# high voltage dc - 220vac solar sine inverter....help!

#### debe

Joined Sep 21, 2010
1,251
The cost of the driver board with LCD display showing Volts Current & Freq was quite cheap, about $35 hardly worth copying. Thread Starter #### Chris Blizzard Joined Mar 12, 2016 24 The cost of the driver board with LCD display showing Volts Current & Freq was quite cheap, about$35 hardly worth copying.
$35 is a weeks wages here. just for the record, suggestions to buy 'cheap' stuff to finish the project falls on poor and deaf ears. to put this into perspective, just add a 0 to anything that seems cheap, and that what it is for me...if i can even source the part or the cash. in days where i had more$ than i could spend, i also just bought random cool stuff hoping it may come in useful in the future. thats the stuff i am working with now. even so, it kind of defeats the idea of a hobby and learning experience to just buy modules and wire them up.
also most ebay sellers are loath to send stuff to my country as they think the eastern blok is evil or something, and postage charges are ridiculous, and the post workers love to top up their $28 a week pay by 'losing' parcels from western countries. hope that dont come across as moody, but its important people understand this. #### debe Joined Sep 21, 2010 1,251 Apreciate where you are coming from, Makes me realise the difficulties you must face. Thread Starter #### Chris Blizzard Joined Mar 12, 2016 24 been working hard on the farm. also got 5pcs of m57959L driver modules. i put some code from the web on a pic but my scope shows glitches where the sine doesnt quite match up at the end of the cycle, which i guess means the hex file used look up tables. i also cannot see the pwm as being tristate to protect the igbt from shorting the supply for a nanosecond from risetimes, maybe my driver modules have this protection. i will need to find better code that works out the sine pwm on the fly and i hope give the short delay between high/low side outputs. would it be better to use pic or olinuxino for this? i suppose olinuxino could allow a UI that can adjust voltage and frequency so the pump fully uses all the available sun or lack thereof. while the project is single phase it would be nice to have a 3 phase mode because i have all the bits. where next friends? ps the igbt are mg06100s-bn4mm which i have 2pcs but i have 4pcs of bigger igbt that should together handle 1MW! my supposed 1500/3000w peak commercial inverter is running the 2 500w pumps in 2 wells alternately from my caravan 24v solar but the lift distance shows the 500w pumps eating 875W each and the inverter output starts to fold back to 160v @580W after half ton of water then it gives up. im currently boosting the 24v system by running 600w fuji rolls into 24v wheelchair chargers to help the battery banks on the 500W mppt. its going to blow up soon if i dont do something, then the whole field will dry up and die. Thread Starter #### Chris Blizzard Joined Mar 12, 2016 24 i hope someone can help me out here with getting this inverter idea up. i have spent every penny on the farm project so far and living on cherries and salad. i have 2x 1200L IBC for gravity feed and installed 1700M of drip tape and planted all my food for the year. it better work or im done for. i need 2 ton water a day for the drip lifted 25M to get from the well bottoms to the tanks that are in different heights to give a good head of pressure all through. a friend has been given 8kw of cracked but working solar panels that he is promising to bring me from UK soon. that should top us up to nearly 10kw which is a serious amount of power and enough to feed most of this village believe it or not. it looks like rain has stopped so i better go back up there and try to get those IBC filled again for the reserve. #### tcmtech Joined Nov 4, 2013 2,867 Sounds like for the time and money being invested a basic and common industrial/military surplus diesel gen set in the 5 - 10 KW range would be a better investment. Thread Starter #### Chris Blizzard Joined Mar 12, 2016 24 i already have a 5kw wind up Diesel lister petters thanks, but the whole point is (without wanting to sound like a hippy) sustainability. its all leading to me getting fully off-grid. that means living without money, and if any heavy fuel is available it will go first to the russian tractor for ploughing unless i can electric power that too. the genset is locked away safe as these are not at all common here, and used for real emergencies such as running the meat freezer if the power goes off more than a couple days and theres a risk to lose a years supply of dead animals. plus it sounds like a small war going on, disrupting my little bubble of country peace. all my investment has gone on land, essential farm equipment (i dont have full use of my left side arm and leg so digging an acre by hand is not going to go well), and the caravan + solar setup which i lived in 2 years already. so far the parts for my megawatt inverter have cost me$12 and im sure you agree that it cant be too hard to use the parts i have for a bulletproof inverter to take low voltage (400v) dc into ac without needing any magnetics.
i learnt a lot already thanks to your posts and comments on this thread and if i were not spending every day working in the boiling sun to the point of heatstroke daily i would have come a lot further on my own. next year should be a little easier. this year was turning jungle into a ploughed field and orchard with the required fresh water for living. that took 8 months of hard labour so keeping it maintained should get easier each year as the soil gets clean of seed. my other option was to sit in an office till im 65...75 by then...or que for the doles meager pittance. no option in my book. i bet you could bash those igbt's together into a working box in an hour or two by what i seen about your background, but probably not on my budget.
PS. I dont live in america!

