120V AC Boost-Inverter

[Biatu]

Hello, 1st time on the forums!

I have been pondering this for a while...
Wouldn't it be more efficient create boost circuit to charge up a few caps, then switch the output? Instead of just switching a transformer. I figured you could get better control on frequency, voltage, current, noise, and weight.

[james_s]

I have no idea what you're even talking about. You have to post more than one cryptic sentence.

[Bendba]

I think he means transformerless inverter with a dc to dc step up converter and capacitors in place of the battery array or solar panel array.

[Electronicshacker132]

Interesting...

[NiHaoMike]

Boost converters are not very good for large voltage ratios. Most commercial 12V input inverters use push pull converters for that reason.

Now if the input voltage is higher like above 1/3 the bus voltage, then it can be a good way to do it. The Prius inverter does about 200V on the battery and up to 650V on the bus.

[Biatu]

Right I mean Boost Converter -> Cap -> H Bridge -> Load

[Biatu]

Boost converters are not very good for large voltage ratios. Most commercial 12V input inverters use push pull converters for that reason.

Now if the input voltage is higher like above 1/3 the bus voltage, then it can be a good way to do it. The Prius inverter does about 200V on the battery and up to 650V on the bus.

Ok, so going with a push-pull topology step it up, then invert the signal...it just seems to be that the 'inverting' should be done on the high side.

[jbb]

Hi all

I was hoping from more from the OP, but I can't resist any more 

As NiHaoMike says, a standard boost from 12V to the 170V (approx. what you need for 110V RMS output) won't be mush good.  But there are interesting topologies out there which play around with coupled inductors to achieve higher boost ratios.  These are often used in solar (PV) micro inverters.

Of course, they aren't isolated, which means that you would have to be careful about how your 12V system is grounded, lest your 12V battery become live to 110V with respect to local earth.

As Biatu says, the inverting should indeed be done on the high side.  By using a high frequency transformer in the DCDC stage (e.g. >= 100 kHz), the transformer can be small and efficient.  Then the 170V DC is chopped up by an H Bridge at modest frequency (say 20kHz to avoid audible effects) and filtered by LC network to make 100V RMS sine wave output.

[NiHaoMike]

What kind of load are you planning on? If it's an electronic (rectifier) load, you'll be better off with just DC. If it's a motor load, you can get a lot of improvement in starting capacity and overall system efficiency by implementing V/Hz soft start and scaling.

[Biatu]

Well, the idea would be a multi-purpose power inverter. Using a micro to drive the H bridge, and boost circuit. So i'd have an efficient and relatively cost effective sine wave inverter in the default configuration, but have the ability to achieve variable current, voltage, frequency, and waveform. The inverter could be made to be modular, each stage being relatively independent.
[Driver][Boost][Cap][HBridge][Load]

[Biatu]

@jbb, I was thinking of using a multi-stage boost topology, with the last stage providing the isolation

[Biatu]

The last stage of the boost converter could use a buck-boost layout, and the other stage(s) could be bypassed on demand.

With this setup, one could have an adaptive Boost module...if input voltage is >= the output, then operate in isolating buck mode, and bypass redundant boost stages. Else, operate in boost mode, and only enable additional boost stages based on a maximum power point.

[Biatu]

Having a handful of these should suffice...

http://www.ti.com/lit/ds/snosd10b/snosd10b.pdf

[jbb]

@jbb, I was thinking of using a multi-stage boost topology, with the last stage providing the isolation

I would say that a boosting isolated converter is more appropriate. Chaining many stages on the LV side can wreck efficiency (e.g. Each diode will drop say 0.6 V, or 5% of the incoming supply.

The last stage of the boost converter could use a buck-boost layout, and the other stage(s) could be bypassed on demand.

With this setup, one could have an adaptive Boost module...if input voltage is >= the output, then operate in isolating buck mode, and bypass redundant boost stages. Else, operate in boost mode, and only enable additional boost stages based on a maximum power point.

More switching like that can be done. But you might do better by making more use of the buck characteristic of the output inverter stage.

Having a handful of these should suffice...

http://www.ti.com/lit/ds/snosd10b/snosd10b.pdf

Those are very sexy switches. Unfortunately they will be very sensitive to layout etc. in the design. Have you done power converters before?

FYI we are steadily approaching the standard design for a quality true sine wave inverter..

[Biatu]

Have you done power converters before?

