Adding a GTI to my PV water heating system blog

[fourtytwo42]

Having finally gotten the water heater to work reliably (see other blog) my thoughts have turned to a grid tie inverter to use the surplus PV energy once the water tank has reached it's set temperature (62-64degC).
I have already had a go at my own design but got bogged down in the control system trying to resolve the complex feedback compensation issues caused by the output LCL filter poles (this was/is a high frequency ferrite transformer design), I cannot afford a copy of Mathcad . So I decided to cut my losses and side step the issue and use a commercial unit instead.

Wary of the instant blow up versions available from certain parts of the world and needing more than a Kw to use the PV power I watched a well known auction site for well known names and came up with a Fronius IG15 (I deliberately avoided transformerless designs). Fortunately using my existing Semi-sepic converter I can adapt the PV voltage to practically any converter input range.

The next issue I looked at was how to avoid exporting excessive power and wearing my equipment out for no gain (I cannot claim payment for anything exported). To this end I have designed a system that measures power and direction at the meter and sends this information to the existing Sepic converter, that in turn will adjust the amount of power it makes available to the GTI to minimize the export dynamically as household loads change.

Cool huh, I will let you know if it works in due course, meanwhile I have to wait a few days for the GTI to arrive that I shall teardown and bench test for starters

One thing I did note about commercial GTI's is there is NO information or firmware available for the hobbyist, all of that is jealously guarded by the companies and there service agents.

[fourtytwo42]

IG15 Teardown
Made a bit difficult due to no component designators and the necessary conformal coating obscuring some component part numbers, I should also mention the boards are at least 4 layer and the conformal coating makes probing difficult, however
Starting at the power connectors they are screw type with no pressure plates so a bit wire unfriendly (EDIT sorry they have pressure plates), the multiple DC inputs are all connected in parallel.  Immediately behind the connector bulkhead lies the FILTER board containing common mode chokes and various caps for both AC & DC sides, I notice the Y caps are Wima yellow ones that I have encountered exploding before.
Following the AC path from the FILTER to the AC board it is connected to a small 12Vrms mains trafo via an opto-triac, this could be for any number of purposes. The main power path passes through a relay to a large 4.7uF X capacitor and through two huge pot core potted chokes via a current sensor (LEM HXN 20-P) to an Hbridge using IXFX48N50Q 500V mosfets. Each half of the bridge is driven by an HCPL-A314J twin opto-igbt driver, power is derived from an isolated winding on the APU to be described later.
The DC-buss has three 470uF/450V reservoirs in turn fed via a large choke from a bridge rectifier formed of four unidentified TO220 diodes. High frequency AC comes from the chassis mounted main transformer.
The main transformer is Litz wound secondary first then two primaries, windings are right to the cheeks so the only way the 4mm creepage can be met is if the Litz is itself insulated to re-enforced standard (commonly known as TIW).
The DC board (underneath the AC) drives the transformer primaries with what appears to be another H-bridge with unidentifiable IR mosfets driven in exactly the same manner as already described. The feed to this bridge comes via a current transformer (Pulse F1S115NL) from a set of three 390uF/500V reservoirs that are connected directly to the DC feed from the FILTER board. The transformer primaries are linked to the Hbridge through two power relays and I speculate this is to allow series/parallel operation to better cover the wide input range.
The APU consists of a small ferrite transformer driven by an ST B7NK80Z 800V mosfet. Power comes from the DC feed from the FILTER board. There are many rails, at least a floating +15V for the AC board, +15 & +5V for the DC & AC boards.
The main processor is a TI TMS320LF240 DSP, there is also an Altera EPM30302 PLD almost certainly doing PWM & sine modulation. There is another ~28 lead wide soic but I think its just some kind of power regulator going by the number of power tracks to it.
The INTERFACE board is located behind the LCD display at the bottom of the chassis and carries 4 option board connectors (all empty). It is connected to the DC board (processor) by a ribbon cable labelled LTG-BUSS and also to the LCD via another ribbon. The option connectors are potentially dangerous as they seem to have a grid feed from the FILTER board via a connector on this board labelled AC. For some reason this board has at least two chips with associated crystals. One is an NXP P89V664 (flash based 80C51 MCU). The other is a TI TL16C550 UART (why, surely MCU has one ?). There are also about 7 SSI cmos logic chips.
Finaly the LCD board has several buttons, a pretty LCD and a NXP PCF8576CT I2C LCD controller.
Remaining pictures in next post.

