformer.
A 120v nominal battery source for 12Kw would be pretty ideal, only 100 amps flat out.
If you run the original unmolested primary (230v) off a nominal 115v dc, or something close to that, a 450v rated secondary would then provide the desired 225v square wave. There appear to be enough tappings to play around with to get it close to being right.
So it might work pretty well as is !
There is a fair bit of wiggle room selecting the exact ratio of the largest transformer.
Once that has been determined, the next smallest inverter (4Kw) can be wound to provide exactly one third of the secondary voltage of the largest transformer.
A 120v nominal battery source for 12Kw would be pretty ideal, only 100 amps flat out.
Last edited:
Hedges
I See Electromagnetic Fields!
- Joined
- Mar 28, 2020
- Messages
- 20,964
Trouble is, power transformers generally have secondary closest to core, primary outside. That reduces transformer inrush, among other things. The tests I posted in my thread show that driving secondary at rated voltage is much worse saturation and no-load current than driving a primary.
So you're less likely to find a 240V primary, 480V secondary transformer that works well. Typically available ones will do step-down only not step-up. Which is why I proposed 96V battery for 120Vrms output. Two transformers with secondaries in series could give you double, but that's a lot of transformers. Well, one more for largest.
True, the other transformers are being switched at higher frequencies so eddy losses could be higher. I haven't played with that. Hmm, we use Chroma at work for 20 Hz to 1kHz as well as variable voltage. I don't have one of my own yet.
I was applying 120V to a 240V winding. To apply 240V, use 480V winding. In our case, 240/480 primary windings can be series or parallel.
Warpspeed
Solar Wizard
When running at very low flux density, as we will be doing, flux doubling and inrush is never a problem.
The flux CAN actually double and still remain safely below saturation.
Try starting up your 230v winding through a lamp from 110v. Inrush should be absolutely minimal.
We will need to soft start the Warpverter anyway, those big electrolytics need time to come up, and that also takes care of any remnant flux in the transformers. Both those effects work together, so don't fret about transformer current surge at start up.
The flux CAN actually double and still remain safely below saturation.
Try starting up your 230v winding through a lamp from 110v. Inrush should be absolutely minimal.
We will need to soft start the Warpverter anyway, those big electrolytics need time to come up, and that also takes care of any remnant flux in the transformers. Both those effects work together, so don't fret about transformer current surge at start up.
Warpspeed
Solar Wizard
When I started playing around with all this many years ago, I went to a lot of trouble thinking about the higher frequencies involved with the smaller transformers. It turns out that, at higher frequencies the flux swing is even lower than in the largest transformer, and we don't need to worry about eddy currents or any of that.
If all four transformers are designed for rms operation at 50/60Hz in the normal way, then they will all work perfectly well.
That is a complete fiction of course, the actual voltage waveforms are rectangular, of weird duty cycles, and there are no sine wave transformer voltages anywhere.
But the whole transformer design process can be greatly simplified by ASSUMING 50/60Hz sine wave voltages for each transformer.
All the errors and over simplifications are on the safe side.
So transformer design goes something like this:
Largest transformer primary, minimum dc input voltage might be for example 40 volts at minimum design dc battery voltage.
Largest transformer secondary 225v square wave required at 50/60Hz.
Just use Faraday's law or any transformer design software to design a 40v rms primary and a 225v rms secondary at 50/60Hz.
It will work perfectly well in a Warpverter.
Same with the other transformers.
Transformer two, 40v rms to 75v rms, at 60Hz
Transformer three 40v rms to 25v rms, at 60Hz
Transformer four 40v rms to 8.33v rms at 60Hz
All the existing Warpverter transformers have been designed this way by many other people and they all work great !
It hugely simplifies things which is absolutely necessary for an open source project.
The purists will be tearing out their hair in frustration, but I don't care !
It definitely works, its practical, and that is all that matters.
There is also another hidden advantage.
Any of the four transformers can be plugged into any of the four inverter outputs without any issues.
It makes testing and de bugging easier too, as the basic large inverter 60Hz simple slow square wave can be used to test any of the smaller inverters or smaller transformers.
The simpler waveform is much easier to see for people that have old limited bandwidth analog oscilloscopes for example.
Last edited:
Similar threads
