Tesla switched inverter to Delta (no optimizers) from promised solaredge (with optimizers)

[nwdiver]

Shading is problematic for all string inverters but based on various forum posts it looks like Delta has particularly poor/limited MPPT that can't handle even moderate shading.

That doesn't make sense from a functional perspective. I don't 100% understand how the MPPT 'sweeps the band' but I can see the results and I know what it's looking for. Current through a series string needs to remain the same and the current a solar cell passes is proportional to the sunlight it receives. So a shaded cell may only pass 1A while its neighbors in full sun are pushing 8A. The MPPT will 'see' that at (1A)(350v) it gets 350w OR (8A)(270v) it gets 2160w. It's gonna go with option 2.

The only area this gets complicated is with parallel strings. But Deltas have 3 or 4 MPPT trackers so having independent strings is possible. If they're paralleling strings then that's a problem with the installation not the inverter.

 

[holeydonut]

It's gonna go with option 2 ... which I believe you mean to be 350w OR (8A)(270v) it gets 2160w

If this is the case, then how do you explain why the OP's production decays so rapidly compared to his neighbor's production? It seems like the Delta has some logic to avoid "seeing" those "1A" cells as you identified.

 

[nwdiver]

If this is the case, then how do you explain why the OP's production decays so rapidly compared to his neighbor's production? It seems like the Delta has some logic to avoid "seeing" those "1A" cells as you identified.

Either he is mis-reading the production data or they paralleled the strings on the roof. For Example; If you have parallel strings of 10 and 2 or 3 panels on one string are shaded then the remaining 7 - 8 on that string will have significantly reduced production. I have an older string inverter with 1 MPPT so my rooftop array is 3 strings of 14 in parallel. In the evening one string loses 3 panels to shade and my production drops by ~30%. With a newer 3 channel inverter I would only lose those 3 panels for a ~7% drop.

It's not physically possible to shut down an independent string by shading a few panels unless you drop below the minimum voltage for the input but that should require shading ~70% of the string.

 

[SaveOurPlanet]

Either he is mis-reading the production data or they paralleled the strings on the roof. For Example; If you have parallel strings of 10 and 2 or 3 panels on one string are shaded then the remaining 7 - 8 on that string will have significantly reduced production.

It's not physically possible to shut down an independent string by shading a few panels unless you drop below the minimum voltage for the input but that should require shading ~70% of the string.

Not meaning to hijack the thread, apologies in advance. I have 2 strings each wired in series, then the 2 series are wired in parallel before the junction box. Is there a reason for that, I just want to learn something.

Actually I have 4 strings on the roof, 2 strings in series with 5.32A down to the inverter, 2 other strings in series then parallel with 10.64A down to the inverter.

 

[taphil]

Saves Tesla $20 in copper and 5 mins of labor. And spares a few mouse clicks in CAD drawing.

 

[nwdiver]

Not meaning to hijack the thread, apologies in advance. I have 2 strings each wired in series, then the 2 series are wired in parallel before the junction box. Is there a reason for that, I just want to learn something.

As @taphil mentioned. Laziness or ignorance. Frustrating. If you have an inverter with multiple MPPT channels and they're only using one that's just dumb.

~99% of the time it won't be a problem but the small expense of running them separately is 100% worth the cost for the 1% it is an issue. It's also a great diagnostic tool. You can compare the strings to ensure everything is fine. A failed panel can be masked when the strings are wired in parallel.

 

[SaveOurPlanet]

As @taphil mentioned. Laziness or ignorance. Frustrating. If you have an inverter with multiple MPPT channels and they're only using one that's just dumb.

I edited my my post, string 1 and string 2 in series down to the inverter. String 3 in series, string 4 in series then parallel down to the inverter. On the inverter string 3 and 4 are jumped. My production is fine with shading and all, just curious.

