High power DC to AC inverter without transformer

Discussion in 'General Electronics Chat' started by naizmenic, Apr 27, 2010.

Thread Status:
Not open for further replies.
  1. naizmenic

    Thread Starter New Member

    Apr 27, 2010
    12
    0
    20 lead automotive batteries (12V / 96Ah each) are connected in series forming 240V DC source.

    It is easy to create converter to 240 V square pulse AC source by full H-bridge using powerful MOSFETs. But how to create good 50Hz sine waveform? Suppose AC load could consumes up to 20A (approx. load of 5kW). Precise sine generator from a micro controller over DA converter and optocoupplers is not an issue. The main problem is which elements to use for
    energetic part. Some ideas with thyristors or IBGTs ?

    Any kind of idea?

    thanks
     
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Look at a Class D amplifier for ideas.

    You really don't want to use a "pure" sine wave output from a bridge, as power dissipation will be very high. Use a form of PWM with a low-pass filter on the output instead.
     
  3. kkazem

    Active Member

    Jul 23, 2009
    160
    26
    Hi,
    You came to the right guy. This is what I do for a living, I design power inverters in the 100 watt up to 10KW power level. In this case you have two good choices for H-Bridge switches: 1) Power MOSFETs, and 2) IGBTs. Either would work well, and for the MOSFETs, you may need to parallel a few for each leg of the H-Bridge whereas for IGBTs, usually one will do for each of the 4 switches. You would not use a D/A converter as such, you would use PWM (Pulse-width modulation) from an analog chip or from a DSP. However, a 240 VDC source is not enough for transformerless operation. You need 240V*Sqrt[2] = 339V and that really isn't enough to compensate for drops in the circuit. You really need about 375 VDC to make a 240 VAC Sine wave from a PWM circuit. You'll also need an output LC filter to remove the switching frequency and the best choice is usually about 20 KHz to 25 KHz. This type of design requires the best breadboarding technique. Actually, I wouldn't make it without a PC Board for at least the most critical parts of the circuit, which are the gate drive and the power H-Bridge layout.

    Gate drive, for either a MOSFET or IGBT are both very similar and it requires the proper drive circuit. You can't just drive it from the logic or DSP output. If you reallly are set on building this, then I highly recommend the IR2110 half-bridge gate driver. For a full H-Bridge, you'll need two of them and a 12V to 15V DC supply to run the chips. It's critical to keep the layout as short as possible and you must use a series gate drive resistor between each Drive output from the 2110's and the gates that they drive (that's 4 equal valued resistors of between 10 Ohms and 50 Ohms each). These help prevent ringing in the drain (FET) or Collector (IGBT) circuit. Selecting the proper MOSFETs or IGBTs is also critical. Use a breakdown voltage of at least 500 VDC to 600 VDC. THe battery power supply must be well bypassed as close to each 1/2-bridge section as possible between the upper fet drain and the lower fet source or its equivalent for an IGBT. You'll need a fairly high value low esr aluminum electrolytic cap in parallel with a couple of polypropylene film caps of at least 500 VDC each. One at about 2.2uF and another at about 0.47uF, both in parallel with a 10,000uF to 20,000uF, 500VDC Low ESR aluminum elytic cap for bulk storage. The impedance coming out of the batteries is way too high to use directly with the H-Bridge without all the bypassing I just discussed above.

    Don't forget safety! It is critical to keep in mind that the potentials you'll be working with are very lethal and won't give you a second chance if you put your body across the 375VDC supply. Always use a fuse in series with the battery positive. For a 20A RMS output, your battery drain will be in the area of about 20ADC or a bit above, so use a 25A to 30A fast-blow fuse, or better yet, a superfast blow semiconductor type fuse. This may save you a lot of money on blown FETs and drivers, and may save your life as well. Be sure to put test points on your breadboard or PWB for easy hookup of scopes and DVMs. This is also a good safety practice.

    Be sure and read the manufacturer's datasheets completely for the FETs, IGBTs, Gate Driver ICs, etc and do not exceed their ratings. In fact, make sure you have at least a 20% derating below the breakdown voltage and below the max current ratings, the derating should be more like 50%.

    Good luck and reply if you have any specific questions for me.
    Regards,
    Kamran Kazem
     
  4. naizmenic

    Thread Starter New Member

    Apr 27, 2010
    12
    0
    Thanks for instant help but
    trapezoidal emulation of sine waveform is NOT applicable to this task. I didn't mention device should work like some kind of UPS for very expensive laboratory equipment to compensate main power loss. Real pure sine is required for proper work of measurement equipment otherwise this wouldn't be a question for this forum.
    Lab device is some kind of laser pump with vacuum tube connected to some measurement equipment. How it works, I don't know, I am not a physicist, what do they measure, I don't know. They call it "Temperature controlled impact on high saturated ionized gas in plasma gaps".

