PV inverter from Prius inverter
 

I'm looking to build a bigger solar power system, with thermal storage. Since going full grid tie is a bit involved, I'm planning to make it "zero export" with limited battery storage instead. Looking at off the shelf inverters, so far I have only found one (Sol-Ark) with full "zero export" support using external sensors. (The rest that claim such functionality merely have the capability to power loads on the output side from DC power even if AC power is available, without backfeeding power to the "input". They are completely unaware of loads on the input side.) At over $6000 for just the inverter, it costs more than I plan to spend on the solar panels (~38c/W in bulk) themselves.

Therefore, I looked into repurposing a Prius inverter for solar power use. Having found a few going for just over $100 (far less than the cost of building a comparable power stage from scratch), that's certainly a great value for something that can very easily power the whole house. It would take more parts to make a whole PV inverter, of course, but even after adding the capacitor bank and filter inductors (the other expensive components), there's still a lot of room to stay within my $1000 budget, probably more like $500. (That's for the completed inverter unit itself, not including the battery bank or other system components.)

Looking at what the EV community has done, it looks like the inverter, with the stock control board removed, just accepts a bunch of logic level PWM signals and some enable signals. In other words, pretty much exactly what a FPGA board can be programmed to output. It even allows independent control of the high and low side IGBTs, good for enforcing a certain current direction.
https://openinverter.org/forum/viewtopic.php?f=14&t=51

Basically, I plan to use two phases of MG2 for interfacing to the two hots and the third phase for a dedicated 120V UPS output. (The first two phases could also serve as UPS outputs with the addition of a contactor to disconnect them from the grid.) Neutral will be connected to the center tap of the capacitor bank, so each phase will be switching between +-200V or so. All 3 phases of MG1 will be used to run the thermal storage compressor, eliminating the need for a separate VFD. The boost converter can be used to interface to a battery pack that is less than the 340V or so needed to get 240V output. The DC/DC converter will also come in handy for low voltage output.

The control side of things will involve a current sensor installed inside the breaker box, transmitting the waveforms in real time to the inverter using PLC. The inverter responds to load current by sourcing current to offset the load current. (That possible since FPGAs can respond very fast, far faster than any microcontroller can.) It will not export power to the grid, thereby avoiding the hassles with going full grid tie.

By zero export do you mean that the power will be automatically switched between the inverters and the grid as needed? I was wondering how long it would take for people to start doing this type of mod. What do you think the chances are of actually getting this project permitted? Regardless, it looks like lots of fun. I'll be watching.
JJ

No, the inverter works by sourcing a current to offset the load current, more like "blending" than "switching". For example, if the load is drawing 10A, the inverter will supply 10A to offset it. (In practice, it will be more like 9.9A to prevent accidental export due to inaccuracies.) Or if there's a 60A load but the inverter can only supply 40A at that time, it will supply that 40A and cut down grid power use to 1/3. That is done seamlessly, so it's basically grid tie without the hassle.

The inverter connects to the house wiring using a 240V plug so it will be considered an appliance as opposed to building wiring. The solar panels will be on a frame anchored to the ground so they will not be part of the house either. The UPS output will (at least at first) only serve loads in the same room and won't interface to building wiring. (And most likely will stay that way, since for my use, it would be pretty easy to move all or most of the critical loads close to the inverter.)

To prevent the disconnected plug from being a shock hazard, it will have various circuit integrity checks including voltage/frequency checking and neutral-ground continuity checking, opening the input contactor if a fault is detected. (The inverter will also lose the PLC signal if unplugged, which while not intended to be a safety mechanism, will be interpreted as no load current.) That said, US plugs aren't exactly the safest power connector in existence - pull them out a little and there will be the full mains voltage on the exposed pins. (Yes, that's even true for 240V US plugs!) Another option is to use a "doubly safe connector" where both and plug and socket are insulated against touch when disconnected, for example Anderson Powerpole.

The first part of the project would be building the current sensor, then I can more or less capture power data on a cycle by cycle basis and run simulations to determine how much current the inverter will have to supply to offset X% of usage. I can also capture the PLC signal using a high speed ADC card connected to a coupling circuit and see if there is any particularly problematic interference. (Since the inverter uses a FPGA for low level control, I could install some Tiffany Yep DSP tricks in it in order to improve tolerance to interference.)