 

[cali8484]

That doesn't make sense from a functional perspective. I don't 100% understand how the MPPT 'sweeps the band' but I can see the results and I know what it's looking for. Current through a series string needs to remain the same and the current a solar cell passes is proportional to the sunlight it receives. So a shaded cell may only pass 1A while its neighbors in full sun are pushing 8A. The MPPT will 'see' that at (1A)(350v) it gets 350w OR (8A)(270v) it gets 2160w. It's gonna go with option 2.

The only area this gets complicated is with parallel strings. But Deltas have 3 or 4 MPPT trackers so having independent strings is possible. If they're paralleling strings then that's a problem with the installation not the inverter.

The MPPT search algorithm is critical. Some string inverter MPPT algorithms do poorly because they don't have the horsepower to continuously search across the entire MPPT voltage range in granular steps so they make assumptions about the search voltage range near some previous reference point that can get stuck with a local maxima on the I-V curve. Such situation can happen when shading happens too quickly for primitive inverter MPPT. Using your example, if the inverter has been operating at 350V for a long time then quick shading happens and the inverter only searches +/-50V around the reference voltage point (i.e. 300V-400V) then it will not find the MPPT voltage at 270V. Delta inverters appear to have a wide MPPT voltage range (50-480V) so I wouldn't be surprised if they have this problem. Of course, there could be other problems like parallel strings as you mentioned.

 

[nwdiver]

The MPPT search algorithm is critical. Some string inverter MPPT algorithms do poorly because they don't have the horsepower to continuously search across the entire MPPT voltage range in granular steps so they make assumptions about the search voltage range near some previous reference point that can get stuck with a local maxima on the I-V curve. Such situation can happen when shading happens too quickly for primitive inverter MPPT. Using your example, if the inverter has been operating at 350V for a long time then quick shading happens and the inverter only searches +/-50V around the reference voltage point (i.e. 300V-400V) then it will not find the MPPT voltage at 270V. Delta inverters appear to have a wide MPPT voltage range (50-480V) so I wouldn't be surprised if they have this problem. Of course, there could be other problems like parallel strings as you mentioned.

For the example given above the nominal voltage would have been ~300v so it's far more likely to settle on 270 than 350. I've seen MPPTs settle on the wrong input voltage but that's always with parallel strings that can have dozens of MPPs that might trick the algorithm. A single series string is significantly more straight forward.

The OP needs to post the string data. If he can show that one panel is in shade and the string is operating at a higher voltage than it normally does then.... yeah... his MPPT isn't working correctly.

 

[dareed1]

There's a couple of technical issues here, at least. The Delta M series inverters have 3 MPPT channels for the M6 and M8 inverters, but just 2 MPPT channels for the M4, M5, and (surprisingly) M10. The OP has 36 panels, so 3 strings on the roof, two of which have to be connected in parallel, probably on the roof. The two strings connected in parallel will likely not do well in MPP tracking during shading.

Secondly, the OP posted a solar generation curve that appears to be clipped at 8.8 or 8.9 kW. The M10 max current per MPPT channel is 20A, so we expect clipping in good sun at 34V*12*20A = 8.16kW. The inverter likely clips at a little more than 20A, so it would appear that either 1) the string that isn't connected in parallel is not actually wired to the other inverter MPPT channel, or more likely 2) all three strings are connected in parallel and connected to just one MPPT channel. If 2) is the case, I suppose that the MPPT tracking could get easily confused.

If it were my house, I would get the M Tool app (password is the inverter date code seen on the side of the inverter; activation code is 6532), and query the inverter history to see the voltage and currents in full sun and with the roof partially shaded. I would also use the M Tool app to check the firmware level (I think 3.1.7 is reasonably current) and that the clock is set to PDT as seen on the graphs time axis. All of this has to be done in very close proximity to the inverter. I use phone screen shots to record the graphs.