A long time ago I got my hands on 4 of these...
http://r.ebay.com/GGTdc4
The 'Cornell Dubilier 12000uF 450 VDC 450V 500CE1325' are the beefiest things I could get my hands on...and originally my design was to utilize a variant of a Joule Thief circuit and had a secondary running off into one of those caps...quickly reached ~400VDC.
I got 4 of these http://www.mouser.com/ds/2/149/HGTG27N120BN-1010523.pdf to form an H-Bridge, and an arduino, and some optocouplers in push pull configuration to drive the IGBTs.

Unfortunately, the project failed as the optocouplers were being parasitic and kept'd bringing down the low side, didn't even begin to test them with the high side, so I looked for other solutions, and eventually left he project alone until I had a better understanding on what I would need.

The most it would do is power a 18W CFL Bulb, and if I let the cap get high enough it blew up a 120V flood lamp.

That project was 5 years ago, and got me interested in power supply design, and multipurpose power packs.

One of the designs in the works are based on a stumbled upon effect while playing with camera flash circuits...basically get a cable with at least 3 wires in it of small gauge, and break or unsolder the flash bulb, then attach 1 of the 3 wires to the output of the trigger coil, and the other 2 to the cap, and then cut the other end of the cable. When triggering the circuit you get a bolt of plasma. Now scale that up with 4 of the aforementioned caps and a 1kW boost stage. (assuming proper equipment so u don't kill ur self.)

Could also apply to coil gun.

Now my target design will not be solely intended for directed energy weaponry, and I'm *NOT* looking to get anyone hurt...even indirectly. My intent is to design a multipurpose power supply than can be adapted to many use-case scenarios, with an efficient and modular base design.

So yes, I do have some experience, but still an intermediate. 

More switching like that can be done. But you might do better by making more use of the buck characteristic of the output inverter stage.

How do you mean?

Chaining many stages on the LV side can wreck efficiency

What if you don't use a plain diode, but rather a mosfet? like so:

[Biatu]

As for the mosfet in place of a diode, I just found this article: https://www.digikey.com/en/articles/techzone/2014/feb/synchronous-rectification-gives-step-up-converters-a-boost, and it's called Synchronous Rectification. But I think it would be better implemented without the IC no?

Update:
Excerpt: "the calculated power dissipation in the high-side SR MOSFET with 10.5 A input current and D = 0.5 is approximately 0.28 W. By comparison, if a Schottky diode with a 0.5 V forward voltage drop were used in place of the SR MOSFET, the predicted power dissipation in the diode, based on equation (2), would be about 2.6 W."

But what about overcoming the limitations of that IC? in the case of a multi stage boost, the input may easily exceed the IC's tolerances. An internal regulator to drive that IC and the FETs would impact that efficiency. But I guess a low power buck...ie, just enough to drive the FETs and the IC would be efficient, unless bypassed when input is within tolerances.

[jbb]

More switching like that can be done. But you might do better by making more use of the buck characteristic of the output inverter stage.

How do you mean?

A traditional output H Bridge stage uses 4 switches, and can deliver a modulation index d between +1 and -1, i.e. 100% duty cycle +ve and 100% duty cycle -ve.  Over moderate time scales (e.g. 0.5 - 1ms) we can average this to estimate the output voltage v_o.
v_o = d v_dc
Where v_dc is the DC bus voltage.  This means that v_o will always be less than or equal to v_dc, which is what I meant by 'buck characteristic.'  Sorry if that was a bit abstruse.  It means that while you might only need 170 V DC to get a 120 V AC output, you can operate that same converter at 300 V DC (but efficiency will drop a little).  So that offers a degree of freedom to your design.

As an aside, d changes with time to make the sine wave, for example:
d = D * cos (\omega t)
Where D\ is often called the modulation index and sets the AC output voltage magnitude (actual control systems vary a lot). D_peak can be between 0 and 1.

Chaining many stages on the LV side can wreck efficiency

What if you don't use a plain diode, but rather a mosfet?

That's a good question.  Certainly, using synchronous rectifiers can help with conduction losses - especially for low voltages.  However, there are also switching losses to think about. If you put down a big MOSFET, it will have lots of capacitance, take longer to switch on and off, and dissipate more energy while it does so.

When considering overall efficiency, consider the following argument.  You could do three stages (first image) carefully and hit 98% each.  Overall efficiency around 94%.

Or, you could do two stages and make some compromises to the DC DC stage to achieve wide input range with 96% efficiency. The inverter retains 98% efficiency. Overall efficiency around 94%, but you've used less components (so smaller, cheaper, less to go wrong).

Hope this helps.

EDIT: fixed quote format problem.

[Biatu]

well said, thank you. naturally I'd go with a 2 stage then