H Bridge Grid Tied Inverter is good despite non-sinusoidal current?

[ocset]

Hello,
I have done this “poor-person’s” Grid Tied Inverter simulation in LTspice. (Schematic and LTspice simulation are attached, waveform of current supplied back into the mains is also attached)
It puts power back into the mains.   
Admittedly, the current waveform is not sinusoidal, though it does have a decent power factor. 
The thing is, surely it is not so absolutely  massively important for a Grid Tied Inverter to have a purely sinusoidal output waveform? After all, the duty of drawing sinusoidal, in-phase current from the mains rests with the loads, not the inverters. ..And  a great many of the loads on the mains draw non-sinusoidal mains current (eg sub 75W SMPS’s in the EU). If all of the loads on a mains phase are drawing non-sinusoidal current, then a pure-sinusoidal inverter for that particular phase is a waste of time…….because it simply will not be able to deliver sinusoidal current into that particular mains phase.
So do you agree, that the attached H Bridge inverter is of some  good  (admittedly not perfect) and  can play a  role in energy saving by pushing power back into the mains from eg a renewable source?

[ocset]

Also,
If you do object to the previous version of the H Bridge  grid tied inverter,(due to its lack of sinusoidal current)  then does the following one please you?...
This H bridge grid tied inverter (attached) pumps a sinusoidal current back into the mains….however it simply does this by having a sinusoidal reference voltage into its pwm controller….there is absolutely none of the complicated control software which exists in all  the microcontroller solutions for grid tied inverters. This is an extremely simple grid tied inverter. Admittedly, it is devoid of the rather faster dynamic feedback loop of the usual  software based   GTI solutions…but in all truth, as long as somebody on the same mains phase is drawing more  power than this simple H bridge is pumping back into the mains, then it will operate  perfectly well.
So what’s wrong with it?
Why is there a need for  GTI’s with reams of complex software control algorithms?
Attached please find the LTspice schematic, schematic and current waveform of this   “simple sinusoidal  grid tied inverter”

[ocset]

And if the previous GTI did not please you, then what about this next  one….all this one does is simply sense the output voltage of the inverter, and uses a simple analog error amplifier to make it sinusoidal and in phase with the mains voltage. (ie this one's regulated)
…and all this is achieved without using any of the reams of control software that most GTI solutions recommend.
So why are people not using these simple techniques to do GTI’s?
Why must GTIs have so much control software associated with them?...i have just shown that its not always necessary…you can run the simulation (in LTspice) and see for yourself.
For a huge number of applications, the simple GTI methods that I show here would be fine.  A superfast dynamic feedback loop using masses of software  is just not always needed for a GTI………so why di ti.com and st.com etc etc all advocate reams of control software  for GTI’s?

[jbb]

Micros are favoured because they can deploy complex and adaptive control loops.

I would be worried about the following:
1) There’s an IEEE standard which requires 5% output current THD or lower (approximately). This should be achievable with a distorted line voltage.
2) Your control loop is a basic Proportional type. Try adding about 2% of 3rd harmonic voltage to the ‘grid’ (I.e. make that sine wave a bit flat topped) and see what happens.
3) Is that a 2 Ohm shunt resistor? That’s crazy! Try 0.1 Ohm. Is your control loop still stable? Or are you relying on a very lossy shunt resistor to damp things?
4) Where are you getting your sine wave reference from? It’s an important piece of the puzzle. Remember the line will have noise on it and you can’t trust the zero crossings.
5) How are you going to connect a de-energised inverter to the grid? That LCL output filter will ring like a bell if you connect it direct on line.
6) Rumour has it that future rules may require the ability to inject controlled reactive power into the grid. How will you make your current reference?
7) Anti-islanding is required so that the inverter will notice if the grid disconnects even when the local load exactly matches the inverter output. Complex detection logic is required.

And that’s why we use micro controllers and lots of software.

[Phoenix]

6) Rumour has it that future rules may require the ability to inject controlled reactive power into the grid. How will you make your current reference?