I can agree with the OP that a SolarEdge system should have been installed in preference to a M10 inverter, or as an alternative, use two Delta inverters so that at least 3 MPPT channels are available. I imagine that showing Tesla repeatedly that system isn't performing correctly will be superior to hectoring them with emails. BTW, I had the best results by writing an email with graphs and pictures, and following that up with a phone call, waiting until I spoke to a live person. The first person you can talk to is not likely to have either the knowledge or agency to do anything, but I found that pleasant persistance would get me to a technical person.

 

[cali8484]

There's a couple of technical issues here, at least. The Delta M series inverters have 3 MPPT channels for the M6 and M8 inverters, but just 2 MPPT channels for the M4, M5, and (surprisingly) M10. The OP has 36 panels, so 3 strings on the roof, two of which have to be connected in parallel, probably on the roof. The two strings connected in parallel will likely not do well in MPP tracking during shading.

Secondly, the OP posted a solar generation curve that appears to be clipped at 8.8 or 8.9 kW. The M10 max current per MPPT channel is 20A, so we expect clipping in good sun at 34V*12*20A = 8.16kW. The inverter likely clips at a little more than 20A, so it would appear that either 1) the string that isn't connected in parallel is not actually wired to the other inverter MPPT channel, or more likely 2) all three strings are connected in parallel and connected to just one MPPT channel. If 2) is the case, I suppose that the MPPT tracking could get easily confused.

If it were my house, I would get the M Tool app (password is the inverter date code seen on the side of the inverter; activation code is 6532), and query the inverter history to see the voltage and currents in full sun and with the roof partially shaded. I would also use the M Tool app to check the firmware level (I think 3.1.7 is reasonably current) and that the clock is set to PDT as seen on the graphs time axis. All of this has to be done in very close proximity to the inverter. I use phone screen shots to record the graphs.

I can agree with the OP that a SolarEdge system should have been installed in preference to a M10 inverter, or as an alternative, use two Delta inverters so that at least 3 MPPT channels are available. I imagine that showing Tesla repeatedly that system isn't performing correctly will be superior to hectoring them with emails. BTW, I had the best results by writing an email with graphs and pictures, and following that up with a phone call, waiting until I spoke to a live person. The first person you can talk to is not likely to have either the knowledge or agency to do anything, but I found that pleasant persistance would get me to a technical person.

Why would Tesla knowingly install an inverter that only has 2 MPPTs on a system with 3 strings with shading?

 

[Heretic]

but just 2 MPPT channels for the (surprisingly) M10.... The OP has 36 panels, so 3 strings on the roof, two of which have to be connected in parallel, probably on the roof. The two strings connected in parallel will likely not do well in MPP tracking during shading.

yes the array is 3 strings of 12, see the roof picture i posted it has colored lines arounf the 12 and the right 24. the 24 are joined in the Jbox in the upper right corner of the picture. also, yes it clips at 8.8kw roughly. total out drops to about 5ish when some of the first 12--panel-string panels are compeltely coverd (not the whole string) and then when any panels in teh 24 is also fully dovered, it drops to zero. The (second) installer taht was here was saying almost what you just did, ie. about the 24 reacting very poorly to shade, and suggested somehow getting a 3rd mppt into the system.

 

[Heretic]

yes the array is 3 strings of 12, see the roof picture i posted it has colored lines arounf the 12 and the right 24. the 24 are joined in the Jbox in the upper right corner of the picture. also, yes it clips at 8.8kw roughly. total out drops to about 5ish when some of the first 12--panel-string panels are compeltely coverd (not the whole string) and then when any panels in teh 24 is also fully dovered, it drops to zero. The (second) installer taht was here was saying almost what you just did, ie. about the 24 reacting very poorly to shade, and suggested somehow getting a 3rd mppt into the system.

Confirm, yes theres are only 2 blocks to tie into in the M10 inverter, no option to up it (one thing the guy i just mentioned was grumbling about) the M6 and M8 both have 3. data sheets list 2 mppt for m10, 3 for 6 and 8. I bet its not the appropriate inverter for the situation, and i posted my suspicion about why they did it above (based on what the instellar the monring of, told me about what was happeing at the warehouse... just got what they could and tried to get it working together).