The PLC signal I'm planning to implement is a pair of FM carriers at around a few hundred kHz, one for each phase, with the current signal companded to improve performance at low current. There will also be pilot tones to allow proper scaling of the waveforms and instant detection of a SNR too low to give a trustworthy signal. The pilot tones could also be frequency shifted in order to digitally encode data such as a more accurate power measurement signal although that might be adding a lot of complexity for little additional gain. (I'll build a test circuit and do some measurements to decide if I want to go that route.) I decided to make the current sensor analog since it's pretty difficult to do it in digital without adding a lot of latency - a Wifi link is on the order of a few ms of latency while the inverter needs to react to the current waveform in a few hundred us for it to work properly.

NiHaoMike,
That sounds like a great project. I cant do that in my neck of the woods because anything that touches the grid is subject to their inspection and approval. They are concerned a lineman might get electrocuted. I got around that by keeping all my generated power completely separated including neutrals and grounds. Keep us posted on your progress.

The nice part about not exporting to the grid is that it avoids the problems with exporting to the grid. In practice, the chances of a grid tie inverter powered from a varying source like solar panels staying within voltage and frequency limits without the help of the grid while powering varying electronic loads is very close to zero. With the "zero export" inverter programmed to only source power up to the point where there would still be a (small) draw from the grid, it would be impossible for it to continue operating without the help of the grid.

BTW, as for commercial inverters that can do "zero export" to loads on the output, internally many if not most of them use a "line interactive" architecture that works basically the same as an inverter that can "zero export" to loads on the "input", only difference is that the current sensors are inside the unit rather than outside. There are a few that use a "dual conversion" design but those are designed to primarily operate as a UPS and many of those are incapable of exporting power to the input at all. (Those are mostly used in commercial environments where the load is great enough that having excess power to export basically never happens.)

NiHaoMike that is a great project. The high voltage of the Tesla inverter always scared me off because of the expensive cost of the HV components (and HIGH VOLTAGE DC). HV solar components are available, but its limited to more commercial type gear.

Low voltage systems are just easer to build. But, if you can manage to build it would usher in a great use of the Tesla Inverter.

On a side note I use a Schnieder XW 6848 to do exactly what you are wanting to do. My Electrical Coop makes interconnection way to expensive to do. So I have the XW set to Grid Interactive Net-Zero and set the SELL amps to Zero. It will then use my battery power, and excess solar to power all my loads until I hit a DOD that I set. The inverter will pass through extra energy if I go over the inverters rated output. I know that both the XW and SW can do this.

Logan

instagram .com/wlbryce

I have looked at that one, but unless something has changed, it can only "zero export" to loads on the output side, not the input. I have also watched a review on it and the owner criticizes that the settings are not flexible enough (e.g. can't set the low voltage cutout high enough) despite the hardware being perfectly capable of it.

It can "export" on the input but that would be "net metering" you can set it to export by setting the "sell" amps. so if you know the load amps you can do that. But, if that load drops it will export to the grid. On the Plus side it is 1741 certified so if the grid goes down you will not backfeed.

As to the net Zero setpoint. On the most current firmware version I can set the battery DOD from 0 to 100% so its not a issue. You have full range of when the inverter goes from Net-zero to grid use.

Maybe I don't understand what your really wanting to achieve in regards to powering loads on the input side.

Logan

instagram .com/wlbryce

Basically, the inverter plugs into the existing breaker box via a 240V outlet with a dedicated breaker. It then powers loads on that breaker box but not backfeed the grid, which is done using sensors installed inside the breaker box to detect the current from the grid. Thus far, I'm only aware of Sol-Ark having that feature.

It's also worth mentioning that the reviewer of the XW has reconfigured his system to run in off grid mode most of the time, since with an unbalanced load, that could cause an export condition on one of the phases in "zero export" grid tie mode! (For example, put 2kW of load on one phase and since it is only capable of sourcing to both phases at once, it will source 1kW to each phase, with one phase exporting 1kW to the grid and the other phase importing 1kW from the grid.) The design I'm working on will be able to adjust the phase currents independently, avoiding that problem.

With my xw+ 5546, I just installed it in between my meter and my main load panel in my house and I’m using it as a whole house transfer switch. Then all of my loads are on the load side of the inverter. As long as you are DC coupling to your battery bank and XW plus will do exactly what you want to do if you can put it between the grid and your loads it has a 60 amp breaker and a 75 amp rated transfer switch in it.