Already enforced in Australia as AS/NZS 4777.2:2015 along with a whole host of other power quality control measures that need to be configurable (e.g. Fixed power factor, fixed VAr, Volt-Watt, Volt-VAr, Freq-Watt, various over/under voltage ride-through and trip conditions, dW/dt ramping). I'm pretty sure this was based on an EU standard too.

Other things that are easier in a micro:
All the safety requirements like earth leakage detection (online), impedance to earth (offline), stuck relay test.
General inverter operation like synchronisation, startup procedure, shutdown procedure, protection (over current, internal voltages).
General PV operation like MPPT, multiple peak scanning, voltage limit, current limit, more protection.
Did I mention protection? What about the recovery sequence from faults.

I wrote a good part of the firmware for the first solar inverter in Australia to be certified to 4777.2:2015 a couple of years ago! I think there was something like 16 (or more?) control loops that operate in various conditions to optimise the behaviour. Anything from a cloud covered day generating 10W to full sun day with 30% over-paneling without missing a beat.

Even just focusing on the power electronics - the micro (TI Piccolo) was used to generate more advanced PWM waveforms for mulit-level topologies. (Not that the basic H bridge has huge common mode).

[ocset]

2) Your control loop is a basic Proportional type. Try adding about 2% of 3rd harmonic voltage to the ‘grid’ (I.e. make that sine wave a bit flat topped) and see what happens.

Thanks, this is interesting, because virtually all the PFC control chips (eg LT1248 etc etc)  use the rectified mains voltage itself  to act as the sinusoidal reference for the PFC control…and it might be flat-topped mains, but they still use it as the sinusoidal reference.
So it makes you wonder why  using the  mains itself to provide a sinusoidal reference for a grid tied inverter  controller is so drastically  unacceptable?  Surely, the  prime responsibility of drawing power with ~unity power factor should logically lie mostly with the loads, not the inverters?
It does sound as if these regulations are being made way too strict…obviously, very strict , lengthy regulations suit the big, rich electronics corporations, because they can afford to pander to such strictness without the bottom line taking a big hit…smaller companies often  just can’t afford this.
Seriously I think the electrics regulations are getting way over the top…they are even talking about 98% efficiency needed for PFC stages now (the titanium regs).
All this unity power factor stuff, and low THD, pertains to efficiency, because yes  I agree, if power is drawn with non unity power factor, then overall, there is  more dissipation in the supply distribution system.
But hang on, if we are really interested in saving the worlds energy, and using less of it, then why aren’t more relevant targets being chosen?…for example, why are people not paying more penalty to drive around in big gas-guzzling range rovers when there are far more efficient cars available……why aren’t people who fly places for non essential reasons being taxed big bucks to prevent this high use of energy…..why aren’t carpets being banned…because after all, you then need to expend energy with vacuum cleaners to clean them……..many of the (mostly hot) countries in the world manage fine without carpets, and just have a few rugs…..because such kinds of things (gas guzzling cars, planes, vacuum cleaners, etc etc) use far more energy than a few more percentage of THD in the mains. And if it’s the distribution capacity that’s the reason why THD must be reduced to microscopic levels, then why  is for example the UK admitting 300,000 migrants into the country every year?, because all these extra  people will mean more electricity  being drawn through the electricity grid. Ditto Germany, where recently Germany admitted 500,000 migrants in during one year.

Seriously, these regulations are going way over the top...and its all coming down from the big electronics corporations who want to get rid of their smaller competitors.

Oh, and while we are at it…lets ban  people using electric showers for more than 10 minutes at a time…(put  a time switch on them) electric showers use around 15kW, so they use loads of energy in our countries…many people shower for an hour at a time…….ditto baths……some people are in the tub for hours, constantly topping up the hot water…lets limit this by having a time switch on people’s hot taps ……..such measures would save far more energy and electrical capacity than making the THD regs  extremely tight.

Seriously, when I was in the Army overseas. ....We had to shower with less than one minute of water flow……..we had to wet ourselves fror 30 seconds maxiumum, then turn off the shower and get soaped up….then we were allowed another 30 seconds to rinse off the soap…seriously, its perfectly possible to get very clean like that….and save loads of energy.