 

[Heretic]

The reason I posted this second level of detail, is 1) i feel people should know how this colar contractor space could play out for them, and 2) the kind of production behavior that might happen if this happens to them. Originally i was hoping for ideas how to contact teh right kind of people at tesla or what to say to get them to "adjust" it. Now I think the right answer was like a guy on another post here did for same situation carefully inspect the bill of materials the morning of and if it doenst look right, refuse install. He got them to come back another day with a SE (cant find the post right now, i am sure it was this website). I am not entirely sure my BOM the morning of would have shown a Delta, (see the comments the original installer made in earlier post; bet he had a list asking for SE, and warehouse didnt have it).

 

[Heretic]

Here's a bit more info about how the Delta inverter behaves/reacts. This picture is a recent (6/14) day's pattern where there was wispy clouds passing by off and on in the morning. Emphasis on "wispy"; very thin. (there is no shading in the mornings, at least not for the past month or so, the chimney doesnt shade it this time of year.)
I observed the wisp that went in front of the sun at about noon (timing corresponds to the larger slot just above the second "o" in "noon" in the output graph). I could barely tell a brightness differnce of sun on the array. But array output went to zero for about 15-20 mins. I measured the house current voltage, it was typical 117-118VAC line voltage I have here. I thought it might be something weird like line votage making the inverter kick off. I do get this pattern farily frequently though.But the history of line voltges in my UPS and there are no big spikes in that voltage data. I suspect interaction between how the delta inverter's mppt adapts to the panels outputs, the array topology and the individual panel output characteristcs themselves. i see this similar bahavior on any less than totally clear day; huge spikes of output, (dropping to zero and briefly popping back up to 50 or 75% of what would happen if there had been no interupption. the right side of the graph is the typical totally clear sky day, until the array starts getting shade on some panels of the left 12-panel-set about 2:45, then the second (paralleled?) 24-panel-set get knoocked out about 3:40.

 

[Heretic]

Secondly, the OP posted a solar generation curve that appears to be clipped at 8.8 or 8.9 kW. The M10 max current per MPPT channel is 20A, so we expect clipping in good sun at 34V*12*20A = 8.16kW. The inverter likely clips at a little more than 20A, so it would appear that either 1) the string that isn't connected in parallel is not actually wired to the other inverter MPPT channel, or more likely 2) all three strings are connected in parallel and connected to just one MPPT channel. If 2) is the case, I suppose that the MPPT tracking could get easily confused.

If it were my house, I would get the M Tool app (password is the inverter date code seen on the side of the inverter; activation code is 6532), and query the inverter history to see the voltage and currents in full sun and with the roof partially shaded. I would also use the M Tool app to check the firmware level (I think 3.1.7 is reasonably current) and that the clock is set to PDT as seen on the graphs time axis. All of this has to be done in very close proximity to the inverter. I use phone screen shots to record the graphs.

@dareed1 The the second installer that came to check ( took the array apart and cehcked the wiring), he gave me an array map, and showed me where the two portions (the left 12 and right 24, joine dn roof) are coming separately to the 2 mppt terminal blocks in the inverter. So I think its unlikely the whole 3 sets are joined to one mppt, unless he's all wrong.

I have obtained mtool (with some difficulty) and seem to be able to log into the m10 with blue tooth, i'll try getting the data you suggested above.

I have been sending them (maybe too much) data about the performance, many pictures, gave them hard copies. walked though it all with the team lead beginning of 05/2021 (at my property in person, holding hard copies of shading and output graphs that we discussed, example graphs and shading patterns, across a number of different day/sky patterns). I dont think its the technical people that are the issue here i think... (ecept the first guy that just swapped in something off the shelf, who installer lead number two said was a "contractor" and not tesla, maybe its true). Instead, I think its the mgmt people.

 

[Heretic]

The OP needs to post the string data. If he can show that one panel is in shade and the string is operating at a higher voltage than it normally does then.... yeah... his MPPT isn't working correctly.

