AC/DC Inverters

Discussion in 'General Electronics Chat' started by Wendy, Jul 17, 2010.

  1. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    I would be the first to say this is a case where it is easier and cheaper to buy a commercial inverter. It is hard to beat a design that has been professionally designed and built with safety and specs in mind. However, we've been getting a lot of this schematic.

    http://www.sentex.ca/~mec1995/circ/555dcac.html

    [​IMG]
    ..................................................................Figure 1

    This circuit bugs me. It has so many flaws that I'm not going to bother going over them in this paragraph, except many of the posts at All About Circuits sum up to "Why doesn't this work?". It doesn't help it was stolen and posted here. I'll post several ideas that will work to various degrees and discuss their advantages and disadvantages, with the proviso that the first paragraph states how I feel about this subject. I am writing this to give folks at AAC a place to point the many inverter requests we have gotten over the long haul.

    Feel free to trash my ideas as I come up with them. If I like the criticism I will add it to the tutorial. This is both a how to and why you shouldn't kinda article. I would not use a home made inverter normally, it would take a major disaster such a plague or meteor strike to make me do it. Therefore, I am not responsible for any of the circuits I am about to present or your use of them! If you use this article, you do so at your own risk!

    Another common theme from posters is a lack of real understanding of waveforms, frequency, RMS, P-P, and power. All of these are critical, so I will go through them.


    • Waveforms - Most simple switching circuits generate square waves. The signal sent to you from the power company is a sine wave. Sine waves are pure, they do not generate much in the way of RF, while square waves aren't, and can spatter the spectrum. If you have a piece of equipment that uses RF (stereos, TVs, satellite receivers, any wireless gear) this can be a real nuisance. It also has major ramifications in voltages, which will be discussed in the RMS and P-P paragraph.

    • Frequency - We take the AC delivered by the power company for granted, but they pay close attention to the accuracy of the frequency of their power. In many cases they count the number of cycles in a day, and subtly adjust the frequency to make it come out right. This is because many pieces of equipment, such as digital clocks, use this as their time bases. Back in the days of TVs using tubes it was also used to make the internal oscillators of the TV stay on track, as this simplified the circuits considerably.

    • RMS - This is the amount of real power (how much heat a resistor dissipates) a waveform will deliver. It the case of a sine wave the power is X 0.707 the peak value of the wave form. In the case of a square wave it is X 1 the value. Problem is, many power supply circuits use the peak value, so while a square wave may deliver the same amount of power you may be feeding a piece of equipment some seriously wrong voltages. To make it worse, a digital volt meter (DVM) or a volt ohm meter (VOM) will not be able to tell the difference on the AC scale!

    • P-P - This closely related to RMS. Many people confuse Peak to Peak with RMS. Generally P-P is dictated by the battery voltage. Another number, the peak voltage, is ½ the P-P value.

    • Power - The power delivered by an inverter is the same as the power input by the inverter, plus losses. Losses are a fact of life, they don't go away. If a circuit has a 20% loss it is probably doing pretty good. So lets say you want to power a 120W light bulb at 120VAC. This means the bulb will use 1 amp (I = P / V). If you have a 12V power supply, and it is feeding a 120W, then you are taking in 144W (remember the 20% loss?). This means your battery is providing 12 Amps! It only gets worse as the wattages go up. Battery current is a common problem with all inverters. There is also the fact the more amperage you want the larger and more expensive the transformer needs to be.
    If you have problems understanding any of these concepts please read the AAC book, as they are all important.

    It is also important to understand AC wiring. These vary from country to country, both in power specs and safety specs. I will assume the US standards, 120VAC 60Hz with a ground. The ground requirement is fairly universal as far as I know, so I will show how to do it on a portable inverter. It is up to the builder to make it safe. Voltages and currents in these ranges are LETHAL! Please do not take shortcuts as far as the AC output is concerned. Build them as close to code as you can.

    Now for the first schematic:

    [​IMG]
    ........................................................Figure 2

    This schematic is old, when I started studying electronics over 30 years ago it was around. It has big problems with voltage output stability, frequency stability, duty cycle, and in producing a square wave. While it might work on a large number of applications the only thing I would trust it with is light bulbs. When I first drew it I used single transistors, but after a little reflection I realized they would not have enough gain, so I switched them off for Darlington Pairs, which come in prepackaged formats. Something like a TIP-100 or an equivalent would work for this. They will get quite hot, so heat sink them properly. They will also cause about 12 V drop on the AC output compared to conventional single transistors. The actual output voltage is dependent on both the components and the power supply voltage.