There's  no need to make THD regulations extremely tight  to save energy, as discussed, there are much better ways of doing this.

LT1248 datasheet
http://www.analog.com/media/en/technical-documentation/data-sheets/1248fd.pdf

[ocset]

I wrote a good part of the firmware for the first solar inverter in Australia to be certified to 4777.2:2015 a couple of years ago!

Thanks,  in UK, there are no  British companies that make solar inverters. I believe that this is the same in Australia…ie, there are no Australian owned companies that design and make solar inverters.
So, with the greatest of respect to you, what you are saying  is that you wrote firmware for a non-Australian company’s solar inverter. Why do  we think that there are few if any electronics companys owned by either Australia, or UK?...well , I believe its partly because the regs are made so tight  that the chances of UK or Ozz startups actually making it are very slim.

[dmills]

Treez, grid tie inverters are jellybean commodity stuff these days, and you really don't want to be playing in a market that is pretty much undifferentiated product sold on cost, especially if you don't have the scale to pull it off.

Leave the commodity stuff to the big players, they can hit BOM that us small time players can only dream about and have volumes to amortize the NRE costs of complex engineering projects to the point that NRE is a non issue (Remember this is product marketed to a price point). Besides, bashing out the control code for such a thing is not actually that hard, just tedious writing all the test cases and proving it fails safe.

Same thing with power supplies, uninteresting, not a major part of my products cost, and I don't want to deal with certification, just buy the stupid things in. 98%? Watch XP or TDK or Meanwell or CUI or Fidus or whoever meet that at basically the same cost I am paying now, it will happen, and I will need to change nearly nothing. Designing my own power supplies would be like designing and making undifferentiated LED drivers over here, an epic business fail.

Far better to do the systems engineering inherent in taking those undifferentiated boxes and adding value rather then designing the boring stuff yourself - For example, take an inverter, add a control processor and some monitoring, write some code, maybe add a radio link or something, add a small diesel generator with auto start and a modest fuel tank, package as a unit... Now I have a product that can sit in the middle of nowhere, supply a 100-10kW load 24/7/365 and only runs the generator as needed to charge the battery, saves fuel and run time on the engine, maybe add some solar to reduce engine use even further, I could see a use for that, a temporary site lighting and small power unit that needs half the fuel and servicing compared to one running the prime mover 24/7!  The generators, inverter/charger and PLC are pretty much off the shelf, or at least can be commissioned from specialists for small premiums, and the value is in some code and putting the thing together. No idea if it is a viable product, but it is far more likely to be viable then a locally made GTI.
The compelling value for the users is only needing to get in the landrover and head out into the middle of nowhere to refuel the thing half as often, people are EXPENSIVE, kit is cheap.

Sell solutions not undifferentiated products.

I think our CEO would act surprised if I told him we were not a UK electronics company.

Regards, Dan.

[Phoenix]

I wrote a good part of the firmware for the first solar inverter in Australia to be certified to 4777.2:2015 a couple of years ago!

Thanks,  in UK, there are no  British companies that make solar inverters. I believe that this is the same in Australia…ie, there are no Australian owned companies that design and make solar inverters.

There is a company in Ausrtalia that designs and builds a grid tie solar inverter, but they certainly aren't mainstream; I doubt anyone here knows of them. I don't wish to state who (I no longer work there anyway). They are targetting more niche or value add opportunities.

The regulations are only a part of the engineering effort! In fact the regulations aren't that difficult to meet with a competent design.

[ocset]

The regulations are only a part of the engineering effort! In fact the regulations aren't that difficult to meet with a competent design.

Thanks, the point is that it was implied (i admittedly dont think by yourself) that the mains itself cannot be used to generate the sinusoidal reference that is needed for a grid tied inverter....if this is the case, then that  makes cheap analog solutions impractical, and means that grid tie inverters will only realistically be do-able by  software based controllers.

But my point was, that is it really that bad to derive the sinusoidal reference needed from the mains?