The only reason I have to keep my net metering with my utility is the fact that I am AC coupled. I have a SolarEdge inverter that back feeds my main home panel and When the grid is up the XW plus will not control the backflow to the grid. So any extra solar power just flows right back out to my utility. if the grid is down the extra solar will go into the battery bank. The XW plus is also smart enough that once the batteries get full in an off grid/ grid down AC coupled scenario it will start to frequency shift up enough to knock the grid tie inverter off-line.

I was told by several Schneider engineers that the new XW Pro has a lot more export related features than the old XW pluses. they told me that the reason the software was more limited on the XW plus Was that were totally out of memory space.

You seem to know control wiring a lot better than I do but I have my xW plus paired with the new Canex gateway device which has all sorts of useful features if you were a software/automation geek. They don’t offer a lot of features for programming ....... 😎. but it has the pin out for:
12v Digital signals
two can buss circuits
two iso RS 485 circuits
0 to 10 V analog inputs
0-20 mv imputes
And two sets of relay Nc, no and ground pins

They claim in the manual that it is mostly for battery BMS hook ups.... but the device has usb and an ST card slots, networking and it looks to me like you could probably use it to do all sorts of cool things.....

I would love to see somebody figure out a way of using the device to mod an xw+. The gateway is also a master configure tool for the XW plus. I can Wi-Fi login and change all the master settings on the XW plus.

The XW plus is monitoring power current and frequency flows on both AC lines 1 and two. It can also detect power coming in on both of its AC imports the second one is usually designed for a generator but you can prioritize to if you wanted to do things a little differently. It is also monitoring battery voltage and temperature .so it has a full picture of the power flow and would be able to take a mod bus power meter input signal if you wanted to have remote sensing.


It would be nice to set up some sort of PLC in her face or modify the program so that you could start a manual charging up the battery bank based on current flows go into the grid and in a sense Self consume all your AC coupled power.


I think a Xw+ Would do something similar to what you want and as long as you are do DC coupling. the software is all built in to prioritize self consumption of solar and export limited amounts back to the grid four loads on the load side of the unit. if you want to get into programming you could probably tweak the firmware or add remote current sensing and figure out a way of on the fly changing the sell amp setting to match your loads on the import side.

NiHaoMike

Building your own inverter is a huge undertaking. We had let the smoke out of ours more than once.

I've read your post though 3 times and still unsure to how this will work. Your indicating that you will be using a Prius Inverter but thats 3ph power stage. and you mentioned you will utilize some capacitors for a neutral tap. Hmmm

A method of sensing line current and offset to negate exporting power. Interesting!!

I will be interested following the build.

Randen

The reason why I'm looking at reusing a Prius inverter rather than building one from scratch, apart from it already having the hard work of all the gate drive and critical current path layout already done, is because it's actually cheaper to buy that rather than the parts to build a comparable inverter power stage from scratch!

DIY EV builders have pushed Prius inverters a lot harder than my application ever will. In fact, it will only get pushed to a fraction of what it would operate at in its original application.
https://www.youtube.com/watch?v=QRxbT9NQT4U

At this point, I'm building the PLC transmitter for the current sensor. It will use a pair of PLL synthesizers with a crystal reference to ensure the carriers don't drift due to component aging and temperature variations. (They'll be fairly close together in frequency so I need to make sure they won't drift into each other.) The signal will be applied to both phases and the inverter will use receive diversity to improve robustness to EMI.

First try at prototyping the PLC transmitter using 2x 74LV4046A. It works but the jitter is more than I like, giving a noisy signal. The CD4046, based on an older process technology, is rumored to have a better VCO section, which I'll be testing when I get them. (Digital circuits favor newer processes while analog circuits usually favor older processes.)
https://i.imgur.com/rQkvTZe.png
I used GNUradio to build a signal analyzer on my PC. In the actual application, the signal processing code will have to be ported to Verilog in order to run it on a FPGA, the sort of stuff Tiffany Yep is good at. Sadly, she's too busy to really help me with that, but she did give me a list of resources she used to learn how to do DSP in Verilog. (Now I get to have a bit of a feel of what it's like to be Tiffany Yep, but I don't think it would make me as good looking as her...)

I think I understand what you are trying to do. I'm still unsure how you are going to make this system undetectable from the grid. Here in GA if the power company sees PV's on the property they look for the signature of an inverter in the line. A few years ago several DIY systems ended up on craigslist after the power company shut them down.

use the solar system to power a grid-synchronized generator head? that won't have an inverter signature...