[fourtytwo42]

IG15 Teardown, remaining pics

[capt bullshot]

 

[fourtytwo42]

IG15 Bench Test
First the equipment, a bunch of transformers fed by a Variac and ballasted by an electric fire, rectified with reservoirs form the PV source (see pics). The grid is simply connected through a power meter.
In a nutshell I find this thing quirky but then it is the first pro grid-tie I have encountered, also there is a nasty warning about never disconnecting the DC while on load, hmm so not very fault tolerant then 
The APU starts up at about 140V but in unhappy mode displaying "wait ps" and a flashing panel symbol. Pushing it up to >170V gets you past that and into the extremely long (3 minutes) "startup" phase, at the end of this it zaps the PV's with a ~1Amp current pulse and if they drop below 170V you get "power low" that lasts for 30 seconds then sends you back to the slow "startup" again.
If your source impedance/starting voltage is low enough to get past the current zap it will actually start regulating nicely and delivering output power (you also get a nice green led), the lowest the MPP will drive the input voltage is 160V and if the panel falls to ~155V you go back to "power low". Anytime you are naughty the led glows orange (assuming the APU is alive).

Operationally that's it but what I want to know is if someone hits the big DC breaker I have on the wall like all sensible people does this stupid thing blow up  ??

[capt bullshot]

The warning about not to disconnect DC under load is usually driven by the danger of a DC arc that may hurt you, not the GTI. This is at least true for a big brand name German manufacturer of GTIs. Long waiting / startup time is usually due to regulations.

I've got also a quite small one (called Gridfit or the like), intended to use with a single panel, this one really quickly starts up and has no issues with sudden DC disconnection.
It's a (deliberately not mentioned) part of this setup: http://wunderkis.de/pvbat/

[fourtytwo42]

Woww that's some system you have there, I have thought about a DC bus and POL's so many times but am always put off by adding the extra wiring! On arcing I have a proper DC breaker not some re-purposed AC breaker (I have heard of being used), its an IMO SI12-PE64R-2 rated 12A/700VDC http://www.farnell.com/datasheets/1633615.pdf?_ga=2.126420362.688012521.1501071687-921658267.1472301606.
Well my intention is also to switch off the converter with a mosfet in it's return leg when I want the PV power for something else (heating) so we will see how it behaves (or not)

[fourtytwo42]

Continuing with the IG15 control arrangement.....I had intended to switch the PV input to it BUT the problem is it has ~1100uF strapped across it following a measly 2uH filter inductance so with an IRFB4227 switch the turn on spike at 180V >160A, definitely destructive! Adding an external inductor is really expensive as it needs to be at least 50uH/100A to do any good.
Another alternative would be to switch the grid on/off using a hefty power relay, doesn't do too much for the efficiency!
I found another alternative, there is a plug link on the AC board labelled RUN, and indeed if removed stops it running......well care is required as this is at dc-link potential (grid) however as it is simply a logic signal requires much less power to switch, indeed I am thinking opto-coupler as it appears simply to be connected to an LM239 comparitor input.
Ohh the secretiveness of these people, all these hidden features that can be used
I still might use the power relay as breaking AC on load is no problem and it avoids tampering with the innards, hey whats a few watts anyhow, just depends on what can be found in the junkbox

If you are still with me this also means leaving the GTI permanently connected to the booster output but its APU doesnt start till ~140V and that represents ~750W into the heater, so a few watts wasted in the APU at that level is no worry.

Well there's only one way to find out if it will work reasonably but it's gonna get a blocking diode on its input to prevent all that stored energy from causing a hazard elsewhere!

Is it a blog or a ramble.......

[fourtytwo42]

So we ended up with the power relay (junk box yielded) easily fitted in the option board space and the reverse protection diode (prevents IG15 reservoirs discharging into booster fault)  on the bulkhead using it as a heatsink see pic.

[fourtytwo42]

All installed on the wall, massive compared to the booster and that box is half empty waiting for my own GTI (one day). Meantime done the hardware mods to the booster to add the GTI relay control output, opto-coupler input for the meter power data stream and change the TVS devices to allow higher voltages.

ATM working on increasing the output voltage to be compatible with the IG15 and I have a software problem in the voltage regulator (see scope pic) unfortunately it keep hitting the overvoltage limit and going into the generic 8 second shutdown-restart cycle. I know why this is as when there is no load the MPPT is still doing perturb and the upper limit of perturb progressively charges the output caps till they hit overvoltage. Have to fix this as the GTI wont start if the booster is in a shutdown-restart cycle.

In retrospect I wish I had made the cpu card of the booster bigger, I wasn't considering it becoming such a hub of energy control! I have had to add the new circuitry as a separate board wrapped in transformer tape, not very elegant but effective!

Testing is held up a bit presently as we have some cloudy rainy weather.

[fourtytwo42]

Today successful GTI start from booster (after disabling the IG15 PV leakage test), new booster voltage regulation works a treat and with the ~100W throttle I have in the booster ATM they settle down to ~155V buss voltage. I have loads of energy meters around here and they all agree the GTI is working. The most important is the grid meter itself that proved true to it's specification and did not meter export power, this was also proven by an indoor unit that uses it's LED flashes to measure both real-time power and accumulate units. Meanwhile the booster also has an energy meter and that once again confirmed the GTI consumption.