    However, devices work well when powered from main AC, but using PWM it works but gives useless (wrong) measurements.

    The only conclusion: PWM is not suitable for such equipment. Pure sine is required. More ideas?
     
  5. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    Sure, use your DC batteries to run a DC motor to an AC alternator.

    Doesn't get much easier than that. Pure sine wave.
     
  6. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    PWM with a filter gives you a pure sine wave, that's the whole idea.
     
  7. naizmenic

    Thread Starter New Member

    Apr 27, 2010
    12
    0
    Believe or not, they already do that. They use gasoline generator (aggregate) but it is not appropriate since lab has moved in the high school. It is loud and nasty. Maybe DC motor will be silent enough. Good thing is equipment works fine using generator. So it would be last solution if all other options failed. This with DC motor and AC generator is good but it is more mechanical than electrical solution. I'll keep it as a final solution if all other options failed. Is it possible there is no efficient way of making pure sine with such load?

    edit:
    to: kkazim, thanks a lot for a suggestion and very detail description. I'll have all that in mind when similar project come on. This one is very specific and PWM doesn't do the job.
     
    Last edited: Apr 27, 2010
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
  9. naizmenic

    Thread Starter New Member

    Apr 27, 2010
    12
    0
    Thanks, as I can see the key point is on output low pass filter.

    Q1: Can low pass filter be defined independently on load for 50Hz sine? Even for very inductive loads?

    Q2: Suppose micro controller is used for PWM generation and duty cycle pulse with resolution is from 1/256 of full width to full width. So 50% of duty cycle is 128/256 = 0.5 of full width. What frequency should has such signal? Should it be in range of few kHz or few hundred kHz?

    Q3: If minus pole of batteries is connected to the load too as common then using half bridge MOSFET circuit on plus pole controlled by PWM I can imagine low pass filter on output "smoothing" square pulses to aprox. sine. But such config doesn't have a negative pole. Sinus generated on output is always positive in respect to ground. From the point of view of "load" current will always has the same direction and that will not be adequate substitution for real AC source. Using full H-bridge (off course without common ground) will make polarity change on the load but I can not imagine low pass filter in such config. Can someone point me in some direction of such topic?

    thanks

    edit:
    thx for links, it is very valuable infos

    I found in it
    http://www.irf.com/product-info/audio/classdtutorial.pdf

    at page 9

    filter configs for full H-bridge with two inductance in series with a load and a capacitor in parallel to load but there is no calculation method for those elements of such filter.

    Help.
     
    Last edited: Apr 28, 2010
  10. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    You need to generate two supply voltages when using a half bridge. It's often done with a capacitive divider from your single supply, using the middle (half supply) as the output's 0 reference.
    50% duty cycle gives you zero output voltage because you equally apply + and -
    In an H-bridge the output voltage is this (D is duty cycle):
    V_{out} = V_{supply}(2D - 1)

    For a sinusoid you make the duty cycle sinusoidal with an offset:
    V_{out} = V_{supply} \cdot[2(0.5sin(wt) + 0.5) - 1]\\<br />
V_{out} = V_{supply} \cdot sin(wt)
     
  11. naizmenic

    Thread Starter New Member

    Apr 27, 2010
    12
    0
    Thanks a lot for formulas,
    full H bridge is preferable since it doesn't require twice as many batteries.
    With such formula it is easy to create values table in memory of uC and use it for duty cycle of PWM.

    That is ok.

    But how such high freq. output to filter to 50 Hz sine in full bridge topology? I saw some solution in pdf from links of SgtWookie but I must admit I don't understand how such filter works since polarity on load in full H bridge has changing.
    Two inductivities in series with a load and capacitor parallel to load.

    Can someone explain that, please.
     
  12. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    Look up values probably won't be good enough. You'll have the voltage changing with load especially if the load is varying.
    You'll probably want it to be regulated and for that you'll need feedback control.

    RLC filters don't care about polarity, it doesn't make any difference.
    It should be easy to filter because you're looking to get 50 Hz while the switching will be tens of kHz.
    Inductors reduce the high frequency switching current going through the load while the capacitor does the same thing by shunting some current away.
     
  13. naizmenic

    Thread Starter New Member

    Apr 27, 2010
    12
    0
    Thanks Ghar.
     
  14. kkazem

    Active Member

    Jul 23, 2009
    160
    26
    OK,
    There have been a lot of confusing and inaccurate statements made here. I'll try to clear them up.

