mchiriciuc

I do not see why one would want to seek maximum power points at voltages lover than 70% of Voc of the string. If you look at P-V curves below that is almost no usable power. So, in my opinion, if the MPPT is not able to learn dynamically this voltage it would be best to be able to set it as an inverter parameter because it varies with string configuration (and then with temperature). Actually, a pv panel is a current source. It's kind of unpractical to shunt a curent source and wait for the voltage to increase rather to let it extremely light loaded until it reaches max volts (this happens very quickly) and then start increasing the load until the voltage drops a bit. Bad MPPT algorithm. This is the main problem. The corect startup voltage is string configuration dependent.

I never experience AC Out Losses no matter how hard the MPPT spikes: I only had AC dropouts when I had two inverters in parallel operation and only when F80 occurred. As you can see when it spiked (4 seconds spike), it switched to Battery from Grid input (you can tell by the nice smooth 230V line since the grid is all over the place). Again, when it actually started production and MPPT Voltage went to 380V. It starts to make a little more sense since I had F80 just for switching operation mode from Solar Assistant and Boom F80, AC out drop and all went to bypass. But I suspected Solar Assistant was doing something wrong or pure coincidence. I'm using the inverter with the factory firmware, never patched it or messed with the firmware (73.04). I missing a lot of power production, but for now I thought it may be good to have one untouched and monitored for reference until this things are sorted out. So I think that you might be right about the sag. Maybe the supply voltage of the parallel board sags too much during the spike and when the master inverter tries to tell the slaves to switch modes, the communication is not working right. If you have soldering skills I would suggest to power the parallel boards from an external power supply. Since is isolated from the power side you could use the same power supply for all the boards. You would have to disconnect the small two pin white connector (CN1) and solder the external psu wires directly on the filter cap (C1 - as per schematics in Coulomb's link). This way no matter how hard the DC bus sags, the CAN driver and the optos would not be able to mess comms and F80 should not occur.

I think your bet is the wining one. If the optos are degrading after less than a year of service.... damn, I just hope the invertes will work enough to pay for themselves. TI makes also great isolated can transceivers wich cost less ten 1$ more than the SN65HVD230D and are automotive grade and in four years of working with them in automotive environments none failed. It is really a shame. But without any competition in this price range there will be no reason to strive for an increase in quality. Thank you for the link on the schematics and for all your effort. Greatly appreciated.

Regarding the F80 error (you are right it is CAN bus comms failure) I had two inverters in parallel operation which started to throw this error gradually (like 6 months no error and then it started to appear more often) without any change in cabling and/or software settings with the same firmware they came from factory. When throwing an F80 error, the inverters went to bypass and stop working in parallel (this should be normal it avoids the risk of them working out of phase). I also suspect a hardware failure on the parallel boards because in standalone operation there is no error whatsoever (parallel boards still installed). It seems that there is a component regarding to CAN comms that is gradually deteriorating. Maybe the opto-isolators, or maybe a faulty capacitor on the opto's supply line and too much noise on the rail which can also lead to comms problems. It's winter here and I have no means of investigating this further at the time. Will try to hook up my oscilloscope to the parallel board in spring/summer to see what is going on on this boards. It may also have something to do with induced noise in the paralele cables. Extra shielding them seemed to improve it a bit. Does anyone has the schematics for the parallel boards and the CAN comms circuitry from there to the controller?

Hi guys! My battle with axperts so far is like this: Initially had 2 x MPP SOLAR PIP-5048MKX running in paralel until F80 disease got them. Tried everything came to mind: cleaning the connectors, resoldering the connectors, shielding the communication cables between the inverters, nothing worked. Tried to split my loads into a two-phase system but managing the loads and leaving enough room to not overload the inverters was almost impossible. So, I decided to buy an MPP SOLAR MAX 11kVA and sell the MKXs. As it turns out just to run into a complete new set of problems: - as the grid (I live in the country side 30km away from the city) is shit (between 180V and 220V most of the time) the MKX way of converting it to DC and back to AC 230V was a nice and welcome feature since I vade electronic devices which are made for 230Vac input only and some died because of the unstable grid. The MAX does not do that.... so, I ended using a MKX after the MAX just to bring the voltage to 230V for the sensitive loads. Acceptable workaround since I had the MKX lying around - MPPT 90V stuck issue (well, seems that can be solved or at least tamed) - the inverter is injecting power into the grid! I have a smart energy meter from Shelly directly after the main breaker so I will know the power my house is drawing from the grid and turn loads on and off without exceeding the maximum allowed. Surprise, surprise, MAX is dumping energy into the grid as much as 1800VA when in Solar/Utility/Battery output priority and solar power is greater than the power absorbed by the load. Instead of charging the battery it goes back into the grid! And this can be REALLY dangerous. Any ideas? (my first thought is to put a power relay and disconnect the inverter from the grid with an automation, but really... )