@nwdiver , can you be more specific about what form the "string data" you are referring to should be? voltages output from the inverter? i dont have any system to monitor panels, the delta system has no way of providing that. ... measured levels on the roof? an array connection map and shading? (I tried to do that part already in earlier posts). Here is the drawing he gave me. "mtool" app sucks bad (i have done app development and this smells like it was slapped together and never finished/cleaned up). but i'll muddle through it.

 

[Heretic]

@dareed1 and @nwdiver

Here is my attempt to get yesterday’s (6/15) history of volt/amp graphs out of my delta M10-TL-US with one of versions of delta's mobile app (Mtool itself on the apple store or android was hardly helpful, i was able to find via a archive site mtool prof (M professional). and how to activate it (act code dareed gave actually). Firmware version seems to be 2.2.13.
The set of graphs are: volts/amps for one 24 hr window (06/15) PV1, PV2, (and PV3 :?)
PV1 (volt, amp, 24 hour window, dies off when the left 12 panels get shaded, so probaly tied to those)

PV2 (maybe the MPPT with 2 strings tied to it? amperages are near double PV1, and its output drops at the time the right 24 panels in upper pisture start seeing shade. )

PV3 ??? RE PV3: maybe just the firmware or software has the variables in it but no terminals to it. The app thinks there are 3x PV, or at least some layers of the software/firmware do. My "checkup" installer was pretty sure there was nothing in there to hook a third input up to, and it looked that way to me when we were poking around in there together ( i was an electronics tech for a number of years an age ago, so i am not totally new to this stuff, just new to solar.)


NOTE the error log top page, I am not familiar with what "arc faults" mean in solar-speak, but sounds suspiciously like some wire, connector, or panel is loose and voltage is arcing across a gap. Please correct me if that is not the right idea. Looseness could be INSIDE the inverter too I guess, even bad solder joint, faulty circuitry. How would the firmware be trying to detect that? Maybe voltage drops faster than a threshold? Or maybe its just crappy firmware?

IN NEXT POST: AC active power, reactive pwr, current, freq, volts; plus the tesla apps Solar CT leg info for the same time window.

 

[Heretic]

AC volt/amp

AC freq, act and reactive pwr

Tesla apps CT deteercotrs current/wattage graph, same period

Temperature of filter boards 0 and 1

 

[dareed1]

I apologize for calling the app "M Tool". As you found, the app that works is called M Professional. I should have looked at my phone before typing
I suspect a problem (maybe the primary problem) is the firmware version, which you say is 2.2.13. A neighbor of ours who has 4kW system with M4-TL-US was seeing clipping at 2.8kW rather than the expected 3.8 or 3.9kW and production wouldn't start until about 10:30AM and would stop around 3PM in late March. The roof has no shading. We checked with M Professional and his FW was version 2.2.11. Tesla upgraded the FW to 3.1.7 (or 3.1.8--I don't remember) and production no longer clips and reaches 3.9kW with a very nice rounded profile. So it is definitely worthwhile to get a FW upgrade. I don't understand the exact issue at their house, but they had never received the small black communication box. I don't know what is called: ours is only identified by the label Tesla and connects by Ethernet cable to our router. Apparently the system was able to send out data to Tesla--maybe by cellular connection?--but wasn't able to receive FW updates. The service person updated the FW and arranged to have the black box sent to the house. No other changes were made. If you can't get Tesla to upgrade the FW, I suspect that you can call Delta and get them to provide you with the current version. M Professional lists an option to upgrade from the app, but I haven't tried it.

I don't know what an arc fault is. Sorry.

The M Professional app is written for all of the Mx-TL-US inverters and maybe others. My app shows PV1 through PV4 on the main page, and History graphs for PV1 through PV3. Our M8 has 3 MPPT channels, but only 2 channels are connected. Anyway your graphs show about 400V when good power is being produced and that's what we expect.

Get the firmware updated.