Interesting problem over paneling with parallel strings

[hwy17]

So what inverter is this?

This is a Schneider MPPT 100 600 standalone charge controller. I use an XW+ inverter as well.

 

[fourtytwo]

So what inverter is this?

See post #7!

 

[hwy17]

I don't understand this effect in MPPT mode where my power point seems to start off high, 400v, early in the morning, gets into .7-1kW output still on the high side at 350v, and then actually dips mid morning, running 270-290v at 1.5kW now, before it will come back up into 330-350v in the main curve of 2-3.5kW production.

No doubt there's a perfectly reasonable explanation for that in the dynamics of PV output. I just don't have my head wrapped around it.

Edit: Ohhh, this is probably that shading I mentioned earlier that goes away by 11. It's a shading effect, and what you were getting at about how a fixed voltage would specifically not handle shading well.

 

[fourtytwo]

I don't understand this effect in MPPT mode where my power point seems to start off high, 400v, early in the morning, gets into .7-1kW output still on the high side at 350v, and then actually dips mid morning, running 270-290v at 1.5kW now, before it will come back up into 330-350v in the main curve of 2-3.5kW production.

No doubt there's a perfectly reasonable explanation for that in the dynamics of PV output. I just don't have my head wrapped around it.

Edit: Ohhh, this is probably that shading I mentioned earlier that goes away by 11. It's a shading effect, and what you were getting at about how a fixed voltage would specifically not handle shading well.

Yup you got it exactly! There are some plots on another thread of mine illustrating this here

 

[hwy17]

Well by the sounds of it you really are going to rely on that fuse for fire protection so I hope you take into account the mounting and cooling of that cable in calculating it's hoped for ampacity!

Might as well take the detour here on that while we're at it:

I'm considering using 22.24 ISC combined PV going into 12awg wire, I believe that's roughly 4mm2. That wire is 90C rated and travels through 200ft of 3/4 (20mm) PVC plastic conduit underground and then 50ft of 3/4" (20mm) metal conduit that's in open air, apart from the 2 points at which it passes through a wall for 1" and 4" lengths.

Nobody can say that's a great idea. But I think I'm confident enough that even if the fuse were to somehow fail it's protection, it's not exactly a current that you'd bet on the wire to fail at. I would fuse it at 20 at the array side after the combiner and expect the CC to be self output limited to no more than 330v/16.66 amps input at the fixed voltage point, but like you say with losses in the CC potentially a bit higher.

Now I do have the spare 12 awg wire to double up the section in the metal conduit that's inside the house. I'm only hesitant because our electrical code doesn't allow doubling conductors that small. But that might be a distorted precaution, because I know logically that doubling is electrically better than not. The other hesitancy to double, is that the wire is currently unbroken from the array side all the way to the CC. And every splice adds another potential point of failure, but there is a junction box where I could make that splice outdoors distanced from the house.

 

[wpns]

https://www.electricallicenserenewa...ucation-Courses/NEC-Content.php?sectionID=713 says that’s good for 30 amps, so you should be OK.
Personally, I would double up if you can, or run individual strings back from the array all the way into the house.
I’m not sure that using panels that are good for an ISC of 22 A with a 20 amp fuse is a good idea, but worst case you blow a bunch of fuses and swap it out to 25 or 30 A.

 

[fourtytwo]

So to summarize your commercial inverter doesn't really have the facilities needed to control multiple parallel strings, I am not surprised as the available controls seem very limited.

In your specific case the new string would be connected by a cable having a loop resistance of an Ohm, quite large but you don't say what the existing array is connected by, I assume a lower loop resistance so that will cause a small imbalance. As for current I see no reason it would operate at or near Isc, instead it will always be operating at Impp that is lower (typically 90-92% of Isc). Trying to control cable current by setting a fixed voltage is a non-starter as your system will no longer operate during shading. The only way you could potentially control it is to reduce the maximum load below the maximum production of the combined arrays however you cannot choose how mpp will choose to fulfill the requirement hence my problem as described originally.

So fuse your cable above Impp and leave the inverter in mpp mode rather than fixed voltage is my advice. I think further discussion of this should be in a new thread.

 

[hwy17]

In your specific case the new string would be connected by a cable having a loop resistance of an Ohm, quite large but you don't say what the existing array is connected by

Existing is on the end of the same 250ft wire that the second array would be added onto.

Trying to control cable current by setting a fixed voltage is a non-starter as your system will no longer operate during shading.

It's still not out of the question for me.

So fuse your cable above Impp and leave the inverter in mpp mode rather than fixed voltage is my advice. I think further discussion of this should be in a new thread.

Yup, np, it's an idea that will get its own thread to report the results if I ever actually get to trying it.

 

[hwy17]

So what is your hardware like anyway? With the custom tracking logic I would've guessed something home made, but you also refer to it as an inverter. Is this advanced programming control of an off the shelf AIO?

 

[fourtytwo]

So what is your hardware like anyway? With the custom tracking logic I would've guessed something home made, but you also refer to it as an inverter. Is this advanced programming control of an off the shelf AIO?

It's my own design and it is rather customized in that it is directly connected to a water tank heater that always gets priority any surplus being sent to the grid with export strictly controlled (as I don't get paid for it). The software is entirely assembly language for speed running on a PIC24EP at 40Mips, it is a transformer coupled full-bridge with primary side modulation (just to be different). THD is very low & EMC almost non-existent!