So next up for this project is the grid energy meter and direction indicator that will stream data to the booster so that I don't wear out my kit giving away free energy to the grid

[fourtytwo42]

At last a system diagram Priority is water heating, GTI just uses surplus power but export is dynamically prevented.

I scratched around for a quick and easy to use drawing package and came across http://meesoft.logicnet.dk/ Diagram designer. I couldn't get along with InkScape, to feature rich and I didn't want to install all of Apache office.

[Simon]

Why don't you look at a power vault, the UK answer to the tesla power wall, I am waiting on a call back and price as I start to plan clearing of the area for my panels.

Like you I currently give my power away but I'd like to not do that so if a storage system that will preserve more value in power than the cost of the unit exists I'll be right onto it.

[fourtytwo42]

Thank you for your reply, I have looked at battery systems but find the cost per Kwh stored high, instead I store energy as hot water (free), I am even thinking of adding more storage capacity in the form of cold (freezer and fridge, also free) but thats another story

I now have the GTI feedback system working to minimize the export power. Making the grid power monitor work was hard given the severe waveform distortion often present (see pic) however got there in the end. The booster caps the power available to the GTI dynamically as household loads change.

As the feedback loop is slow due to the limited transmission rate between the monitor and booster it can take several seconds to adjust but as the object was to limit unnecessary wear and tear on the equipment that is not an issue.

Both the booster and GTI can supply the nominal household load of 50-200W with no fans running. Due to various measurement inaccuracies the practicable reliable minimum export is about 100W when the house is on base load of about 50W.

When I get the chance I will try and concoct a scope shot of the feedback loop in action!

[Simon]

Well ecochute do a battery system and wanted 3750 pounds for it with 3.86KW of useable power and 6000 cycles, at 17p/Kw that could break even, but they supply a grid tie inverter as well because the way all of these systems work by taking solar to mains, mains to battery and then battery to another GTI, all very wasteful as one step could be avoided but I suppose it makes retrofit easier. The power vault does not include the first GTI so hopefully will be cheaper. All of these systems work like yours by trying to balance output versus demand to preserve the battery power for your loads only.

[fourtytwo42]

I see many people building PV systems with batteries and as I have no direct personal experience of that I would not like to comment either way, I do think it's a matter of personal choice. For me I try to keep the costs within foreseeable payback in perhaps 5 years or less and I am very concerned about fire-risk particularly as it seems some at least of the cheaper batteries on the market are from gone bust or cancelled production of the infamous hover boards!

All things handling significant power at highish DC voltages are a fire hazard including my own converters and GTI so they are housed outdoors in steel enclosures. The thought of sharing my living space with Lithium of several Kwh with Kw's of DC flowing in and out on a daily basis is not something I would be comfortable with but then I may be just an old fashioned stick in the mud

OTH if you truly have to go offgrid I can see there is virtually no alternative unless your prepared to run a traditional mechanical generator set.

[fourtytwo42]

When I get the chance I will try and concoct a scope shot of the feedback loop in action!

Well here are a couple of scope shots, first shows DC buss between booster and GTI, pretty uninteresting as remains stable irrespective of load or insolation, good news is that means the two regulators don't fight

A more reveling plot is the second showing panel current, the insolation was very variable whilst the house load was light,the two peaks are caused by the sun suddenly coming out, during the shaded period the power cap has been wound up as the house is importing, then when the sun comes out it gets cranked down as the export is excessive.

Following the two blips is a period of normal regulation where the power oscillates about the ideal point of a small export.

It's a real rubbish day so not much chance of getting any load switching plots, but then it's a uk bank holiday, always rubbish weather hahaha

[Simon]

Well I've had some good weather today made 2.11 KW so far with 1040W peak output. I have a powervault on it's way, I know it's not economically viable yet but i wanted to start dabbling.

[fourtytwo42]

Funny weather patterns I got 2.3Kwh so far and 930Wpeak, yesterday I got 3Kwh for the first time Of course it partly depends on how much electricity we use (rom a 1060W array).

Where about's are you, I am in West Norfolk Lat53N

[Simon]

I'm in northamptonshire, I think my latitude is 51 degrees so only slightly more south, but mine lay on the lawn and grass keeps growing in the way, I mowed the grass and went from 500W to almost 1KW, apple tree that shades them has started to be cut down today, battery bank arrives soon ready for the trial starting 1st october

[fourtytwo42]

OMG Apple Tree, please don't cut it down, think cider, wine, apple pies, there like gold dust, I plant trees hehehe
suppose I might be a tree hugger hahahaha well at least fruit trees
Good luck with the battery bank, hope it does what it says on the tin!
Actually the chap I bought my panels from in Colchester has his on the lawn to as he is in rented so cannot put them on the roof.

[Simon]

Yes it was with a heavy heart that I started to cut it but it is huge and wrecks the lawn with rotting apples and will share my panels. They are cooking apples but I'll put in an eating apple tree further away.

Sent from my Moto G (4) using Tapatalk