The high frequency noise it outputs is just like any other switching power supply, so simply trying to detect that will cause a lot of false positives. (I will also design the output filter to be very effective at attenuating noise, which is required in order for the PLC receiver to work.) Also, it never exports to the grid - in fact, the inverter will not completely offset the load. It also doesn't send impedance test pulses, since it doesn't need that with the lack of export.

Mike I would bet there are many wanting to do the same thing here but like me are not good at electronics. I had thoughts about going grid tie at one point. Its very convenient to be able to store that energy on the grid till you need/want it. In my case, the deciding factor to not go grid tie was the wording in the contract that basically gave my local government free access to my house to "inspect" when they deemed necessary. The other issue is you don't have access when power is out or even worse if the grid goes down. My goal now is low key urban partial off grid.

Storing some of that energy as heat is a good idea. I'm guessing you want it for domestic hot water and radiant heating maybe?

I'm going to be watching this thread with much interest. Perhaps one of the solar manufactures will also be reading and come up with a heat pump tied charge controller. Diverting to grow lights would also be another good source for many solar folks. There is a lack of load diversion controller options in the solar industry. Someone smart could find a good business nitch here.

I may adapt part of your idea. I like the idea of dumping some load for hot water. Resistive heating is a poor option. How about a simple charge controller that detects battery voltage and turns on a compressor/heat pump. If voltage still rises, it turns on another? Not as smooth as a variable but simpler to source parts. A large enough battery bank would smooth out the gaps in load size. Panels would be directly connected to the batteries as the controller would be purely diversion based. I would also include secondary load of LED grow lights once water temp was reached.

What are you going to do with the cold side of the heat pump? How about pulling heat from a cooler/freezer upstream from the source heat? Kill 2 birds with 1 heat pump.

How about going off the DC side? There are a few options with DC 3 phase brushless compressors. You could build a controller interface to pull power off of the batteries to control charge rate. It would be easy to make it a variable vs on/off. Thats basically what the newer inverter heat pumps are, 3 phase DC motors.

I like the idea of not having all your eggs in one basket. My systems tend to have lots of simple parts rather then one large device doing everything. That way I can move things around if there is a failure and keep things going. Its from a grid down self reliant angle. Down side is it requires more baby sitting.

Looking forward to see what you build

One thing that occurred to me, the ideal load diversion would also need to be fairly fast switching. Turning compressors just on and off has the problem of waiting for pressure to drop on the high side so the compressor doesn't stall at start up and also efficiency losses with pull down time. My setup would have to have a delay timer and multiple small compressors. Variable drive would solve allot of this.

Problem is with other large loads that come and go on the system.

Gadget

I as well am on the edge of my seat to see what Mike will come up with. The liquid cooled IGBTs will be a big help. When you start to draw current to keep an efficient home powered the resulted heat for the inverting will create a lot of heat localized in those semiconductors.

I understand what you are thinking trying soft start and multiple compressors and this is a valiant effort however these will only be subject to diminishing returns as the resultant steady draw will be the problem.

If you put your hotwater tank, cookstove clothes dryer and home heating on propane and operate the other loads on solar maybe!!!

It will always be not enough panels and not enough storage.

During the summer we usually have enough for everything the geothermal are air conditioning, cars are charging clothes are getting washed computers lights and all the creature comforts.

Put two or three days of overcast it all comes to a screeching halt.

Randen

So true, there is always plenty of solar when you don't need it.

I'm still working on getting the PLC transmitter to work as I want it to. The CD4046 did not really perform that much better than the 74LV4046A since it was getting pushed close to its operating frequency limit, what performed the best was the 74LV4046A with a very carefully tuned VCO circuit. Next comes the compander circuit, which I'm going to base around photoresistors illuminated by LEDs as the basis for a variable gain cell. I'll have to tune the gain control rates to get the best overall signal. (This is very much getting up to the Jim Williams level of analog circuit design!)

To help with testing, I have built a current signal simulator by connecting a coil of wire to an audio amplifier that I can play signals through with a PC. With 100 turns, the amplifier will only have to source 2A to emulate a 200A current.

I also got a 90A contactor that will be used to connect the grid, running the coil from 240V AC as intended, it uses 12W just to stay on and buzzes pretty loudly. But if I use a 170V DC pulse to turn it on and then supply 12V DC to hold it on, it's silent after switching on and uses just 0.8W! (I selected the voltage based on the current that was drawn when powered from AC. I have found that even in its worst case orientation, it will stay on with a mere 4V applied to the coil.)