 

[SeaGal]

What I chose to do was to implement thermal current limiting by averaging input current over a minute and forcing mppt to increase voltage a step if the limit were exceed (increasing voltage will reduce current) until such time as the average current subsides.

I guess the long averaging period has to be weighed up against changing illuminance We seem to have so many sporadic clouds in East Anglia recently my power output is up and down like a yo-yo.

 

[fourtytwo]

I guess the long averaging period has to be weighed up against changing illuminance We seem to have so many sporadic clouds in East Anglia recently my power output is up and down like a yo-yo.

Yup just like here in Lincolnshire, I wanted to absorb those peaks without chopping them off by interfering with MPPT straight away and in many cases the average is nowhere near any sensible limit hence the thermal limiting by averaging over a long period as what I am really trying to do is protect PCB traces & the PV current sense resistor. It all came about when adding a second parallel string as Impp/sc suddenly became twice the original design criteria, fortunately I tend to build with slack in hand but it still needed something to protect critical paths in worst case scenario's of MPPT going off on a bender! (software even written by oneself remains unpredictable at times IMOP lols).

 

[hwy17]

It's my own design and it is rather customized in that it is directly connected to a water tank heater that always gets priority any surplus being sent to the grid with export strictly controlled (as I don't get paid for it). The software is entirely assembly language for speed running on a PIC24EP at 40Mips, it is a transformer coupled full-bridge with primary side modulation (just to be different). THD is very low & EMC almost non-existent!

When I was a teenager and I thought about learning some programming, I wanted to start with assembly, because it seemed there was no point in learning on top of a bunch of abstractions. I didn't get anywhere lol.

That was either an indication of how I am not suited for programming, or that I'm well suited and just too poorly motivated. And the irony is I ended up in the world of "regular IT" systems operation where I work on the application ships that float on such a deep sea of abstraction I don't even hope that I could ever see the bottom.

Do you care to share any pictures of the hardware? It's an AC inverter? Are you tempted by Lifepo4 and DC systems? Sounds like you've probably been at this since before the Lifepo4 takeover.

 

[fourtytwo]

Abstraction is ok if you don't need exact timing and total control of the hardware! When you have self-destructing hardware with high energies around it's a bit better to have cast iron control of what's going on or BANG usually expensive!!

Ohh I wouldn't countenance anything containing large amounts of Lithium anywhere near me
IMOP unless your going fully off grid storage is not worth the hassle or investment.

Here are some pic's please excuse my prototyping........

 

[hwy17]

Abstraction is ok if you don't need exact timing and total control of the hardware! When you have self-destructing hardware with high energies around it's a bit better to have cast iron control of what's going on or BANG usually expensive!!

Ohh I wouldn't countenance anything containing large amounts of Lithium anywhere near me
IMOP unless your going fully off grid storage is not worth the hassle or investment.

Here are some pic's please excuse my prototyping........
View attachment 214069View attachment 214070View attachment 214071

Awesome stuff. I got into building some 12v relay logic off my BMS with chassis mounted relays and it clicked to me how the story of circuitry is all about shrinking them. My chassis mounted relay bank is already growing out of my box. I would like to get into the scale of that kind of board next. Something I can wire bits into by hand but use board mounted components.

 

[SeaGal]

Here are some pic's please excuse my prototyping........
View attachment 214071

Ah! Proper electronics where you can see the components without an electron microscope And Veroboard and directly soldered IC's

 

[Warpspeed]

Abstraction is ok if you don't need exact timing and total control of the hardware! When you have self-destructing hardware with high energies around it's a bit better to have cast iron control of what's going on or BANG usually expensive!!

Have to agree there.
When the current is rising at a thousand amps per microsecond, servicing an interrupt is not really an option.

The "smart" stuff is excellent for making a system user friendly, but for direct real time control of high powered electronics its often a disaster waiting to happen.
I much prefer to see brutally simple hardware that can take care of itself, then add all the the gee whiz features to that as a separate non critical system that can crash or go totally nuts without letting the smoke out.

The younger generation of software magicians, have sometimes accused me of being a Luddite, a silly old bugger, and being behind the times.

 

[pollenface]

I guess the mppt is operating outside it's design parameters, hence the interesting behavior. In a perfect world, one might have both parallel strings in matching azimuth/inclination.

 

[Warpspeed]

A lot of these problems simply go away, if the power tracking system just adjusts panel loading to produce a constant solar panel voltage under all conditions.
It will easily beat a perturb and observe algorithm for speed of response with passing cloud, and be immune to false power peaks caused by shading.
It gives away nothing in performance where there is any worthwhile solar to be had.

The only time a perturb and observe algorithm beats it, is under extremely marginal conditions, like 20 watts from a 450 watt panel.

If your 450 watt panel is only getting ten watts per panel with constant voltage, instead of twenty watts in total nasty grey cloud, its not going to make a huge difference in the greater scheme of things where you will be running off your battery or the grid anyway.

 

[fourtytwo]

I guess the mppt is operating outside it's design parameters, hence the interesting behavior. In a perfect world, one might have both parallel strings in matching azimuth/inclination.

Well actually the purpose of the second string is to cover a different part of the sky to deliberately extend the solar day or mitigate some of the local shading effects however during the overlap period on an ideal day they can both be capable generating maximum current hence the problem.