Dear Powerforum community members.

I recently installed a Deye 8kW inverter with 10 x JA 540W panels and a couple of batteries. Panels are two strings of 5 each facing the same direction and having the same slope. Works like a charm except a curiosity that I picked up.

In the morning one string is subject to partial shading (PV1) due to the house structure until about 10h15. On a cloudy day with distinct clouds interspersed with full sun both strings track one another closely until about 7h00 when the unshaded string (orange color) moves ahead slightly and attains the maximum power as appropriate for the irradiance level and cloud cover. The partially shaded string (blue curve) produces less as one would expect but still produces well. At about 10h15 when both strings are unshaded, they track one another almost exactly. See figure below. It seems that the MPPT’s are “awake” and the panel diodes are working well.

On a really diffuse cloudy day most of the irradiance that the panels see is due to diffuse reflections and there is really very little difference between the two string outputs even though the one is supposed to be partially shaded up to about 10h15. The overall level is however significantly lower that full sun. See below.

The curiosity that I picked up is during a cloudless day. In this case all starts well but then the partially shaded string gets bogged down “goes asleep” at a very low power level, 250W basically, up to the point it is in full sunlight when it suddenly takes off with gusto, See below. I believe that in this case the MPPT gets trapped in a local maximum on the overall string Voltage-Current curve and is unable to extract itself from this nonlinearity in the curve to achieve the global maximum. At first, I thought I had a defective MPPT, maybe I do but I swopped the strings between the two MPPT’s with no change apparent. Therefore, either both MPPT’s are defective, I have an issue with some panels, or this is a bug in the firmware.

Interestingly if I cycle the power on the partially shaded string with the DC isolator the MPPT suddenly comes alive and then starts to track at a more appropriate power level for the partially shaded string (see below). It then actually tracks the partially shaded string beautifully until it eventually attains the same power output as the unshaded string.

So, what do you guys and girls think, defective hardware or a less that efficient MPPT algorithm. I personally think it is the latter. A few clouds around, but not to many seems a good thing, early as it "awakens" the MPPT it would seem. Production is still good and in the greater scheme of things (full production day) I could probably live with this but being a perfectionist is a bugger. Ps. I have the very latest firmware installed for the 8kW Deye i.e.

Protocol Version：V0.2.0.1     Lithium Battery Version Number：V0.0.0.0     MAIN：6022-1724     HMI：0000-C35D      LCD Type：8100

5 minutes ago, rfl said:

less that efficient MPPT algorithm

Yeah, some mppts have deep shadow scan modes, but that also has downsides. There is no easy fix for this problem, you have to pick tradeoffs.

14 minutes ago, P1000 said:

Yeah, some mppts have deep shadow scan modes, but that also has downsides. There is no easy fix for this problem, you have to pick tradeoffs.

I assume "deep" refers to how far they scan either side of the current maximum for "something better". I have seen somewhere that some companies actually punt the fact that they scan the full curve continuously or at least very often but that it does lead to increased inefficiencies. 

15 hours ago, rfl said:

Interestingly if I cycle the power on the partially shaded string with the DC isolator the MPPT suddenly comes alive and then starts to track at a more appropriate power level for the partially shaded string (see below).

Heh. That sounds very much like the "stuck at 90 V" problem that all high PV voltage Axperts have. Though it's possible that Voltronic have finally changed the firmware to fix this.

I wonder if there has been some copying of firmware, or at least copying of ideas and/or algorithms.

I have a similar problem, better tracking of two strings (each 6x545w) on a cloudy day, but unbelievable separation on a cloudless day, until there is no shading at all.  The extent of shading for most of the period (until about 1pm)  is a small section of just one panel from an adjacent chimney.  Quite considerable loss of generation capacity throught he morning when we have most of our loads active.

Would a DC optimizer or whole-panel bypass diode on the offending panel help?  Any suggestions besides moving the panels at quite considerable effort and cost.

I haven't tried cycling the strings with the breakers to see if I have the same pattern as rfl. Using essentially the same inverter fwiw (Sunsynk 8kw vs Deye)

 

 

Your solar panels are likely to contain 3x bypass diodes.  Each diode will have a forward voltage of about 0.6V.  This means that a shaded panel will have 1.8V drop across the three diodes.

