Author Topic: DIY Low EMF 5 step Sine Inverter  (Read 10049 times)

BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
DIY Low EMF 5 step Sine Inverter
« on: July 29, 2018, 03:12:30 AM »
I thought I should play nice and quite stuffing this in Glort's PV -grid tie work.  While I did add 1500 watts of panels to my existing 875 for this project, I'm now down to the inverter work.

A little review:

I live off grid and have lived with my own 120VDC system and Listeroid for AC and compressed air for a long time.  In preparation for adding inverter power to my system, I upgraded my 120VDC PV charge regulator to handle up to 3000 marketing watts of PV.  It's linear, not pulsing, for very clean DC.

My 120VDC input inverter design is based on the venerable Trace SW series which does low frequency switching of transformers with secondaries in series to create a sine.  That general concept is about all I'm using.  I'm using very soft (slow) switching plus snubbers and minimal passive filtration to generate AC with no audible (via AM radio) EMI on the AC or DC supply.  I'm only using 2 transformers of equal output secondaries, and they are used to make a 5 step sine wave; zero plus two steps up and two steps down.  I originally did a 7 step sine but found 5 step better. The 5 step sine gets the THD down below my ST3 generator head, and eliminates the typical 20% motor heating loss of a MSW inverter.  I use 120VDC for my home and shop power, and have been using a Listeroid 6/1 for AC generation for well pumping and washing machine.

I have previously done quite a bit of testing of the prototype on motor starting at 120VAC output; it would start a 1.5HP tile saw like grid power, instant and effortless.  I ran into a bunch of trouble when I switched transformers for 230VAC output; ultimately I tracked it down to huge inrush surge current.  Until I worked up a software solution of soft starting the transformers, I could only test by starting the processor and H-bridges with 0 volts input, then switching on the 120VDC through two huge chokes and 30K uF of DC filter caps. This provided a slowly increasing voltage which kept inrush current minimal. The massive DC filter was needed for keeping AC off my clean 120VDC. With the filter in place, I now have 1 millivolt of AC ripple with 1000 watts of load.  Backfeeding 230VAC to my step down transformer outside the shop, I've run a couple loads of laundry in the washer (about 1100 watt load also) with no trouble and barely warm heatsink on the H-bridges.  I've also been testing with 500, 1000, and 1500 watts of heat lamps in the shop.

I got the prototype jury rigged on the floor of the battery bank shed and then found that due to my new short 120VDC supply wires to the 10K uF fiter output capacitor, the old Antek transformer inrush current problem came back and bit me.  I should have tested with small fuses, but I got cocky...with 4 mosfets, 3 ICs, and 2 transistors fried the result.  I got it repaired, changed the soft start to extend from 50 usec to 5 millisecond pulses, and now she starts and stops nicely on small fuses, again.

Tomorrow I'll test on my 1/2 HP Franklin submersible well pump. It pumps from a depth of 200 feet.  It normally runs at 1150 watts, with starting peak over 3400.  The issue to be tested will be surge load capability and compatibility with the operation of the Franklin QD solid state capacitor start relay.


BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #1 on: July 29, 2018, 04:01:13 PM »
My intentionally minimalist design with PICOscope measured 8% THD does not comply with the THD specifications for grid tie inverters (5%).   My primary goal for an off grid inverter was to have the lowest possible EMI, while being able to run single phase induction motors efficiently. I intentionally sacrificed THD to allow for slow switching times and reduced EMI, as long as motor efficiency was not significantly affected; keeping induction motors happy was my requirement.  Most generators are in the 9-15% THD range. My ST-3 THD measured at 12%.  So "generator" class THD was sufficient for my needs, and I did beat that, just, when I got rid of the waveform distortions of slow switching a 7 step sine and went to the simpler 5 step. 

The large toroidal transformers will always be costly, and this will always pressure designers away from this approach, especially when EMI is not even considered.

For the low THD required for GTI inverters, the designer has no choice but to do 100x faster switching, even while using a low frequency design per the original 3 transformer Trace SW series. The sine wave must be built with many small steps; Trace used 32 steps per half wave.   The same Trace SW approach with a modern microcontroller can be done, but even with very well designed snubbing, the faster switching design will require substantially more passive filtering of the AC and DC.  AC filtering in particular is costly in efficiency.  For example, only 10 microfarads of total capacitance line to line at 230VAC draws 1 amp (230 watts) continuously.  Capacitance to ground in microfarad levels is largely a "beat the test" farce since the earthing system in the real world is NOT a copper plane from which all measurements are referenced. Capacitance to ground in larger values, for AC or DC typically just turns the entire grounding system into an EMI radiator. I combat that problem by spending a fortune on the earthing system; I use copper flashing, buried,  with rods connected, and copper flashing extending all the way to the filter enclosure, typically. 

