low insolation GTI startup failures

[fourtytwo42]

Happy birthday to me! As the weather gets worse I get to cast a jaded eye upon why my (Fronius) GTI spends so much time offline, it doesn't help itself by taking 3-4 minutes in between tries BUT why oh why does it cycle around trying so much and getting nowhere!

It consists of an input filter & reservoir connected to the panels followed by a high voltage isolated SMPS followed by an LC filter to a DCbus that is connected to the grid via an Hbridge and LCL filter all standard stuff. BUT it has lots of software shortcomings, one of them is its startup sequence.

As soon as the input voltage exceeds its UVLO threshold it starts its boring 4 minute connect timer, when that expires it starts up the SMPS!! And that's the problem, it just starts it, no attempt at any kind of soft start ramp! Its trying to charge 1200uF to ~400V from only ~1000uF charged to <200V, unsurprisingly unless the source impedance (governed by insolation) is low enough the ubiquitous DCLOW or POWER LOW message flashes up and the whole boring 4 minute restart timer starts again!

This explains why these things can follow the sun down and happily run at <20W but completely fail to restart until the sun is bright, very frustrating, just give me the code so I can put a soft start in there please

Enclosed is a pretty picture of a simulation at about 1.25mS the SMPS starts, the green trace being the DC input voltage and the red the PV power, by 1.75mS the DC input trips the 150V UVLO.

Of course simulation is an extrapolation of reality according to one hypothesis but it beats sitting out in the cold trying to capture it on the scope.

I hope this helps others understand some of there weird GTI behavior when the sun is low and maybe encourage some manufactures to improve there software and testing 

EDIT :- I added a scope shot, not of an actual failure thats very hard but a typical start at 50% insolaton and at ~2.8 seconds there is the sudden drop in voltage caused by there hard SMPS start, when insolaton is lower <20% this causes a start failure.

[f4eru]

Interesting behaviour, yes they seem to have a bug in their system design. Try to get an update from them by reporting that bug

[fourtytwo42]

Try to get an update from them by reporting that bug

I don't know if in your country you have expressions like tighter than the rear of a duck, or like trying to get blood out of a stone but as far as Fronius are concerned both expressions apply This is not unusual, most silicon and equipment manufacturers I have come across behave the same unless you are a VERY LARGE customer!

If I can stand the cold I will try and capture some scope shots when the insolaton reduces tonight (we actually have sunshine ATM) 

[fourtytwo42]

Brrrrr cold hands, feet and all other extremities but got some useful scope shots and tried a modification!
As you may have surmised from the simulation and scope traces the PV panels are not directly connected to the GTI for various reasons including Vmpp of the panels being below the GTI operating range (a common problem in small systems) so there is an MPPT boost controller of mine in between  better to fess up as they say So really it's a shared problem if you like!

Anyway to cut a long story short the transient response of the boost controller is dependent upon it's MPPT rate that was 32hZ so a peak transient response of ~31mS, given the hard start of the GTI this was not always sufficient to cope adequately.  I found I had enough headroom in the ADC acquisition and software processing to be able to double the MPPT rate to 64hZ and so far things seem improved, as I said before it is very difficult to capture an actual failure on the scope so only long term observation will tell if the GTI downtime in low light seems improved.
If the panels were directly connected the transient knocks them way of there MPP hence there impedance rises and I am sure this would give the result I found in simulation hence Mr Fronius you should improve your software, I am just lucky enough to maybe have a workaround!

As an aside a pretty picture of PV voltage with varying insolaton and load from this afternoon

[f4eru]

This is not unusual, most silicon and equipment manufacturers I have come across behave the same unless you are a VERY LARGE customer!

From my experience, it is often possible to get awesome support if you can describe the problem really well, and communicate in a professionnal manner.
Seems your MPPT has a big influence on the problem, what I suggest you :
make a clear and complete report with schematics of your installation and documented oscillograms, and send them this as soon as you have a contact with the second level support. Give concise, useful, and clear information, that really helps.
Keep us informed of the outcome, it'll help potential buyers to choose reading this posts...

[fourtytwo42]

I find we have to deal with things as we find them and devise workarounds

In this case writing a load of indecipherable rubbish here helped me understand what the likely simple solution was and has largely proved to be so, that being increasing the MPPT rate to 64hZ to improve the transient response whilst also realizing that with low light the default minimum boost pwm was insufficient to overcome the minimum startup voltage of the GTI and that aspect has been overcome by adding a minimum as well as a maximum to the voltage regulator within the MPPT.

Most of these issues are of little help to anyone else other than to illustrate that apparent intermittent problems can be caused by unusual corner cases in the operating parameters and are a bit hard to test specifically.

When this system starts up in the morning with low light it is always heating water first, so the GTI is not powered up till later in the day after the water is hot, this normally corresponds to good insolation (as otherwise the water would not be hot enough). The circumstance of attempting a GTI start at low insolation is confined to winter or weird weather where there has been enough energy to heat the water (hence the GTI start) but the insolation has fallen off a cliff edge (late afternoon). That also makes an excuse for why it took me so long to realize there was a problem

Picture is a good start at 270W, at 1.25S the Fronious makes a sudden current demand (I believe that's when it starts it's main SMPS) then at 1.9S it almost goes open circuit (Probably finished charging DC-link cap but not yet enabled grid H-bridge), it is hard to maintain the bus voltage between limits during these two events.

Zzzzzzzzzzz