#### tcmtech

Joined Nov 4, 2013
2,867
All things said and done if it was me and pumping water was my primary concern I would be looking at either setting the pumps up to run on three phase and use dedicated VFD units fore each or build a simple but rugged motor driver inverter.

A simple self-oscillating push-pull unit based around a common iron core low frequency 1 KVA or so power transformer would have no problems running a pump in your wattage range all day long once you have the proper design set up. Basically it's just two sets of properly sizedswitching devices, a few resitors, diodes and capacitors and the proper amounto f capacitance between tyour power transformer output and the pump motor to get eh freqency to stay close to where it needs to be and your good to go.

In my teens I played with that sot of stuff all the time and making a application specific DC to AC inverter to drive a specific motor at load is not very difficult.

#### Chris Blizzard

Joined Mar 12, 2016
24
three phase would have been my choice but there was these 2 bore pumps on sale for about \$80 each so i had them.
im expecting to have a lot of losses and problems using a simple oscillating inverter and the transformer to do 400-240 would be big. and hard to source.
if it ran in switch mode the induction motor wont like it, may overheat from the energy on the squarewaves corners cant be put to use. i currently only have 1100w available power.
what do you think of the driver hybrid modules i bought? will they do the job? interface to the pic/olinuxino drive levels?
just as a sidethought and to simplify a design, might it work to use an eprom to repeat the pwm on 2 (or 3 ) data lines. a simple adjustable clock to set the frequency and a counter on the address lines. that might be very robust. any reason not? i just thought; the resolution you could get using a whole 64k block per phase cycle would be as near perfect as you could get with digital synthesis
i do like the idea of having a pic or linux control on the igbt bricks so i can reprogram them to do other stuff like driving a tesla coil or vlf radio now the spectrum has been released for hams.
but at the end of the day i need water on crops reliably. its great fun!

#### tcmtech

Joined Nov 4, 2013
2,867
Regardless of the voltage change, a 1 KVA transformer is the same size which means that whether it's a 12/24:120/240 or 240/480:10,000 volt if they are of equal VA or KVA ratings there are no major size variations between the two.

As far as 400 VDC to AC is concerned it's dang close to being a direct conversion match for 277 VAC which is a common transformer voltage used in industrial three phase 277/480 Wye systems which make finding 277:xxx voltage transformers pretty easy.

Now relating to square wave going to an induction motor a simple LC filter circuit in between the square wave source and the sine wave motor is an easy way to clean things up.

So given that I say a 400 VDC H-bridge feeding 277:120/240 VAC 1- 2 KVA step down transformer using an old but reliable Inverter driver IC like a LM3524 or LM3525 and a pair of IR2110 high/low side driver IC's plus some basic feedback circuits to use the PWM control of the LM3524 IC keep the output voltage regulated to some degree and a simple LC filter between the two and you would be good to go.

#### Chris Blizzard

Joined Mar 12, 2016
24
well i just went to the field and the caravan full of smoke and sickly smell of burnt silicon. my old inverter burnt out. supposed 1500w continuous duty 3kw peak true sine by dopower. never ran over 1000w.
i better learn a raindance fast. cant afford to replace it. might be the end of the farm project.

#### tcmtech

Joined Nov 4, 2013
2,867
So that makes you about the 5 billionth person on the planet who some how equated harvesting 'free natural' renewable energy meant living for near free and to be a cheap and reliable way to live only to find out they would have been time and money and to have lived like a normal person and just paid the electric and fuel bills.

So just out of curiosity if the amount of time and money that you spent building this system to live for free had been put towards paying a typical monthly electrical bill how many years/decades could you have run those water pumps for on utility power?

I know that around here the cost of a typical mid grade 1500 watt sine wave inverter would pay my electic bill for a good 4 months or for roughly 6 megawatt hours or power which would run a 500-watt pump nonstop for about 11 years.