    The frequency output is only an approximation, if you need to tweak them add two identical resistors in series with R1 and R2 (one for each resistor) to drop the frequency. Conventional power diodes (such as the 1N4001) will work for all the diodes. Do not use resistors lower than 120Ω, as they will get much hotter and possibly burn up. Use 150Ω for R1 and R2 for 50Hz.

    You may run into the problem of "flux walking". This happens when the duty cycles of the two transistors are not exactly the same. Current flow is imbalanced, and the transformer core saturates. This causes extremely high current flow in the transistors, resulting in smoke.

    The Figure 3 is based off of Figure 1, but it has been cleaned up as much as I was able. I left L1 to minimize the RF interference produced by the square wave. It will not help much, but it is better than nothing. Neither Figure 1 or Figure 3 will produce much power. If you get 10 watts out of either designs I would be surprised. Figure 1 has no protections for the transistors, either they or the transformer will blow if pushed beyond their limits. Figure 1 also looses more than 2.4VDC with all the BE drops (over 20% of the drive voltage), which my design corrects. C4 is another source of loss, it is 1Ω at 60Hz and 1.2Ω at 50Hz, but while it adds to the loss it is necessary. It will loose 24 volts of the AC output for a max of 96 VAC out, when C4 is taken into account it is even more. The symmetry was a bit further off than I liked too (just over 52%), so I tweaked it to around 50.2%. Basically both are little more than science fair projects, showing how it is done without being able to do much.

    [​IMG]
    .......................................................................................Figure 3

    So far we have used BJT type transistors. They are common, they are cheap, but they have problems with high power. They generally use 10% of the Collector current for the Base. It you have 10 amps through the collector (which will light a light bulb) you will have 1A going though the base. This is a limiter for how much current you have going through the transformer, as well as being very wasteful and generates extra heat. Technology has solved this problem in the form of a MOSFET.

    MOSFETs are voltage controlled devices, and when they are on they have very low resistances. Their only drawback is they take a substantial amount of current to force the gate to change in the form of a surge. The gate looks like a capacitor, after it is charged it takes almost no current to keep it there, for all intents and purposes no current. I have substituted the 555 with a CMOS 555 because a CMOS 555 will go rail to rail, unlike a conventional 555.

    [​IMG]
    .......................................................................................Figure 4

    Generally Push Pull outputs like these have to have something to prevent shoot through, a condition where both transistors Q1 and Q2 are on at the same time during a transition. If this happens long enough one or both the transistors may fry, a bad thing. Figure 3 uses the zeners to allow one transistor turn off before the other turns on, with this design (Figure 4) the inductors L2 and L3 slow Q1 and Q2 turning on while the other turns off.
     
    Last edited: Jan 4, 2011
    crazyengineer likes this.
  2. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    There are many ways to do the same things in electronics, my solution using the coils is not the best out there. I'm trying to keep the parts count down and stick with common components. I think it will work, but a better approach would be to use one of the many PWM MOSFET controllers (such as a UC3824).

    In general, using capacitor such as C4 (C5,C6, and C7) shown in Figure 4 is not a good idea. The reason I added capacitors is with 10 amps flowing through them they are under a lot of stress. Another variation that doesn't use these capacitors, one that is a lot more common, is a true H Bridge output feeding a transformer. It eliminates C4 and its ilk completely, and would look a lot like Figure 5. U2 in this case is a simple inverter/MOSFET driver.

    [​IMG]
    .......................................................................................Figure 5

    These designs are not tested, so while I think they would work there may be bugs with them. If anyone tests them please let me know the results.

    All of the designs shown so far are square wave inverters, probably the worst designs there are to power sensitive electronics. For precision you need a sine wave type inverter. Before I show a possible schematic for a design like this lets go over some problems you might have.

    First, sine waves are analog. What we've shown up to now is digital, a square wave. A square wave RMS is 1.0 X the peak voltage of the square wave. Figure 5 doubles that, but it is still digital in form.

    You might think you can use a simple audio amplifier to feed the transformer. This would almost work, except the part of the waveform is analog, and will cause extreme inefficiency and heating. Since this is meant for portable power it does no good to loose half your power inside the box, cooking the transistors and other components, while the load only gets what is left over.