The large toroidal transformers used in my design are heavy and expensive. The mass market for GTI's has driven them to make them smaller, cheaper, lighter.  Low frequency, transformer isolated designs were well proven as being the most durable and reliable. 

My usual plea to those adding inverters to their homes:

To avoid developing Electrical Sensitivity and to greatly reduce the risk of accelerated cancer growth and autoimmune diseases (arthritis, thyroid, diabetes, cardiac) I urge everyone with an inverter (GTI or off grid) to add a commercial grade two stage common mode choke filter to both PV input and AC output.  This will greatly reduce the EMI radiating from the home and PV wiring.  These passive filters are available at any big electronics component supplier, and they run $70-$250 each depending on size.  They will not adversely affect inverter efficiency or performance. They are likely to provide -20 to -30dB worth of filtering up to 50MHz, where must of the switching EMI energy will be.  DO NOT buy plug in "filters" such as Stetzer or Greenwave.   They are nothing but 22uF motor run capacitors and will void the warranty of your inverter, will likely cause it to fail, dramatically reduce it's efficiency and will provide absolutely zero common mode filtering. Anyone promoting them is utterly incompetent, knows nothing of the engineering field of EMC, and should be given a good tar and feather job. 







BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #2 on: July 29, 2018, 08:35:29 PM »
I did the inverter test on my Franklin "1/2 HP" well pump this morning.  Success!  I'm pleased with the results.  It starts faster than on the my ST-3/Listeroid which sags a little.  Running current at 235V was down to 5.9 amps in 2 seconds for a total of 1386 watts. That's a bit higher than expected but the pump is over 10 years old and main winding resistance is out of range, slightly.  The pump controller lid specs say 6 amps at 230VAC, even though Franklin literature claims 1150 watts for the pump. 

I did a comparison to ST-3/ Listeroid power. With the same RMS volts and RMS clamp on amp meters in place; results- 6.1 amps 232VAC.  1415 watts.   The frequency was 60.1 Hz which might account for the small increase. Certainly the inverter has performed at least as well as ST-3 generator for motor driving efficiency.

I did find a software bug which is intermittently affecting remote shut down.  Otherwise, I'm ready to get on with proper job of packaging and installation; at present it is just a jury rig of boards and wires on the floor of the shed.








BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #3 on: July 30, 2018, 02:24:51 AM »
Franklin rates it's submersible motors on pumping HP, apparently.  It's nuts.  The rated current/volts for their "1/2 HP" is more typical for a 1.5 to 2 HP motor.  Don't ask my why.  Horse power is generally untrustworthy in American marketing- you can only look at rated current and voltage and service factor and go from there. 

For the typical two stage commercial filter power filters, here's a page full of models about right for a 5KW output GTI.  If you drill down further you will see that the more expensive ones use larger common mode chokes, which have better performance, especially at lower frequencies.  Data sheets give you an insertion loss plot for each one.  I make my own filters, but I wanted to keep this simple for those who might be daunted.  They are not that expensive and even the cheapest one's will reduce the home power EMI dramatically. 

https://www.digikey.com/products/en/filters/power-line-filter-modules/838?k=&pkeyword=&pv2155=u25A&pv2155=u30A&pv2155=u36A&FV=ffe00346%2C5400d2%2C33c02b2&quantity=0&ColumnSort=0&page=1&stock=1&pageSize=25

For  installation, these need to be mounted inside a junction box, since the terminals are exposed.

Commercial electronics are not designed for ease of maintenance, just as cars are not anymore, either. It's all about the race to the bottom.  When I build something, the components alone might cost me 20 times the price of a finished product from China, and they have a custom molded plastic case.   

The challenge for my inverter install is to make it easy for myself for any troubleshooting and maintenance.  I have the luxury of not caring about size, weight.  I want to locate the two H-bridges so I have ready access the top side of the boards for o-scope checkout; the processor board can just unplug, and I'll add some test points on the top side of the power control/remote control board.  The little voltage sense transformer plus rectifier for computing RMS voltage needs a top side test point also.  Fortunately, what commercially would be stuffed in a small shoebox, I'm putting in a 15x15x6 inch box.  I haven't decided on whether the heat sink should be split in two and extend outside the box, or whether I'll just have lots of screened vents on the box and let the heat sink help drive the thermally driven airflow.  The latter probably, as it gives me better access.