#### Chris Blizzard

Joined Mar 12, 2016
24
the land i farm HAS NO GRID and to get mains electric brought to it by powergen involves 15000 euro and the installation of pylons.
if i wanted to live like 'normal people' i would spend my days in an office in the middle of a city i hate working for a boss i hate to get money to buy my chemical food substitute in a shop, both of which i hate.
instead i like to live in the middle of nowhere, away from all that because i have freedom to do what i like and there are few humans around to trigger me.
you sir are a troll, and i will feed you no more. please find another thread to comment on.
not interested. you made first on my blocklist

Last edited:

#### tcmtech

Joined Nov 4, 2013
2,867
Well, it's not like you are the fist person who convinced themselves that RE was a way to live cheap and easy only to find out it's far from it due to the simple over sight of not actually doing any realistic cost of implementation and operation study or basic numbers first then tried to build that dream from largely incompatible and poorly designed processes and equipment in a location that is a poor choice for what they are after only to fall on their face in the end because to set a system up like so many of you want unfortunately takes a great amount of time, knowledge, skill, resourcefulness, and a pile of money to boot.

As far as where I live and how well I'm about as far from New York and that lifestyle and environment as a person can get and still be a part of the civilized world. Near dead center of the North American Continent to be more precise.

As with your dream I actually do live in the middle of nowhere on a few hundred acres of land and I live my life as I please without being tied to a "normal capitalistic job" but then I actually had the sense to do realistic and rational cost Vs gains/losses analysis of every aspect of my life while getting set up here over the last 15+ years to live the dream life so many like you so desperately want.

If that makes me a capitalistic troll so be it. It's a great life!

#### tsan

Joined Sep 6, 2014
116
Here are my comments to some of the questions on this thread. Perhaps too late but anyway...

Chris Blizzard said:
what happens if i run that type of ac pump motor on a higher or lower frequency? could i vary the power of the pump, or water flow? or speed it up and have it pump more (it will run nowhere near its maximum head)
Induction motor flux is determined by voltage to frequency ratio, often called V/F ratio. If you change the frequency it is best to change the voltage too so that V/F is constant and therefore motor flux is constant. Induction motor runs well on smaller than nominal frequency in v/f mode (also called scalar mode). On the start from standstill in v/f mode, voltage drop in stator resistance is big when compared to supplied voltage. Outcome of this is reduced starting torque. Voltage can be increased on the start to get better starting torque. This is often called IR compensation but it should not be required on the water pump as breakaway torque is not big.

You can speed the motor above nominal motor frequency but not huge amount especially when the motor is loaded. Typically DC voltage is selected so that about nominal motor voltage can be reached with pwm. Above nominal frequency v/f ratio decreases (also called field weakening) and maximum torque of the motor drops rapidly. At the same time power increases cube of speed on the water pump.

Reducing the frequency below nominal frequency reduces the power a lot on the water pump. Power is proportional to cube of motor speed on a pump. It would be easy to adjust power demand from solar panel by changing frequency. Motor shaft cooling fan is less effective on smaller frequency but the load is much smaller too so it doesn't matter.

Chris Blizzard said:
can i feed igbt modules direct from the pic, or will it need buffering?
It is best to have a driver circuit made with discrete transistors or using specific drives ic.

Chris Blizzard said:
the junction likes a hefty negative pull down to switch off
On this case current being switched off is so small that perhaps negative supply is not necessary. 500w is about 5A peak and the IGBT is 400A. If the motor is started like direct online manner then current can be 30 A peak so not that small anymore. Gate threshold voltage of MG06400D-BN1MM is about 5V so not that small either. Driver M57959L recommends -7 to -10V.

Chris Blizzard said:
i dont know what the pwm frequency is for the hex file i blew to the pic and i would like to try and calculate
Some multimeters can measure frequency. Perhaps multimeter can measure the pwm frequency from the pic output pin.

Chris Blizzard said:
s being tristate to protect the igbt from shorting the supply for a nanosecond from risetimes, maybe my driver modules have this protection.
M57959L is for single IGBT and can't provide dead time. Some bridge drivers that have low side and high side driver has deadtime function.

Chris Blizzard said:
what do you think of the driver hybrid modules i bought? will they do the job? interface to the pic/olinuxino drive levels?
Recommended Vin of M57959L drives is 5 V but can work with 3.3 V too. Current requirement is 16mA with 5V. There is 185 ohm resistor in series with the opto-coupler. Recommended IGBT module size is upto 200A for 600V IGBT. This is not a problem if you will use 500W motor even when the IGBT is 400A rated.