    Fortunately there is a digital amplifier design that will work well for this problem. It is called a Class D amplifier, which uses Pulse Width Modulation, and it will create a complex waveform simply by switching a transistor on/off at the right times.

    So now we have an added requirement. We need a sine wave generator that has a stable frequency and a stable peak to peak signal. I'll go over some possible designs of sine wave oscillators after I show a inverter that could use a sine wave.
     
    Last edited: Jan 4, 2011
  3. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    Stash point #3
     
  4. timrobbins

    Active Member

    Aug 29, 2009
    318
    16
    "It has so many flaws that I'm not going to bother going over them, except many of the posts at All About Circuits sum up to "Why doesn't this work?". "

    Maybe it would be better to list the flaws, and review the initial design based on your dot-points, and go through a possible solution for each flaw (or a summary of why a solution for the flaw is not appropriate). People might then get a better insight. I don't think adding alternative designs is a focussed means to review a particular design (eg. the initial design) and then discount it and possibly jump from the frypan into the fire with another circuit. To my mind there is no better way to develop an understanding in general than to assess a particular design in a detailed manner.

    Ciao, Tim
     
  5. tom66

    Senior Member

    May 9, 2009
    2,613
    213
    Why bother with such an inverter circuit anyway... you can pick up at decent 150W modified square wave inverter here for about £30, that's better than the time you spend making one.
     
  6. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    tom6: You are singing to the choir on this one. People think it is cheaper and easier to DIY, when quite the opposite is true. Plus the fact these circuits can be dangerous, and it seems newbies are the ones who want to do it without having the back ground or skill.

    timrobbins: I may go back and rewrite that area, this is a WIP. Mostly it is the losses, which are fairly unnecessary, that bug me. A circuit like this needs all the voltage it can get to the transformer. All the small losses Tony ignored aren't as trivial as he thinks, he's loosing almost 2 volts going to the transformer. And those losses add up as heat. There is also the fact his design can't power up a 100W light bulb. Neither can my version modified off his, but I acknowledge and explain why.

    Looking at Figure 1 I think I need to change the transformer to a 24:120V model.
     
  7. timrobbins

    Active Member

    Aug 29, 2009
    318
    16
    I think the value of discussing the chosen circuit is it's simplicity! That makes it a great vehicle to identify the issues at hand. Most people starting electronics cut their teeth on a 555. The half-H bridge PWM drive is easy to visually and electrically comprehend. I reckon it is a great tutorial workhorse - there are so many topics to probe. In comparison, I suggest your multivibrator circuit would be somewhat bamboozling for many starters.

    Ciao, Tim
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    There are several problems with both circuits.

    Bill, in the circuit you posted, you'll lose roughly 1.2v drop across the Darlingtons
    ' collectors.

    You may run into the problem of "flux walking". This happens when the duty cycles of the two transistors are not exactly the same. Current flow is imbalanced, and the transformer core saturates. This causes extremely high current flow in the transistors, resulting in smoke.
     
  9. timrobbins

    Active Member

    Aug 29, 2009
    318
    16
    Strike one for van Roon!
     
  10. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    I know, it is a problem that can't really be fixed with the design without good test equipment (mostly an oscope). You'll note I mentioned the drop, "They will also cause about 6 V drop on the AC output". If you don't mind I'll lift your comment verbatum and put it in. It was how it was done at during the 60's though, and for the power levels I think darlingtons are absolutely necessary.

    I fixed this in Figure 3, and mentioned the symmetry problem in that paragraph (not quite as you did).

    I could use some help with shoot through prevention in my next design, which will be Figure 4. I haven't started drawing it yet, it is next.

    A common theme with all of these is "don't like it, wouldn't do it", but since we can't discourage these guys I'll show them what it looks like and talk about their disadvantages.

    You miss the critical point, it doesn't work! It can't work, it will not work, and only with major modifications will it work. How many appliances use 10 watts or less? I'm trying to teach people why, what, and how. The folks come in without a clue and expect it to power Lord knows what. I can try to teach folks to think, the rest is up to them.
     
    Last edited: Jul 19, 2010
  11. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Actually, it'll be worse than that. Don't forget, you only have 12v in, 12v-1.2v=10.8v; that's a 10% loss right off the bat. So, instead of 120v out you'd get 108v out - and it would be a square wave instead of a nice sine wave.

    Be my guest. ;)
    But nowadays we have power MOSFETs! Those 3-legged critters can have such a low Rds(on) that power loss is quite minimal.

