Warpverter

[Hedges]

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.

I'm not worried about inrush at half voltage. Full voltage, it can be a problem because they are designed to go part way into saturation with each half cycle. If power is disconnected at the end of half cycle, then reconnected at start of a half cycle of same polarity, that's when they go deep into saturation, can draw many times full load current. Operating at half normal voltage, this effect would only draw a moderate current.

I've measured inrush by repeatedly switching on & off, sometimes got quite high current. Better waveforms with Fluke i2000 Rogowski coil (20 kHz bandwidth) than with CT.

A more obvious and dramatic demonstration is applying DC with a CV/CC supply or any DC source with a series resistor for current limit and measurement. First time I apply nominal voltage, it holds off current for 20 milliseconds before current shoots up. Second time same polarity, only holds off 2 milliseconds. Reverse polarity and again one application holds off 20 milliseconds. (With a different applied voltage, time to saturate would be different, closer to the 8 or 10 milliseconds of one phase.)

 

[Hedges]

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.

The high frequencies won't come near saturation, but I'd expect higher eddy current losses in the core. I haven't looked into lamination thickness vs. frequency, could be typical transformers have much thinner laminations than needed for 60 Hz. It could help prevent heating from loads that generate harmonics.

 

[Warpspeed]

Oh yes !
I have seen transformers quite literally jump when powered up, breakers pop, fuses blow. High drama.
Mosfet exploding current surges in home made inverters at power up are far from unknown.

 

[Warpspeed]

The high frequencies won't come near saturation, but I'd expect higher eddy current losses in the core. I haven't looked into lamination thickness vs. frequency, could be typical transformers have much thinner laminations than needed for 60 Hz. It could help prevent heating from loads that generate harmonics.

That worried me too. But the reality is that the flux density is so very low in the highest frequency transformer, which is also the smallest, eddy current loss is negligible. The remarkably low idling power losses achieved by many others is an excellent indicator.
There are a lot of things to think about in all this, but its all finally worked out very well indeed.

 

[Hedges]

Not that your Warpverter will have problems, operating below half of saturation, but your PWM can be narrow enough to have no inrush and walk magnetization back to the middle. Basically degauss.

What do you think a shorted turn does to degaussing? You have 3 or 4 transformers in series. If you keep all but one in the configuration of transistors shorting out primary, what happens to core's residual magnetization when the one that does get PWM sends current through it?

I ask because we work with magnetic lenses, and attempt to degauss them. I tried to get BH curve of it like I do chokes or transformers, but without laminations I was unable at 60 Hz to drive it into saturation. The lens pole pieces acted as a shorted turn. All I achieved was induction heating.

I built a Chan model with whatever bad Hc, Br, Bs I got from the oval not S-shaped curve, and did simulate magnetization and degaussing. Likely not an accurate model, but I still showed large difference between exponential decay sine wave and a few triangle waves.

 

[CoolWill]

Oh yes !
I have seen transformers quite literally jump when powered up, breakers pop, fuses blow. High drama.
Mosfet exploding current surges in home made inverters at power up are far from unknown.

There's nothing quite like that feeling when you power up an inverter and it sits quietly and you feel quite proud. Then you switch in a lamp and the FETs explode. Quite a feeling, indeed.

 

[Hedges]

If you have a couple identical transformers left over, built yourself an amplifier. Use it for AM transmission (if it works to high enough frequency) or as a 60 Hz light dimmer.

I think the right windings on an E-core transformer is preferred over two transformers. Maybe press a 3-phase transformer into that service.

 

[Hedges]

There's nothing quite like that feeling when you power up an inverter and it sits quietly and you feel quite proud.

My professor's advice was, "try first at reduced voltage"

Then you switch in a lamp and the FETs explode. Quite a feeling, indeed.

 

[Warpspeed]

The high frequencies won't come near saturation, but I'd expect higher eddy current losses in the core. I haven't looked into lamination thickness vs. frequency, could be typical transformers have much thinner laminations than needed for 60 Hz. It could help prevent heating from loads that generate harmonics.

I did some testing of a toroidal core many years back.
If you increase the frequency, Xl obviously increases, and so does the overall impedance of a winding.
The increase is pretty gradual up to about 350Hz where it peaks, above that, eddy current loss rises steeply and impedance falls off a cliff.
The waveforms we are using have a fairly slow repetition rate, and the eddy currents don't seem to be excessive, provided thin grain oriented steel is used. Never tried it with thick cheap iron laminations.

A big existing transformer should work fine for the largest inverter.

 

[Warpspeed]

What do you think a shorted turn does to degaussing? You have 3 or 4 transformers in series. If you keep all but one in the configuration of transistors shorting out primary, what happens to core's residual magnetization when the one that does get PWM sends current through it?

I ask because we work with magnetic lenses, and attempt to degauss them. I tried to get BH curve of it like I do chokes or transformers, but without laminations I was unable at 60 Hz to drive it into saturation. The lens pole pieces acted as a shorted turn. All I achieved was induction heating.

Don't really know.
At the exact voltage zero crossing, for an instant, all four transformer primaries are shorted.
Only current there is out of phase reactive.
The voltage buildup after the zero crossing is gradual, one step at a time.

The drill here to turn the beast off is, first disconnect the load. Then disconnect the incoming dc.
The inverter continues to run off the voltage stored in the big electrolytics.
Those discharge over a few seconds effectively demagnetizing the cores with controlled gradually diminishing drive amplitude.

Startup is via a soft start up resistor without any load on the inverter.
Dc voltage across the electrolytics ramps up, and so does the flux in the transformers.
All very smooth and gentle, no sudden exciting events.

