 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Electronics : Warpspeed’s low frequency stepped sine
Page 1 of 6   |
|||||
| Author | Message | ||||
LadyN Guru  Joined: 26/01/2019 Location: United StatesPosts: 408 |
Warpspeed's low frequency stepped sinewave inverter concepts In the post from 2017 March 16 at 4:02am at https://www.thebackshed.com/forum/forum_posts.asp?TID=9380&PN=1, Warpspeed mentioned a method to synthesize an output sinewave by injecting harmonics at their proper amplitudes. The way this method works is not clear to me (the physics of it) and I also want to build a prototype based off this idea: this thread is to document the journey. Warpspeed, if I were to build a 120V AC inverter using your idea, and feed the output in 4 stages, assuming the input stage supplies +Vin volts, do I: 1. Use a HBridge to generate + and -Vin volts 2. Use a set of 4 switches to output +, 0 and -Vin volts 3. Use one transformer per stage, where: a. the "main" transformer inputs Vin volts to, say 210V output b. the "second" transformer inputs Vin volts to, say (210/2)V output c. the "third" transformer inputs Vin volts to, say (210/3)V output d. the "fourth" transformer inputs Vin volts to, say (210/4)V output 4. Connect all the outputs of these transformers in series Did I get this right? |
||||
mackoffgrid Guru Joined: 13/03/2017 Location: AustraliaPosts: 460 |
Hi N Have a look at this diagram (from that thread) https://www.thebackshed.com/forum/uploads/mackoffgrid/2017-03-16_064055_cascadeInverter_1.pdf Cheers Andrew |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
mackoffgrid, thank you for the diagram, that was very very useful. Let me try and understand it: Vin, in your digram is 24V. I am assuming you have a buck or buck-boot convertor that takes a LiCo battery pack that has a 26V output and regulates that down to 24V? 1. Each HBridge generates +, 0 and -24 volts. Their timing is synchronized to get the final output we want 2. TR1 takes +, 0 and -24 volts and outputs 9x that: 216V. Since it's a 3kW, it can source 14A nominal 3. TR2 takes +, 0 and -24 volts and outputs 3x that: 72V. Since it's a 1kW, it can source 14A nominal 4. TR3 takes +, 0 and -24 volts and outputs 1x that: 24V. Since it's a 300W, it can source 13A nominal 5. TR4 takes +, 0 and -24 volts and outputs 1/3x that: 8V. Since it's a 100W, it can source 13A nominal The secondaries of all these transformers are connected in series so the wiring in them need to support atleast 14A (actually much higher to be able to support surge loads?) Do I have everything absolutely correct? |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Tried to post some waveform pictures but for some reason I cannot. Cheers, Tony. |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
imgur should work. Does it for you? |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
 Cheers, Tony. |
||||
| mackoffgrid Guru Joined: 13/03/2017 Location: AustraliaPosts: 460 |
Hi N I did that diagram when I was getting my head around what Tony was doing. That 24v was just a nominal voltage and in fact in my case its about 26.5 - 27v LiFePO4 battery bank. For your interest LiFePO4 160Ah. 2 cells in parallel (320Ah) by 8 in series. I built my own solar controller which charges the batteries directly from 10kW solar array. Spot on  Hopefully Tony will get his images going for you. Cheers Andrew |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
Thank you Tony for these images. Thank you Tony for these images: my questions are below |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
I will need help with: 1. Understanding the physics behind this magic. Is there a software tool or a software library into which I can feed in a set of inputs and see the sample output? 2. How do I source these transformers? To make sourcing these transformers easier, what are some input and output specifications I can look for to maximize my chances of finding a suitable set of these transformers? They don't have to be powerful to being for now. I just need something I can use for a build so I SEE IT IN ACTION and try and understand the physics and play with it 3. How do I go about designing the EMC filter? Same questions as above regarding sourcing these components It's becoming apparent to me that I need to get an oscilloscope to help me go forward with this. I created a new thread about this |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Thank you Natasha for posting my waveforms and explaining that using the # hashtag as a file name for HTTP is a definite no no. My file names were "Inverter #1.JPG" for example, and that ignorance is what brought me undone. Choice of exact transformer voltage ratios primary/secondary is made a lot simpler if whole integer volts are used on both sides, to work out the turns ratios when designing the transformers. If you add 225v + 75v +25v + 8.33v you get 333.33 volts peak which is an easy number to remember. That produces 235.7 volts rms which should be a workable starting point in most parts of the world. If voltage regulation is added later on, tweaking the exact output voltage can then be done with a potentiometer. The one