Chris Blizzard said:
if it ran in switch mode the induction motor wont like it, may overheat from the energy on the squarewaves corners cant be put to use.
Most of industrial variable speed drives/motors are run with pwm voltage. Inductance of the motor will filter the current to acceptable level provided there is enough high switching frequency, lets say 2-3 kHz minimum. Usually extra losses are just taken account by derating the motor for example 10%. High rise time of the voltage can cause high voltage peaks on the motor windings destroying them. It is possible to reduce voltage rise time on the inverter output with du/dt or LC filter. One possibility is to use a bigger gate resistor (about 10 ohm used on the IGBT data sheet) to make switching slower. It would create more switching losses but on this case IGBT is really big.

Common mode voltage can also cause bearing currents. I don't if common mode voltage/current can be a problem on single phase inverter. Common mode chokes are often used on various single phase power supplies so it would be good to use common mode choke in inverter too.

Industrial motors are three phase. Single phase motor can have a starting capacitor or running capacitor. Typically (based on google images) capacitor is in series with a winding and does not see pwm voltage directly. At least if there is a running capacitor it would be good to measure if there are peaks on the current because of pwm. Or just use LC filter to have more sinusoidal output voltage.

also i am considering using the solar in 2x3 series parallel with centre tap instead of a full bridge or transformer, then keep adding capacitors to the input until i get maximum power out. i would only need a single igbt half bridge would this have any disadvantages?
If you get 2x 400V (or 325-400V) it should work ok. With this method gate driver is easier to make as it does not require a high side driver.

i do like the idea of having a pic or linux control on the igbt bricks so i can reprogram them to do other stuff like driving a tesla coil or vlf radio now the spectrum has been released for hams.
With pic or Linux control it is possible to run the pump in V/F mode. It would make it easy to adjust frequency -and therefore power- according to available power from the solar panel. Start of the pump would also be easy. Direct online start of the small motor takes about 7 times nominal current. Power factor is low but there is quite some extra active power requirement too. It is no problem if solar panels can supply it or there is enough capacitors on the supply. In V/F mode power demand is small during start.

#### Chris Blizzard

Joined Mar 12, 2016
24
Here are my comments to some of the questions on this thread. Perhaps too late but anyway...

Induction motor flux is determined by voltage to frequency ratio, often called V/F ratio. If you change the frequency it is best to change the voltage too so that V/F is constant and therefore motor flux is constant. Induction motor runs well on smaller than nominal frequency in v/f mode (also called scalar mode). On the start from standstill in v/f mode, voltage drop in stator resistance is big when compared to supplied voltage. Outcome of this is reduced starting torque. Voltage can be increased on the start to get better starting torque. This is often called IR compensation but it should not be required on the water pump as breakaway torque is not big.

You can speed the motor above nominal motor frequency but not huge amount especially when the motor is loaded. Typically DC voltage is selected so that about nominal motor voltage can be reached with pwm. Above nominal frequency v/f ratio decreases (also called field weakening) and maximum torque of the motor drops rapidly. At the same time power increases cube of speed on the water pump.

Reducing the frequency below nominal frequency reduces the power a lot on the water pump. Power is proportional to cube of motor speed on a pump. It would be easy to adjust power demand from solar panel by changing frequency. Motor shaft cooling fan is less effective on smaller frequency but the load is much smaller too so it doesn't matter.

It is best to have a driver circuit made with discrete transistors or using specific drives ic.

On this case current being switched off is so small that perhaps negative supply is not necessary. 500w is about 5A peak and the IGBT is 400A. If the motor is started like direct online manner then current can be 30 A peak so not that small anymore. Gate threshold voltage of MG06400D-BN1MM is about 5V so not that small either. Driver M57959L recommends -7 to -10V.

Some multimeters can measure frequency. Perhaps multimeter can measure the pwm frequency from the pic output pin.

M57959L is for single IGBT and can't provide dead time. Some bridge drivers that have low side and high side driver has deadtime function.

Recommended Vin of M57959L drives is 5 V but can work with 3.3 V too. Current requirement is 16mA with 5V. There is 185 ohm resistor in series with the opto-coupler. Recommended IGBT module size is upto 200A for 600V IGBT. This is not a problem if you will use 500W motor even when the IGBT is 400A rated.

Most of industrial variable speed drives/motors are run with pwm voltage. Inductance of the motor will filter the current to acceptable level provided there is enough high switching frequency, lets say 2-3 kHz minimum. Usually extra losses are just taken account by derating the motor for example 10%. High rise time of the voltage can cause high voltage peaks on the motor windings destroying them. It is possible to reduce voltage rise time on the inverter output with du/dt or LC filter. One possibility is to use a bigger gate resistor (about 10 ohm used on the IGBT data sheet) to make switching slower. It would create more switching losses but on this case IGBT is really big.