 

[Hedges]

Except for the excitement of finding AC coming from something that was turned off.
A pilot light would be a nice addition.

⚡Shock⚡ through the heart,

And you're to blame, You give solar a bad name! Meant to be safer than a cheater cord, power cord with shrouded banana plugs + Similar wire and banana connected to equipment under test. Now it's easy to stack banana plugs and plug in HV scope probe. It's just as easy to mix up Line and...

diysolarforum.com

 

[Warpspeed]

The only output indicator I have, is a 230 volt powered temperature control module for the big centrifugal heat sink air blower.
When the LED temperature display for that is up, I know the fun has started, usually about two seconds after I have held down the soft start button.

I will not close the dc breaker until that lights up.

 

[RCinFLA]

Combined core three phase transformer rely on magnetic field cancellation from other two phases, so cannot be used as three independent phase transformers, other than using it as an individual single phase transformer for series stacking.

The three-phase transformer shown in eBay ad is poor choice due to high leakage inductance core. Need full E-core per transformer for tight coupling with center core area twice the perimeter core areas. If you use it as an individual single phase with only center transformer used, its center core area is insufficient unless you reduce the VA level used.

You do not gain much going to four transformers over just using three series stacked transformers. You need to use three level voltages per transformer and ensure DC average across primary inverter AC output frequency maintains net zero bias offset on each transformer.

Just using two voltage levels per transformer is very inefficient.

 

[Warpspeed]

You do not gain much going to four transformers over just using three series stacked transformers. You need to use three level voltages per transformer and ensure DC average across primary inverter AC output frequency maintains net zero bias offset on each transformer.

True, three transformers (with three inverters) will work perfectly well for most practical purposes.
The lookup tables in ROM have 8 bits, three inverters would only require six bits of stored data, so using all eight bits requires nothing extra there.
The fourth inverter is always going to be of very low power, and really adds very little to cost and complexity. The choice is yours...

You are quite right about keeping the positive and negative waveforms across the primaries symmetrical, to prevent flux walking, and drifting up towards potential magnetic saturation in one direction.
Waveform symmetry can be maintained by only jumping between lookup tables after each full mains cycle and right at the zero crossing.
The positive and negative halves of each mains cycle must both come from the same lookup table to maintain this symmetry.
Then you can safely introduce a massive step correction in ac output voltage without introducing any net zero bias offset.

In the Warpverter, the dc input voltage is read, and an appropriate lookup table is selected at 25Hz (for a 50Hz inverter).
The application notes for the ICL7109 dual slope A/D converter say 30 conversions per second is the fastest recommended speed, so a 60 Hz Warpverter is possible using that A/D converter. It would be nice to do this at full inverter speed, but I already had the ICL7109 chips, so that is what I ended up using in the original prototype, and it works well enough not to bother with a redesign.

 

[Warpspeed]

Another Warpverter has just been born !
This one is the largest yet, and weighs in at 15Kw continuous, and should have a surge capacity of multiples of that.
The workmanship is absolutely superb, its a real show piece, and a real credit to Roger.
https://www.thebackshed.com/forum/ViewTopic.php?FID=4&TID=16626

 

[Nobodybusiness]

Another Warpverter has just been born !
This one is the largest yet, and weighs in at 15Kw continuous, and should have a surge capacity of multiples of that.
The workmanship is absolutely superb, its a real show piece, and a real credit to Roger.
https://www.thebackshed.com/forum/ViewTopic.php?FID=4&TID=16626

It’s definitely a big boy..

What do you think the surge capability is on this one?

Love the simplicity of these.

It’s the winding of the transformer that’s so daunting.

 

[upnorthandpersonal]

It’s the winding of the transformer that’s so daunting.

That, and I wish I could allocate more time for these kinds of projects in general. There's just so many interesting things out there...

 

[Warpspeed]

What do you think the surge capability is on this one?

Its fed from a huge fork lift battery, the transformers are so large that they will easily absorb massive short bursts of heat, so I imagine the mosfets would set the absolute limit. The largest inverter has four groups of six HY4008's, but the other inverters contribute something to that as well.

Difficult to say really.
Looking at the HY4008 data sheet it says 153 amps per device continuous at 100 C junction temperature.
https://datasheet4u.com/datasheet-pdf/HOOYI/HY4008/pdf.php?id=1300514
Each device in the bridge is on for less than 50% of the time in the largest inverter.
A rough conservative guess might be switching 6 x 150 amps should be well within their capability with a reasonable heat sink.
900 amps at 45v dc is just over 40Kw. It could be more than that.....

Roger is running a 63 amp thermal/magnetic circuit breaker (15 Kw) so that should open well before anything goes bang.

 

[Hedges]

It’s the winding of the transformer that’s so daunting.

I think you can repurpose an assortment of utility transformers. A number of different ratios are available. It looked to me like I could do it without any custom winding.

 

[Warpspeed]

I think you can repurpose an assortment of utility transformers. A number of different ratios are available. It looked to me like I could do it without any custom winding.

Large toroids seem to be much easier to source here in Australia than the U.S.
Rewinding large toroids has become a bit of a cottage industry here in Australia.
For us, there is really not much other choice.

Your American 110v/220v split supply means 110v transformer windings should be much more common in the larger size transformers than here in oZ.
Definitely agree that for you guys, a bit of hunting around scrap dealers, and a bit of innovation, might be very rewarding and save a lot of work.

It makes more sense at higher power levels, to run a 100v (nominal) battery, rather than a 48v battery.
That also encourages the use of monster IGBT power blocks which will be much more robust than a brace of smaller mosfets, especially at higher dc voltages.