big disadvantage of this type of inverter is the requirement for four special transformers, which you will probably have to wind yourself. Its the cost of the transformers that has also kept this method from becoming popular commercially, although there have been some on the market from time to time. Its not my invention, just old technology revisited. At very high power levels say several tens of Kw up to Megawatt power levels, its really the only practical way to turn dc into ac electronically without rotating machinery. For us, PWM has more attraction because its simpler and uses fewer parts. But switching at 50Hz is FIVE HUNDRED times slower than switching at 25Khz, and many of the noise and physical layout problems make PWM a much more difficult challenge for the novice home constructor, especially at the high power levels. It can be done, but its not as easy as many people expect, and it gets more difficult as the power level and switching frequency rises. Anyhow, the easiest approach to generating the stepped waveforms is from a simple lookup table. A lookup table with 1,024 addresses can have a time resolution of 20uS with a 20mS (50Hz) output sine wave which is plenty. Each inverter requires two data bits which just switch on and off as required, in the correct switching sequence. For instance, the positive peak would be all inverters switch on +ve in the same direction and the data to do that could be 1010 1010 for four inverters. All inverters off at the zero crossing would be data 0000 0000. The negative peak would be data 0101 0101. Now to generate a normal sine wave lookup table we usually wish to know the amplitude, and we already know that the angle (lookup table address) advances constantly. This is different, we know the step amplitudes which correspond to the actual dc input voltage to the inverter. What we wish to find are the angles where those amplitudes are crossed, and switch as close as possible while introducing minimal error into the process. That enables us to program our on and off times with 20uS time resolution. To do that we use an arcsin table to convert amplitude to degrees or radians. There are arcsin calculators available on the internet, and you only need to work out values for the first ninety degrees (256 lookup table values). One lookup table will then give you a working inverter which is a very big step forward. To regulate the output voltage, we can use multiple lookup tables and switch between those at the zero crossing point. If we make every step very slightly narrower, that would reduce the ac output amplitude very slightly with very fine control of amplitude for any given dc input voltage powering the inverter. By having multiple lookup tables, we can very smoothly crank the output voltage up and down seamlessly. My own inverter has 256 different lookup tables, each 1K with a 2:1 dc input voltage range. I am sure it could all be done with a microprocessor in real time, but it all becomes very time critical. I overcame the time criticality by using a dual port ram, but it all became unnecessarily large and complex. A bunch of lookup tables in an EEPROM is fast, simple, and reasonably easy to get going. The LSBs are continuously clocked sequentially by a counter. The high order MSBs are selected by an analog to digital converter. The analog input voltage just scrolls through the lookup tables to adjust output amplitude. I started out with a microcontroller and eventually gave up in doing it all in real time. I just added an EEPROM burning sub routine to the same program and let it loose filling in all of the 256 lookup tables, which took a couple of minutes. This is just a fairly quick overview of how its all supposed to work. Cheers, Tony. |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
My pleasure! Ah, that seems to be an undertaking that I will have to read up a lot on. Are there some guides (i know clockman has a good one for toroids) you know of that I can follow? How much would those 4 transformers cost and how much would they typically weigh each? I am really interested in reading more |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
The transformer drama really revolves around a tradeoff between dollars and sheer cunning and determination....... My situation is slightly different, in that I once had a transformer winding business that went broke. I gave my coil winding machine and my large industrial oven and a lot of other stuff free ( I am hopeless at business) to a very good friend that was also in the same line of business. So for around forty years I can just front up to his door and either get freebies, or get brand new materials at tax free trade price. The grain oriented silicon steel for the transformer cores, and the copper wire both cost about the same, trade price or about $18 Australian per kilogram. That's about roughly $13.50 US per kilogram, for just the materials. So building multi kilowatt inverters with 50Kg++ transformers can get rather expensive if you have to buy the bits, and even more so if you get a professional to design it, and then professionally build the actual transformer. And he has to make a profit to