NiHaoMike

Just curious if there is any update on the progress of your inverter??? Our late winter early spring was quite disappointing for any serious product , alas we had to buy electricity

Randen

I got a lot more parts for the inverter (the inverter module itself, capacitor bank, inductor cores, various control circuit parts) and I'm working on it. It will be a long project for sure, one with a lot of learning opportunities.

NiHaoMike

Just wondered if you were able to get any electrons to flow yet. Here days are the longest and of recent quite warm. Our biggest consumers of current such as the Geo-heatpump and car charging are using most of the days production. Today being the longest is quite cloudy with thundershowers.. Oh well.

One of our fellow members is quite eager to follow in our footsteps with a single phase inverter. At this point of time I'm extremely busy with another huge project and can't possibly take on another.

Thought I would check in

Randen

Working on the wireless sensor - managed to get it outputting a nice, low jitter signal. For a test, I had it transmitting music over the power line and the quality is surprisingly good - not HD but about as good as expected for FM. Next up will be implementing the subcarriers for the phase current imbalance and digital data transmission. I now understand why no commercial zero export inverter uses PLC for the sensors - it's hard to get right.

I also started planning out the FPGA design. The FPGA I chose is a Spartan 6 LX100-3, because it's relatively cheap (about $10 for the board!) and quite large with 180 DSP cores and just over 100k general purpose logic cells. It's not as energy efficient as the (much more expensive) Artix 7 I have, but it doesn't use enough power (just a few watts at most) to significantly impact the efficiency of the whole system and is unlikely to be pushed near the limits of what it can do, unlike the Artix 7 that was stuffed full of BLAKE256 instances and run at basically the maximum clock speed it's capable of when it was helping Naomi Wu.

Pretty big Pi Day update. I connected a pair of AK5558 ADCs to the FPGA and managed to program the FPGA to work with them. I then used a Raspberry Pi to send the data stream over Ethernet to my PC so I can use baudline to analyze it.

There was a lot of work leading up to it. The AK5558 ADCs are very high performance ADCs, able to digitize 8 channels at 768kHz and 32 bit. As such, there's very high requirements on the analog design to allow the ADCs to meet their specifications. Two things that took a lot of research were a low noise power supply and a very low jitter crystal oscillator.

768kHz is not fast enough to directly digitize the PLC signal which is on the order of 1.2MHz. For that, I would use a Quadrature Sampling (Tayloe) mixer to downconvert the signal. There will be two of those implemented in order to allow implementation of coherent diversity MIMO, the PLC version of wireless beamforming. Basically, there are two analog front ends picking the signals from the two hot lines. Noise tends to affect each line differently, so with some DSP trickery, the noise can be largely nulled out making the system more robust. The high dynamic range of the ADCs and mixer also greatly improves robustness against impulse noise.

The FPGA programming turned out to be the most difficult part by far. I started with implementing an internal data bus to allow the Pi to change and read back registers implemented in the FPGA. The Pi side of the interface is clocked by the Pi while the internal side of the interface is on its own clock domain that's tied to the ADC clock. I got to deal with the issues of clock domain crossing right at the start since it would be a lot easier to solve it in one place rather than have to deal with it throughout my design. The MASIE (Mixed Architecture System Integration Extended) bus specifies that the CPU (Pi in this case) acts as the master for the I2C, SPI, JTAG, and eUSB/HSIC interfaces, but the PCM interface can be clocked by whichever side that makes more sense for the design. Thus, the PCM interface in my design is tied to the ADC clock so no clock domain crossing needed in this case. (Which is good because retiming PCM data is really complex...)

While doing all that, I now understand just how smart my friend Tiffany Yep really is. I knew that she was smart but only now do I truly understand how complex the stuff she does is. So far, I'm only aware of one other model - Xyla Foxlin - to have participated in a beauty pageant and have such advanced technical skills. (I know quite a few other models who are engineers, but it's one thing to look good enough for stuff like appearing on TV or at live events, quite another to compete with other models.)

And for the Pi Day bit, I used a 3.14kHz sine wave test signal, one phase shifted so that it's about 90 degrees out of phase with the first. (A sine wave is in fact a 2D projection of Euler's Formula.) The sample rate reported by the Pi is not correct due to a quirk of how it handles the MASIE PCM data.

Mostly completed the wireless sensor, what's left is to validate that all of the features work properly and then conformal coat for installing into an outdoor breaker box.