Let us assume that during hours of shade you are likely in the morning or late afternoon periods where your solar panels will anyway be down on power.  Let us further assume that your panels will deliver 10A at this time.  This means that the 3 bypass diodes will dissipate 1.8V x 10A = 18W of power.  This power will be stolen from other panels.

If you now connect a whole-panel bypass diode, you would only loose 1x 0.6V x 10A, i.e 6W.  A power saving of 12W.

The diodes are cheap, but you would need extra Y-connectors, cables and the schlep of installing this on your roof.  All that effort to "gain" 12W is not worth it.

The above assumes that the internal bypass diodes are all working OK.  If just one of those diodes would be faulty, you could stand to loose a lot of power.  You could test and replace individual diodes.  This would be quite a mission because you will likely need to remove the panel.  You could also measure the voltage drop across a shaded panel.  If that voltage drop is less than 2V that means all diodes work.  If the voltage drop is much higher you have a problem.  THEN the easiest way to "repair", might be to fit a whole-panel (external) bypass diode.

All this from a guy that doesn't even have solar panels.... 😄  If it was my system, I would wait for dusk and then disconnect the one or two panels that get the shade and use a small 9V or 12V battery with a suitable current limiting resistor (say 1Kohm) to connect over the panel while measuring the panel voltage with a multimeter.  I would hope for a 1.5 to 2 volt reading.

Edited May 14, 20233 yr by Modina

Thanks Modina.  My installer is going to love your suggestions!  But maybe I should suggest they swap the panels around in case by bad luck the shaded panel has faulty built-in bypass diodes.

Update: Still waiting on a response from both Sunsynk and the installer on a remediation from either, but I did try the approach of the OP (rfl) of cycling the DC power and sure enough the offending string went from 200w to 2000w with no change to the shading, which does kind of point to it being an issue on the inverter optimization side. 

Is there a safe way to cycle the DC without powering down the inverter - weary of cutting breakers under load?

So this is very interesting...

I have always believed the logic of bypass diodes apply to "short" out the affected PV module, or portion of the PV module. I recently came across the same issue at a client's house.

We have one PV string of 8 panels connected to MPPT1 of a 5kw Deye. MPPT2 was reserved for future panels. Early morning (around 7am) there is no shade on the PV panels, and the MPPT starts up. Production is on par for what I expect at 7-8am in the winter, with a north facing roof.

Just after 8am, a shadow hits the first panel in the string. Output production gets "clipped" at 120W to 250W, from 8am when the shadow hits the first panel, all the way up to 12h30 when all panels are in full sun. Production then shoots up to 2.94kW, and climbs to a maximum of 3.5kw, gradually going down (as expected) for the rest of the day. Its as tho the MPPT tracking algorithm does not allow the power point to be tracked to a reasonable step size (i.e. 1 panel voltage drop ~= 50V).
At other sites where panels start with shaded areas in the morning, and then get full sun as the day progresses, or start in full sun and only get shade late afternoon, it seems that the MPPT algorithm keeps up. BUT where there is a scenario of "full sun, shade, shade, full sun" it looks to me that the MPPT get stuck on the 1ste full sun voltage point, and can not adjust "far enough" to find the actual Max power point.

Asked Deye for a firmware upgrade, and hoped that they might have addressed the problem. No change at all.

I'm heading out to the specific client tomorrow morning to do some tests, and will post my findings here. It does seem that Deye / Sunsynk needs to address this problem - this cannot be "normal"?

 

7 hours ago, FrancoisSnyman said:

I'm heading out to the specific client tomorrow morning to do some tests, and will post my findings here. It does seem that Deye / Sunsynk needs to address this problem - this cannot be "normal"?

On a shaded string there will be multiple power points and the MPPT is suppose to scan the voltage range, if it sees a point at a lower voltage point that yields more power it should demand a higher current output which will drop the voltage and cause the shaded panels to go into bypass mode. This is the theory that is also published in research papers. The deye/sunsynk mppt tracker can only lock on to the highest voltage power point, so even with one panel shaded the most I get is around 200W.

I have seen a couple of videos about shading and Fronius inverters able to implement the above theory. They are able to extract the power from a string if one (or more) panel is shaded. Unfortunately none of the software updates made any improvement to the Deye/Sunsynk inverter. They are probably not even aware that the MPPT needs to scan the whole voltage range and there are higher power points are available and I doubt they will ever code this into their firmware, they are too busy trying to keep the cloud service running. Good luck explaining this to them

So as a test, I split the one string of 8 modules into two strings of 4 each. 