pay his mortgage, his factory rent, and feed his family on top of that. So we come to the do it yourself Back Shed method, of scrounging other people's rubbish, and with great heroism winding your own transformers out of recycled crap. This can be a very rewarding and character building exercise, and highly recommended. Myself I am 70 years old, been designing and winding bloody transfomers for years, and cannot be bothered these days. My very eager mate has all the good gear, and can do it in a tenth of the time, so I just said "Hey Paul" this is what I want, how much ? Its not cheap but Paul will do it for me at cost price. So if you are poor like me, you probably have to decide what it is you are trying to do, scrounge and recycle all the bits, and have hours of fun and aching hands and muscles winding your own transformers. Just about everyone here does it that way. Cheers, Tony. |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
Those numbers were very useful. OK, so questions: 1. the cost of materials to build a single (but HUGE) toroid based inverter like the OzInverter would be far less than the cost of materials for this design (TonyInverter)? 2. If so, what's your estimated differential? 2x as much? 3x as much (3x means OzInverter would be 3x cheaper than TonyInverter)? 3. Could we design the TonyInverter using toroid cores instead of EI cores? I want to get started with the TonyInverter and building transformers are quite an undertaking: I don't like their combined complexities at all. Figuring out one or more source of (almost) ready to go transformers (maybe I need to rewind the secondary) would possibly reduce the starting complexity. I don't expect to design a high power TonyInverter at the beginning and I am perfectly fine with ready to go transformers dictating the TonyInverter specifications if that helps me build the first one. Do we have a subforum here to ask about transformer building? Is there another forum out there that kinda focuses on transformer building? |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Many ways to approach this. The usual "Forum way" would be to obtain a large toroidal transformer usually from inside a trashed grid tie inverter off the rubbish heap, or bought from a scrap metal dealer. Or maybe bought very cheaply from e-bay as not working. There are a lot of blown up or unloved grid tie inverters out there, and its usually the electronics that go bang, the transformer being fine. You then get to unwind the transformer, recover the wire, and rebuild it to suit your own requirements. That is how most people here go about it. The first thing you need to do is have a plan and nail down some requirements regarding dc input voltage and required power level. Suppose you decided on 1Kw for the largest transformer. It would need a 225v secondary and most transformers out there will already have a 220v to 240v winding which if you are lucky is the inside winding (which is usually the case). The primary might already be 24v or something fairly close to what you need, and can have turns either added or removed. That could all be a pretty straightforward job. The second transformer will require a 75v secondary and a power rating of about a third that of the largest transformer, perhaps 350 watts. Again you might get lucky and find a transformer with a 110v winding which would be fine to produce 75v. Just add more primary turns to get the primary/secondary ratio right. Again if the primary is on the outside that should be easy. Third transformer might be 24v to 25v or something similar at about 120 watts. It will be small and should not require many turns on either primary or secondary. That could be wound from scratch either with a toroid or E and I laminations. Ditto for the fourth transformer, maybe about 30 to 50 watts size. See what you can find first for the largest transformer, and start with that. With a bit of luck and some patience it could probably all be done for just about nothing. The infamous "modified sine wave" inverters just use one transformer, and even a two transformer stepped waveform with nine steps is perfectly usable and functional. I ran a three inverter version that has 27 steps for over a year and it ran everything in the house just fine. Cheers, Tony. |
||||
| mackoffgrid Guru Joined: 13/03/2017 Location: AustraliaPosts: 460 |
Natasha, I am building a TonyInverter, TonyVerter, WarpVerter, I been calling it a cascade Inverter. It is certainly more expensive to build than an Ozinverter. Tony's inverter just speaks to me and I wish to be able upscale a bit on the power side (some of the Ozinverter certainly handle High power). Make sure you realize most of the inverter builds on this forum are for 240Vac single phase. Regarding costs, it is definitely more expensive, mostly labour, and electronics which is getting cheaper all the time. "On Paper" There should be no difference in cost on transformer cores, for a given kWatt rating and flux density. The net kg of cores should be the same for the Ozinverter and Tony's inverter. There should be little difference in Copper Costs, The net Kg of copper should be similar, as all windings