Common mode voltage can also cause bearing currents. I don't if common mode voltage/current can be a problem on single phase inverter. Common mode chokes are often used on various single phase power supplies so it would be good to use common mode choke in inverter too.

Industrial motors are three phase. Single phase motor can have a starting capacitor or running capacitor. Typically (based on google images) capacitor is in series with a winding and does not see pwm voltage directly. At least if there is a running capacitor it would be good to measure if there are peaks on the current because of pwm. Or just use LC filter to have more sinusoidal output voltage.

If you get 2x 400V (or 325-400V) it should work ok. With this method gate driver is easier to make as it does not require a high side driver.

With pic or Linux control it is possible to run the pump in V/F mode. It would make it easy to adjust frequency -and therefore power- according to available power from the solar panel. Start of the pump would also be easy. Direct online start of the small motor takes about 7 times nominal current. Power factor is low but there is quite some extra active power requirement too. It is no problem if solar panels can supply it or there is enough capacitors on the supply. In V/F mode power demand is small during start.
serious thanks dude! so many answers and all at once. so my drivers are not so good for bridge because of deadtime. i will need to either find another pic file that uses 4 gpio lines or linux on the sbc and hope not for a crash that switches both on.
i now have a scope and the pic switching at 3khz and i cant see any dead time between the output and inverted output.
i have not identified any comparator inputs working as feedback.

if i go the linux route (olinuxino imx233 micro with arch) will i be able to compute the sine on the fly with pulse width overall adjusted with feedback? would 450mhz cpu be fine, or will i need rtos or assembler level code?
a lot of websites discuss using pointers and tables, but i thinks thats for arduino level stuff rather than a real computer.

if im using tables i might as well scrap the cpu and have a counter and rom chip.
rather have something i can telnet in to though

#### tsan

Joined Sep 6, 2014
116
if i go the linux route (olinuxino imx233 micro with arch) will i be able to compute the sine on the fly with pulse width overall adjusted with feedback?
Do you have an example how pulse timings are calculated on the fly? I have heard selective harmonic elimination only but I think for example sine triangle comparison is better on this low power level. It is easy to use high enough switching frequency.

would 450mhz cpu be fine, or will i need rtos or assembler level code?
I can't say about processing power but to my understanding normal Linux kernel can start to do some housekeeping and during that time there is no new switching commands. Over current will happen very soon.

a lot of websites discuss using pointers and tables, but i thinks thats for arduino level stuff rather than a real computer.

if im using tables i might as well scrap the cpu and have a counter and rom chip.
rather have something i can telnet in to though
Tables on the micro controller gives a benefit that it's easier to control what table row (pulse timings) to use. If solar panel voltage varies 50 V it would be good to change the switching pattern too. One option I think of is to use reference sine wave and compare it to timer to find out switching moments. It reduces amount of data in tables.

#### tsan

Joined Sep 6, 2014
116
i will need to either find another pic file that uses 4 gpio lines or linux on the sbc and hope not for a crash that switches both on.
Google image search with term deadtime shows several circuits that typically have RC network and some logic chip. If you have those components you can make a deadtime generator and use existing PIC program.

#### Chris Blizzard

Joined Mar 12, 2016
24
i like the idea of a millisecond delay from cr and logic. thats a fast route to getting -some- ac sinewave power back to the pumps.
its something i have parts for and can verify on the scope plus a buffer stage for the pic.
and i know its heath robinson but what would be the restrictions on having a reference sine wave that i can vary in amplitude, sample into pwm, shove through an and gate with dead time on a leg, invert for low side and feed to drivers. thats messy but done with a handful of 555 and 741 if it came to it, no?
in the winter i will have time to learn to do some simple coding on the pic and have it run through tables. there is a lot of info there once i get to grips with coding basics.

i will throw the olinuxino back in the parts bin until i can find a better use.

#### tsan

Joined Sep 6, 2014
116
i like the idea of a millisecond delay from cr and logic.
Required dead time is on micro seconds range. 1/3 kHz is about 330 us cycle time. 10 us should be on the safe side if you have negative gate turn off voltage, about recommended turn on gate voltage and gate resistor is similar that was used on the datasheet tests.

Remember that induction motor takes a lot of starting current. Most likely you need the additional solar panels to have enough supply power to start the pump. The picture below is from simulation of single phase capacitor start motor. The motor data is arbitrary 110 VAC machine but it shows the characteristics of current and power behaviour during direct online start. It is not enough to compare nominal solar panel power to motor nominal power.