    OK.

    I have a rough idea for a design that could actually be viable. ;) It involves using a standard switching regulator like a UC3825 driving a pair of MOSFETs that alternately sink current from the center-tapped primary toroidal transformer, a sinusoidal voltage reference generated by a small PIC uC with a look-up table, and use a low-pass filter on the secondary winding. However, I'm in no hurry to build such a thing.
     
  12. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    Since this is for the international types, I'm going to try to keep it as simple as possible for parts availability. I'll change the numbers mentioned in Figure 2 for the Darlington caused drops. I may just make two different circuits, one with crossover compensation, one without, and mention the problems.

    How common are these UC3824s?
     
    Last edited: Jul 19, 2010
  13. timrobbins

    Active Member

    Aug 29, 2009
    318
    16
    Bill,

    "You miss the critical point, it doesn't work! It can't work, it will not work, and only with major modifications will it work. How many appliances use 10 watts or less?"

    With the right Tx and the right load and the right duty cycle - the output will deliver the required voltage - it will work! The benefit of this thread is to allow people to understand where it falls over and why. The thread appears to be about generic circuit design - so a good idea is start with the why's and wherefore's in relation to well known circuits. An understanding of why the output will not regulate, or be the correct voltage for a stock transformer, or have very low efficiency, or have excessive ringing, or have SOA stress, or have problems with reactive loads, are all good items to flesh out - and are often more comprehendable if they relate to a simple circuit (that doesn't do even a vagualy acceptable job).

    Ciao, Tim
     
  14. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    I spelled out the problems, what part do you disagree with? It has a max of 96VAC with a 12VDC power supply, it has a asymmetrical square wave, it can't power squat, it has no overload protections. The most common complaint we keep getting is low AC voltage out. I do not define that as working. The output isn't really a half H bridge either, you need the emitters on the power supply rails to get that, otherwise you loose 1.2 VDC right off the bat (which is not rail to rail on the output), this is part of the problem.

    You keep saying if it does this and that right, the problem with this design is it doesn't do several things right. The definition of doesn't work covers this nicely. In the last month we've had 2 (3?) posts asking for help using this design. Even without Figure 1 inverters keep coming up on this site.

    If you like the design that is OK. The rest of us have to help the folks who can't get it to work or meet the specs they want, and don't know why. I'm working the issue.

    I do not claim my designs are any better. Hopefully they will get better as the article progresses, and I do take the time to explain their problems. I do not plan on going over 120 watts, or multiphase.
     
    Last edited: Jul 19, 2010
  15. timrobbins

    Active Member

    Aug 29, 2009
    318
    16
    Bill - you put up the schematic only, so I'm referring to the schematic. The schematic doesn't present its performance specs. For those coming across this thread, they only see the schematic and go from there. The value to viewers is in identifying the foibles, such as asymmetry and describing what aspect of the circuit causes the the asymmetry. The idea is not to present it as a good design, but to go through the design points so as to provide the viewer with an understanding based on some detail, rather than hand-waving it away as s**t. That then opens the doors for others with good ideas on why it is bad, or with improvements/alternative, to come in and participate.

    Ciao, Tim
     
  16. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    Have you read the website? I have. My comments stand. If you choose to disagree that is your right.

    Too bad the beginners trying to use it can't find the specs either.
     
  17. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    Wookie, or anyone else reading this.

    One thought how to reduce shoot through on this schematic...
    [​IMG]

    Put a small 1µH coil on each source (along with a snubber diode). This should kill the shoot through to almost nothing. What do you think?

    [​IMG]
     
    Last edited: Jul 20, 2010
  18. tom66

    Senior Member

    May 9, 2009
    2,613
    213
    Why have you specified MOSFETs in the diagram but given bipolar transistors as parts?
     
  19. Wendy

    Thread Starter Moderator

    Mar 24, 2008
    20,765
    2,536
    I suspect you got me midedit, the new drawing has replaced the old already, so there isn't any reason to modify the old one.

    Have any suggestions for MOSFETs?
     
  20. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    Having inductance inside Vgs is generally a bad idea but you might get away with it at such a low switching frequency. It generally means exploding transistors in high frequency switchers since the gate charge must flow through that inductor as you switch.

    Shouldn't those diodes be the opposite orientation? When the transistor goes to switch off the inductor current has nowhere to go.
     
Loading...