I was too scramble brained from the live testing this AM to do the software debug work.  Hopefully tomorrow AM I'll be smarter.   It's the damned grossly overtaxed RMS voltage routine that's got the bug- that makes it ignore the shut down request signal...after traversing the voltage timing table upwards to a large amount.  I'm going to have to do some recording with the Picoscope logic analyzer function again, while flipping some spare port bits in the code so I can figure out what's happening in real time.  I'll be able to use an Arduino Uno board hooked up to a breadboard for simulating a few signals and a pot for AC sense voltage.  It would just be a whole lot faster if I didn't have to go into the house to operate the picoscope, but I can't work near it or the computer.
















mike90045

  • Mendocino Metro
  • Hero Member
  • *****
  • Posts: 1594
  • Mmmm BBQ
    • View Profile
    • Mikes Solar PV page
Re: DIY Low EMF 5 step Sine Inverter
« Reply #4 on: July 30, 2018, 06:05:52 AM »
My inverter logs my 1/2 hp Franklin well pump motor right at 1Kw.  Poor Power Factor on the motor is what kicks it up over 800w/hp

BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #5 on: July 30, 2018, 07:41:06 AM »
Good point on PF, Mike.  I've got to try adding some motor run caps and see if I can lower the power draw a bit or if Franklin already has it optimized.

I also thought about reducing power frequency and voltage a bit after it's started, but I'd need a faster and more capable (ram, faster ADC, priority interupts) controller processor as the little ATmega328P is tapped out. 

Franklin has a PV direct inverter for running their single phase pumps. About $3K and no doubt an EMI monster. But I was able to comb through their literature to find out what they are up to.  It slows down to 30Hz and reduced voltage if PV power drops.  They can go no lower than 30 Hz or their pump's hydrostatic bearings fail.   

 I don't need full power flow at all, it will just be topping off a 2400 gallon gravity feed storage tank on sunny days.  I expect their controller also replaces the standard motor control box so they can manage their own soft start on the start windings.  A bit more of a project than I wanted to handle right now.  With 3 of my soft switch H-bridge boards and a bigger, faster processor I could also do 3 phase MSW (3 step).  6 H-bridges for 5 step sine.

I did get my two 1000 watt transformers wrapped with 3/4" mylar strip tonight, and also wrapped my 14mH gapped, laminated toroidal choke.








BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #6 on: August 06, 2018, 09:35:38 PM »
I've got the inverter installed in the battery bank shed and am running my washing machine PV power using the remote start. Voltage and frequency are perfect.  Battery bank is holding in float at 139V thanks to the new 1500W array addition to my old 875W panel array.

It was a relatively uneventful installation but I had an interesting problem with the generator/inverter 230V switchover coil relay; it was chattering during the transformer soft start, even though voltage during the soft start is limited to single transformer, only 150V max.  The soft start failed as it was also connecting to remote step down transformers and blew the fuses.  I figured it out after some end to end testing.  I've temporarily bypassed the offending relay and tomorrow will add a time delay module on the coil so that it won't kick in until transformer soft start is complete and the the output has stabilized at 230V.

I still have some work to do; I haven't installed the earth grounding system for EMI filtering. Grounding the center tap of secondary side the two 1000W transformers completely eliminates the very low end of the AM band EMI. It's like adding a shield ground between the two sets of windings, which is what I'd probably do if I was having them custom made.  Since this is a low budget DIY project, I just rewound the secondaries of some stock Antek transformers.  They aren't ideal for this application since they buzz with the non-sine primary input, even with no load.  The laminated cores aren't adequately bonded for that. 










BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #7 on: August 08, 2018, 10:16:54 PM »
Light white smoke this AM (more distant fire) with no taste and less misery for me so I got up at 4AM and tried to get something done before the worst of the heat and smoke.

I did get the software updated with a slightly better version of RMS voltage sensing.  I added a second 65 sample half wave with 1/2 a sample time shift.  It's adequate now, +-1%. I can improve it more by exceeding the best accuracy speed of the ADC and doubling the sample rate but will wait until I have finished the grounding system as the little bit of common EMI on the AC now is likely affecting the RMS accuracy.  It's also only a 10 bit analog to digital converter (ADC) so that may be the limiting factor at this point.  Hot rodding the ADC may only make accuracy suffer further.

I topped up the water tank with about an hour of running at 1385 watts in 93F heat.  The transformers were barely warm, the heat sink on the 2 H-bridges was also barely warm.  No need for fans. 