I dont think doing it like that would reprsent "incompetent design"

After all,  most offtheshelf analog PFC controllers already derive their sinusoidal reference from the mains....and as far as i know, this doesnt make for loads of problems in the electrical supply system?

But there seems to have been a point made (i admittedly dont think by yourself), that deriving the sinusoidal reference from the mains would end up meaning that THD related regs would be failed.

[dmills]

But a PFC by definition should look like a resistor , that is it should have a current envelope that tracks the supply voltage  Nobody expects a power supply to be injecting harmonic current into the grid to cancel existing distortion.

A generating set by contrast should approximate a sinusoidal source, and grid tie inverters already slightly subvert this by being current sources rather then voltage sources with phase shifts. 

Needing software is just not that big a deal, pretty much everything does there days, and a modest micro is usually cheaper then opamps and passives.

Regards, Dan.

[jbb]

Also, it's much easier to do all the 'basic' protections in firmware (over voltage, overcurrent, over/under frequency etc.).

I once did a test platform which had over/under current protection for three phase and over voltage for a split DC link and that took up half the control board...  I think it was worth it for the test but totally unnecessary for a product when the micro could do it nearly for free.

A generating set by contrast should approximate a sinusoidal source, and grid tie inverters already slightly subvert this by being current sources rather then voltage sources with phase shifts. 

I think that the 'energy springs' concept currently being studied by some power electronics academics is intended to bridge this gap and provide more 'generator-like' behaviour at the grid terminals.


Also, I finally remembered why the voltage-controlled current source is an absolute no-no for mass deployment; it's a negative resistance.  They have a nasty tendency to make things unstable.  Sure one might do OK (it will be stabilised by the grid), but what if someone chooses this negative resistance inverter for the top of a warehouse and installs dozens of the things?

[dmills]

Actually even a simple switched mode supply is a negative (incremental) resistance, voltage goes down, current goes up....

None of this stuff is simple at any kind of scale.

Regards, Dan.

[Phoenix]

But my point was, that is it really that bad to derive the sinusoidal reference needed from the mains?

I dont think doing it like that would reprsent "incompetent design"

After all,  most offtheshelf analog PFC controllers already derive their sinusoidal reference from the mains....and as far as i know, this doesnt make for loads of problems in the electrical supply system?

But there seems to have been a point made (i admittedly dont think by yourself), that deriving the sinusoidal reference from the mains would end up meaning that THD related regs would be failed.

The standards* are different for loads compared to generators and they are also different depending on power level. A 100W power supply will have greater THD and PF allowance than a 5kW inverter. However I also recall that there is a maximum allowable THD for the "grid/grid emulator" an inverter is connected to when performing the inverter current THD tests. It is not tested for compliance when connected to a heavily distorted source. It might be feasible to scale the AC voltage to generate a current reference, but why would you? I would say possible but highly impractical.

(*Australian Standards)

[Phoenix]

Also, I finally remembered why the voltage-controlled current source is an absolute no-no for mass deployment; it's a negative resistance.  They have a nasty tendency to make things unstable.  Sure one might do OK (it will be stabilised by the grid), but what if someone chooses this negative resistance inverter for the top of a warehouse and installs dozens of the things?

For the idea to be feasible the magnitude of the AC current would have to be scaled by the generation requirement and somehow be sufficiently decoupled from the AC voltage magnitude. Perhaps an analog PLL - but at what point is the reference no longer derived from the mains is a philosophical question that can also be debated.

I wouldn't say a an off the shelf analog PFC controller uses the AC voltage without any filtering or THD optimisation/improvement. Modern ones are pretty sophisticated, details on how exactly they work are the manufacturers own "secret sauce".

[NiHaoMike]

It's possible to measure the RMS of the mains and use that as one of the inputs to the gain adjustment circuit. Although for the small grid tie inverter I made for senior design, the negative resistance characteristic was a feature in that if it's unplugged while running, the voltage/frequency would quickly go out of bounds and trip the protection circuit, stopping it from energizing the exposed pins of the plug.