    First and probably most important, naizmenic, is that the load of a bridge circuit, whether a full bridge or a half bridge, connects across the bridge outputs in the center. Think of it this way, if you have a battery and a DC motor, with the two battery wires in your hand (plus and minus), then connect the wires in the normal way, the motor will spin in the forward direction, but since you have a battery, which is only a single polarity, how would you ever make the motor spin backwards? Simple, you reverse the battery wires going to the motor and it spins backwards. You have just created a negative power supply from a positive one. This is exactly how a full (and half) bridge works. There are essentially 4 switches, or two half bridges. One half bridge has the top switch connecting to the battery plus and the bottom switch going to battery minus. The two switch outputs connect in the center. Now take one more of these and you have a full bridge. The load output connects to the two half-bridge center outputs. If you switch the upper left and lower right switches on, you get a positive voltage to the load, and if you switch the lower left and the upper right switches on, you get a negative voltage to the load, despite the fact that you only have one single polarity battery.
    You can use full bridges in either the linear mode or the switching mode (PWM). Since a bridge in the linear mode is quite inefficient, it is almost never used that way. The carrier frequency for an inverter to make 50Hz or 60Hz at 120 VAC or 240 VAC should be about 20KHZ to 25KHZ for maximum efficiency and to make it inaudible. You can use 5KHZ for the carrier if you want, but it will be annoyingly noisy and it will limit you to not-so-great sine-wave distortion of the output due to low resolution. In order to get the PWM signals to feed the H-bridge, one needs a sine wave reference, whether analog or digital. Often, a table in Flash memory or ROM is used to generate a digital sine wave with good resolution so that the inverter output will have low sine distortion. And yes, the LC filter will be largely independent of the power frequency, even with inductive loads, and this is so because the LC filter is filtering the 20KHZ carrier out and has next to no effect on the 50 or 60 Hz power frequency.
    And Yes, this type of PWM solution is in my very own designs of 5KVA 28VDC to 240 VAC, 60Hz, low-distortion sine wave power. I sell thousands of them to the U.S. military for use in the field in their 28VDC vehicles. The quality of the 60Hz AC Power my inverters provide is substantially better than what the utility company provides coming out of the wall. My distortion is less than 2%, it can handle power factor loads of +1 to 0 to -1, it has overload protection, but still allows the user to draw a 3 to 4 times power overload for motor starting, so my 5KVA inverter can for 5 to 30 seconds output 20KVA without damage. These types of inverters are being manufactured not just by my firm, but by hundreds of firms around the world in power ratings from a few watts up to several Megawatts. Try reading the links the others have provided you and do a google search on H-Bridges and not how they work. It does not matter if the switches used are manual switches, relays, SCRs, BJTs, MOSFETs, or IGBTs. In fact, in cars in the 1940's and 1950's the only way they could make car radios work was with vacuum tubes, which require plate voltages in the hundreds of volts. In order to get that, they needed AC and a transformer to step up the voltage. They had none of our fancy modern semiconductor switches and they used something called a vibrator, which was a mechanical device that switched at a relatively low frequency like 60Hz, but was of course a square wave output. But the theory of how a half or full bridge circuit works is the same no matter what you use for the switch!

    Regards,
    Kamran Kazem

     
  15. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    ......or you can use a DC motor to turn an AC alternator ;)
     
  16. id_ruben

    New Member

    Oct 27, 2009
    7
    1
    How to calculate voltage output class d.
    Is this scheme can produce 220V AC, and how to modify them to be reliable and safe inverter. .
    [​IMG]
    Class D

    How to modification this "inductance heater" becomes Inverter
    [​IMG]
    Modification Inductance Heater becomes Inverter

    thank you
     
  17. crutschow

    Expert

    Mar 14, 2008
    13,033
    3,241
    The output across a full bridge with a single supply will be a plus and minus sinewave.
     
  18. demetrios

    New Member

    Jan 21, 2013
    1
    0
    Dear Kazem, I am new in designing an inverter but I can follow a schematic/circuit diagram and build one. Would you be kind to send me a circuit diagram or a link to follow or a pdf file attachment to my email address which is demetrios@nevadatechnology.co.uk. I thank you and greatly appreciate your effort, 5KW inverter would the right choice I believe kind regards demetrios



     
  19. BillB3857

    Senior Member

    Feb 28, 2009
    2,400
    348
    You do realize that it will probably cost more to build than to buy a commercially available unit. Development cost to re-invent the wheel is quite high. Also, the component costs on a unit basis for a single build are much higher than bulk costs used in mass manufacturing.
     
  20. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,804
    1,105
    Welcome to the forum!
    The SKKT92B and MCC95 have typical tq values of 100uS and 185uS respectively, according to their datasheets. Doesn't that make them much too slow for this purpose? Commutating them at more than a few hundred Hz could be problematic, so a 'pure sinewave' would be unrealistic IMO.

    Edit: BTW, it is not good practice to resurrect a dead thread: better to start a shiny new one of your own. Perhaps the mods can move your post accordingly.
     
    Last edited: Dec 7, 2014
Loading...
Thread Status:
Not open for further replies.