are contributing to total power etc. There should be little difference in FETs (or IGBT) because they still need to accumulate the overall primary current requirement. There is a bit more control electronics, a lot more optodrivers and isolated power supplies which are all coming down in price. There is more PCB cm2 as well, which is also coming down real cheap. In reality The devil is in the detail as they say, and if you had to buy the transformers then it would likely cost a lot more. You probably will use more copper, you probably will use a few more fets, and there is a bit more wiring. The Ozinverter is pretty much optimized cost wise. I have certainly not optimized what I've been doing nor have I even tried to keep an account of cost. The big ticket items will always "tend" to come down to the transformers and copper. Cheers Andrew |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Andrew, Did you get that fourth transformer going ? Cheers, Tony. |
||||
| mackoffgrid Guru Joined: 13/03/2017 Location: AustraliaPosts: 460 |
Gday Tony. Not yet.  When I proved the 3 transformers on the bench, it was with no heatsinks or anything. Having proved the 3 transformers was enough to show me I had the electronics right and my understanding was correct. So now I've gone straight to doing the difficult stuff  Getting all the physical hardware going. Cutting heatsinks, Aluminium angle, drilling and tapping threads.Unfortunately we're fair weather sailors, that is to say that if its too hot we don't go down to the shed :-) And find distractions in the air conditioning. I would say we are a week (manana) away from running the whole thing up. It won't be in its final enclosure, but a sub-chassis that holds all the heatsinks, H-bridge PCBs, and negative bus-bar together. Cheers Andrew |
||||
| isochronic Guru Joined: 21/01/2012 Location: AustraliaPosts: 689 |
I am wondering about trying this with old microwave oven transformers. They would need the old secondary and magnetic shunt removed of course. The old primary could be used as the output (secondary), a new primary should be easy enough to wind as it will only have a few dozen turns. I guess say 700 watts from the first transformer, and so on, to a total between 1.5 and 2 kW output ? A fair bit of effort but really low cost (and a good way to recycle)... Maybe two could be wound in parallel as one, to increase power on the first stage (?) I don't know how efficient they are though, maybe they'd waste too much power. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Microwave oven transformers usually run at extremely high flux density and often have aluminium windings. The laminations are commonly welded up, so its not really possible to dismantle the transformer. Microwave oven transformers while pretty easy to find, are not really ideal material to work with. They might have possibilities for the smallest transformer, if run at much reduced power. The voltages on both primary and secondary would both be low requiring few turns so winding it would not be too frustrating. Once anyone has some definite plans and some transformer parts to play around with, I can provide some design guidance if required. Probably the best way to to begin is to cobble together any old junk you can find, get that working, and concentrate on getting the drive waveforms and the electronics sorted out. By that stage you will have some experience and some definite ideas on where to go from there. If you wish to start by playing around with just some ideas in software, you could make a "virtual" inverter without needing any mosfets or transformers. If you generate all the required on/off gate driver waveforms, feed those into a 256 x 256 bit lookup table to convert the gate drives into numeric values. That can go into a digital to analog converter and the output will be a beautiful stepped waveform. The whole thing can be put together on a breadboard and you can view the analog output on an oscilloscope. That is much easier than building four bridge inverters and winding four transformers. For me, the most difficult part of all this was generating the switching waveforms in software. I am not too good at that type of thing, but for the software guys it should all be a lot more straightforward. Cheers, Tony. |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
Exactly. Microwaves were the first thing I thought of and had my brother take it apart, only to get excited by looking at the transformer and quickly disappointed that they were welded. Why do they do that? To further reduce eddy current? That will definitely be me. Let me create a new thread where I basically brainstorm with you and others here what I see available and how much that is of use: https://www.thebackshed.com/forum/forum_posts.asp?TID=11118 Excellent advice! makes complete sense! Yes! Let's take that one step further: simulate the whole thing. I initially wanted to model the DAC in software but quickly realized SPICE would probably be a better way because they have already done that. I will create a new thread for that and let everyone know for this as well. |
||||
| Page 1 of 6 |
|||||
 Print this page |
