Hi All

I have to design (and build) a UPS with 3kW power output, and I have some constraints that don't allow me to simply buy one.

1. Cost. UPS-es with such a power output are usually pretty expensive and I can't afford buying one (plus I'm itching to build it myself )

2. Square wave. The UPS will have to drive some induction motors and I believe they don't handle square wave very well, because of spikes and harmonics that occur if an induction motor is ran on square wave power.

I actually tried to run a small induction motor on a square wave UPS. It had a fraction of it's nominal power and soon I felt a smell of burned insulation, It kept running though, so I can't be sure if that was the cause.

So I have to design my inverter to output real sine wave or at least not very rough step-sine wave.

3. Big transformer. As you might imagine, a 3kW, 50Hz transformer would be quite big and heavy. I'd like the device to be as small as possible and light too.

A particular design is "rolling" in my mind, so I thought I could use a third-party opinion before I throw the effort to build it...

This is how it goes:

1. The 12 volts form the battery are fed into a powerful DC to DC converter, which outputs 300V DC with maximum power output of > 3kW.

2. Those 300 volts, power a class D amplifier, which outputs 220V, 50Hz, ~3kW.

3. In order for the amplifier to to output a 50Hz sine wave voltage, it uses a simple sine wave generator for the input signal.

That's it. Easy to explain.

My question is, how feasible do you think that design is?

Here are my pros and cons.

cons:

1. It's complex - lots of points of failure.

2. Possible reliability issues.

3.Possible reduced efficiency; even though switch mode DC to DC converters and class D amplifiers are very efficient, chaining two in a row might have it's consequences efficiency wise... What you you think?

4. A class D amplifiers are designed to drive speakers which have a constant impedance. I this case the load could change rapidly, and there could be spikes and other fluctuations with possible adverse effects. Although I'm not terribly concerned sine class D amplifiers have and LC filter at the output that would have some protective effect aside of it's main function.

pros:

1. Relatively low cost. That depends of course, but I think it'll be cheaper than a huge near-solid chunk of metal.

2. low mass and size. The DC to DC converter is going to run at 60kHz or 100kHz. So I could use a small ferrite transformer. (toroidal maybe) Certainly it'll be much smaller than a 50Hz transformer... And that of course meets one of my main requirements.

3. (Near)real sine wave. The amplifier doesn't have to be audio-equipment quality, it could be a crude one. As far as I know usually they run at 400kHz, mine could run at as low as 30kHz or even lower. And that covers the last of my requirements.

So that's it. What do you think of that design?
Does it exist by the way? I haven't seen such a design. (I hope I don't show myself as too ignorant )

Thanks in advance

 

Let us start with the basic fact that you power output for the first stage of the process is 3 kW. The power out of a DC DC converter will always be less, sometimes much less, than the power in. Let us assume for the sake of argument that you could design a process that was 80% efficient. The required power in would b 3 kW divided by 80% or 3.75 kW. Such a power level would require an eye popping 312.5 amperes of current. If you know anything about battery capacity you will recognize that most standard automotive and marine units would become completely discharged in a matter of minutes. The reason UPS costs are high is that they are not trivial to design and build. This is not a good first project for someone with limited design, fabrication, layout, and debugging skills.

Without any a priori knowledge of your knowledge and skills, just by the very fact that you came looking for help, suggests that you should not expect a rapid and successful path to your goal. You did not specify a time horizon for this project, but I'm going to suggest two years. That amount of time should be sufficient to learn what you need to learn, research the components, acquire the materials, and construct the unit. You might also decide after some investigation that the project is beyond your abilities.

BTW - a quick internet search reveals units for $300 that claim to provide 3kW from 12 V. I can't speak to the accuracy of the claims, but that is really not expensive compared to the alternative.

 

I agree with Papabravo.
This is not a project to embark on without much experience. You will find a lot of smoke!
If you want a 3KW inverter, don't start from 12V as the current will be too high, along with the losses. Maybe 48V?
It is not a trivial job to do this and I'd be looking for other ways to do it. You haven't really explained the application and if it is motors, you may be able to use DC motors running on a battery that is float charged from the mains.

As for the inverter, a battery to 355V DC supply then a high powered H bridge to generate the AC will work. A lot are run that way but there is a great deal of extra circuity needed to keep the Magic Smoke in.

If you want to go this route, start with a 100W version and remember to wear safety specks! I have had experience with launching smoking debris across the room from over stresses motor controls. High power = quite a bit if trickyness.

 

Thanks.
OK maybe I got too enthusiastic about that project. And yes it was silly of me to list 12 volts as input. Such a UPS would need banks of batteries, connected in such a way that the input voltage is a multiple of 12. As high as possible as long as the internal resistance of the batteries doesn't become a problem.

I do have some very limited experience making SMPS but nothing near that scale. Of course I'd start small if I ever start that project at all.

The problem is that the machinery is already there. It has several motors and heaters which turn on and off at different times. The power requirement only occasionally rises to 3kW and for only few minutes so the power source does have to have that kind power output. Maybe the best I can do is to split the different internal modules and power them with several less powerful UPSes, that however doesn't solve the square wave problem.

OK thanks for giving me a dose of reality. I'll probably think of something else.

 

Have a look at...
www.egmicro.com/download/EGS002_manual_en.pdf
http://www.ebay.com.au/itm/DC-AC-Pu...422854?hash=item2cb3c26646:g:3qoAAOSwZJBX~0Wn