I tried a new transformer soft start routine, with both transformers pulsed at the same time, in opposite polarity so that the net AC volts would be zero until soft start was almost complete.  Alas, it blew a 3A (gentle testing) fuse and I don't have enough 3A spares to continue with trying new variations to see what keeps the Antek transformers happy.  After I cool off for a bit I'll install the time delay controller on the output relay, and then I'll be able to switch between Listeroid 6/1 and inverter power from in my shop. The inverter gets priority- if it's on, it steals the 6 gauge twisted pair coming down the hill.

I'll do a video when we have a windless AM without too much smoke.




BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #8 on: August 11, 2018, 03:03:47 AM »
I got the earth grounding system put in over the last few days. That was a lot of dirty work. I did hook the transformer secondary "center" to ground, but still have to modify the H-bridge output filters to add capacitors to ground.  For all the bench work, I didn't have a decent ground so couldn't test that for effectiveness.

Got the 3A test fuses so back to the start up relay issue. 

Tried the ICM 102 delay on make controller- it is either defective for won't work for a 230V coil relay.  The spec sheet that came with it says 40ma minimum current...a detail left out in the sales literature.  It was just turning on immediately.  I took it down to my shop and tried it on generator 230V, with another Omron relay with 230V coil. Yep, that didn't work.  Coil is being fed 80VAC when "off".  ICM in Syracuse, NY,  blew this one.

I decided to have another go at a zero voltage output transformer soft start.  Why add hardware with some new  software can do it instead?  With keeping zero output voltage in mind, I wrote a new version pulsing one transformer positive while the second negative, then vice-versa, increasing the pulse width by 50usecs each time.  The transformers are the same, so outputs in opposite polarities in series cancel.  I carefully matched the pulse and delay timing of the old successful routine.  Voila, it worked on the first go.  Figures now that I had an open bag of 20- 3A fuses!  I can hear the distinctive buzz-winding down sound of the soft start, with zero volts on the meter, then snap, voltage goes to 230 and the output switch-over relay goes clack.  So I don't need the delay on make relay controller anymore. 

I put the covers on all the boxes yesterday and cleaned up the chaff on the shed floor.  Looks pretty decent. All that's left is the final EMI cleanup. 




BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #9 on: August 11, 2018, 06:07:31 AM »
Thanks, Glort.  Open sourcing this hardware and software and my 120VDC battery management hardware is one possibility.  I'm afraid most need a finished product, competent installation, and lots of support.  Precious few are well enough (or have the resources) to tackle living off grid after becoming ill.

Wish I could use a MIG welder and plasma torch.  Now I have to get back a few hundred feet from those. I wasn't too bad with a stick welder/cutting rod at one time.














« Last Edit: August 11, 2018, 06:23:59 AM by BruceM »

BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #10 on: August 11, 2018, 03:29:10 PM »
Thanks, Glort.  Because of MS and autoimmune thyroid disease that has caused alternating episodes of hypo and hyper thyroid, depression has often been a serious problem for me.  Just in the last year, I found that by taking just 1/4 grain of natural (pig) thyroid plus Cynomel by Grossman, I can manage and not have the horrible aggravation of my epilepsy that thyroid meds have always caused.  I need something that's in the natural stuff, but only a little.  The Grossman generic Cytomel feels MUCH better to me, is head and shoulders above the US generics, and 1/4 the price.  When I'm naturally hyperthyroid, my epilepsy is markedly better (7 years was my longest stretch), despite the difficulty in sleeping; so I know the thyroid medications are the culprit.

"Modern" medicine assumes that hormones that are not the same as your natural ones, but are similar molecules that can be patented, must be good enough, even when patients claim otherwise.  I think that's nonsense, and that for profit medicine with it's patent medicines is hardly in the best interest of the public.  I have similar problems with the hydrocortisone I inject.  It's NOT the same as having normal adrenal output, and even in injectable form causes gastric distress, yet for me, compared to the other two patented synthetic analogs to cortisol, dexamethasone and prednisone it is the lesser of evils. 

I'd sure like to see truely bioidentical hormones available; I think millions would benefit greatly.  Right now our  pharma industry with their regulatory capture makes that impossible,  as does their control of the legislature.

Depression has a dozen different causes, biologically, and is serious business. It may be the harbinger of neuro-degenerative diseases. The conventional approach of just giving SSRI drugs for "depression" will be viewed in the future as dark ages, incompetent medicine.  I highly suggest reading Datis Kharrazian's books.  His functional medicine approach, in looking deeper for the underlying problem, may not be as profitable as 8 minute medicine, but seems a hell of a lot better for the patient.






BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #11 on: August 11, 2018, 06:02:42 PM »
Neurology and Psychiatry are both still pretty much dark ages.  They apply labels to symptoms and call that a diagnosis, and then start trying drugs at random.  There is no science to what drugs may temporarily help for someone and what will not, or worse.  All you can do is just keep trying.  I had a horror show for a life for a few years, trying drug after drug to try to get some control of my epilepsy while new MS lesions were showing at every new MRI. The horrible depression of being seriously autoimmune-hypothyroid, just taking enough Cytomel to keep me from suicide, since taking more turned up the epilepsy dangerously.  The rapid downward slide gradually stopped when someone visited with a meter, showed me the high readings. and I cleaned up my home, EMF wise.  A lot of misery and holes in my brain for the lack of a couple wire nuts in the right boxes.  I owe my life to the person who spoke up and stuck a meter in my face.

Doing laundry on the inverter with the latest software version now.  Did some checking and found an issue I have to work on besides cleaning up some EMI.  The PV charge regulation is going oscillatory at higher loads- this in the late AM when there is plenty of excess amps beyond what the inverter is calling for.  I see a 1V peak to peak sine ripple at about 6.6Hz on the 120VDC.  The battery regulators are flashing visibly.  This is the hassle of closed loop analog control systems.  I can twiddle with the values of the net charge dampening term...I suspect that when I last jacked that up to improve response to sudden shedding or adding of 1200W loads, I may have affected stability for the minutely pulsing nature of the inverter. It's going to take some time to sort it out.   I have a solderless breadboard that can take over that part of the control, getting the essential signals from the PCB and generating the drive signal for the PV current regulator.  I just hope I can make it happen without the inverter on, as I can't stand there for more than 30 seconds with that on.





BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #12 on: August 12, 2018, 04:08:17 AM »
The other thing I found very helpful in the last year is the Gundry low lectin diet. Reduced my headaches, brain fog and muscle pain dramatically.  Don't think I could have finished the inverter project otherwise.  Software was beyond me, totally.

Learned an important lesson today.  I forgot and started the inverter with the step down transformer turned on.  The new soft start can't handle that either, it blew the 16a fuse for each h-bridge, and the same two mosfets on each board.  No other casualties.  I MUST add a time delay relay control to the output so that soft start is ALWAYS with no loads.

 




BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #13 on: August 15, 2018, 03:16:44 AM »
More EMI testing and DC ripple measurement today  led to an annoying discovery.  I use 4- 800 w toroidal step down transformers in parallel (not ideal) outside the shop to get 120VAC for the washing machine, a friend's welder, etc.  On occasion, when remotely switching on those transformers via relay, I get the same old toroid transformer inrush current spike- and it blew fuses and 3 mosfets.  So I will have to add a relay time delay controller (Siemens), another relay, and 400 watts of 50 ohm resistors.  The resistors will limit the start current, then the relay will short them out after 3 seconds. 

Toroids are very efficient, and a bit of a pain.




BruceM

  • Hero Member
  • *****
  • Posts: 3054
    • View Profile
Re: DIY Low EMF 5 step Sine Inverter
« Reply #14 on: August 15, 2018, 05:43:41 AM »
Yes, when I did all my testing, there was a long 12 gauge pair leading from the shop bench to the 230V outlet, which was also 12 gauge back to the shop power panel, and I think 12 or 10 gauge to to the power pedestal in the hard with the transformers.

Now that I'm thinking about voltage losses, and everything is beefed up, it's been an eye opener how high the inrush currents can be for the higher efficiency toroids. The ST-3 never complained.

I found out something interesting today-  I had almost 20 millivolts peak to peak of AC waveform riding on top of my 120VDC when the inverter was running and no other DC load in the shop...and it didn't change in the slightest from no load to 1500 watts load! It took me a bit to realize what was going on.

 The problem is that I ran both AC and DC as 6 AWG twisted pairs in the same metal conduit (something I avoided for my neighbors setup) down the hill about 250 feet...and the 230VAC signal is coupling capacitively onto the DC line.  I'll have to add capacitance from 0, and120VDC to ground at the power pedestal to help remove the AC.  It should be well below 1 millivolt!  The luxury of DC is that there is no loss or performance penalty for adding capacitance.  I rely on clean 120VDC, it lets me use soldering irons and other resistive appliances with no health problems.

There's that old saying about 90% of the project completed, but that last 10% takes 90% of the time.