As a comparative measurement - I attach two screenshots showing the total DC Input Power (i.e. total power produced by both MPPT's).

This one from 2023/06/15 is when all panels were connected in series on one MPPT:

And then for 2023/06/17 when the panels were split into two strings, one on each MPPT:

 

Production increased from 10.9kWh up to 12.9kWh... Same shading on panels, nothing changed as of yet here.

Less voltage on each string.

 

Anyway - thought I would post the update for anyone interested

I also had panels in a 8S configuration, I ended up changing them to 4S2P as the average output with all the shading is better. For about 2 months the sun's angle is so bad that every tree is a problem. For now I just decided to live with the problem. By August my output will triple again. I went from 15kWh average down to 5.5kWh

 

 

I have a similar situation with my two strings, but fortunately only in the month of June when the sun is at is lowest point, thing will start to improve towards the end of the month, tomorrow will be the worst day. winter solstice. herewith my PV in the morning.

Edited June 20, 20232 yr by Antonio de Sa

@rfl unfortunately this is an issue specific to the Sunsynk and Deye MPPTs.  They dont actually do Maximum Power Point Tracking at all, but rather Local Power Point Tracking.  I've been trying to motivate them to fix this issue in FW, so that the bypass diodes are used as intended.

All of their competitors (Vitec, Sungrow, SolarX, SAM, Fronius, Sofar Solar, SKE/ Huawei Luna, etc.) have special algorithms to handle partial shading - mostly by performing a periodic (eg. every 30min) global Voltage scan to find the maximum power point.  I can just imagine how many millions of kWhs goes missing every year because of this issue.  And the most shocking part is that most installers seem to think this is "normal shading behavior".

2 hours ago, cvschalk said:

@rfl unfortunately this is an issue specific to the Sunsynk and Deye MPPTs.  They dont actually do Maximum Power Point Tracking at all, but rather Local Power Point Tracking.  I've been trying to motivate them to fix this issue in FW, so that the bypass diodes are used as intended.

All of their competitors (Vitec, Sungrow, SolarX, SAM, Fronius, Sofar Solar, SKE/ Huawei Luna, etc.) have special algorithms to handle partial shading - mostly by performing a periodic (eg. every 30min) global Voltage scan to find the maximum power point.  I can just imagine how many millions of kWhs goes missing every year because of this issue.  And the most shocking part is that most installers seem to think this is "normal shading behavior".

Yes I fully agree! Someone that actually understands the theory how MPPT tracking works! +1 I will add my vote to any petition and support you. Would love to get my MPPT tracking on par with what I have seen from the Fronius inverters. Deye/Sunsynk is lacking and have been avoiding the issue, so people try work around the issue with optimisers. 

Here is a very nice explanation of "global" vs "local" MPPT tracking....

https://aurorasolar.com/blog/the-importance-of-modeling-global-maximum-power-point-tracking/

16 minutes ago, FixAMess said:

Here is a very nice explanation of "global" vs "local" MPPT tracking....

https://aurorasolar.com/blog/the-importance-of-modeling-global-maximum-power-point-tracking/

Thanks for a great link on maximum power tracking. 

I bet its not just a software fix. 

2 hours ago, Chris_S said:

I bet its not just a software fix. 

The MPPT varies the load(current) via software (probably with a high frequency duty cycle) and as a result the MPPT Tracker Voltage. The fix will most definitely be a software update.

It is probably not an easy algorithm to write especially if you have branches swaying in the wind and casting shadows on different sections of panels causing the max power voltage to change a lot but surely they must at least attempt something better than nothing at all

Has anyone with this issue noticed an improvement of late? Not sure if it is changing seasons and shading, or if one of the firmware upgrades actually made a difference, but am noticing a less dramatic differential on the shaded string than previously, during the time of partial shading. Then weirdly a difference in peak output in favour of the previously shaded string.

vs this previously

 

 

Edited October 9, 20232 yr by zak687

1 hour ago, zak687 said:

Has anyone with this issue noticed an improvement of late? Not sure if it is changing seasons and shading, or if one of the firmware upgrades actually made a difference, but am noticing a less dramatic differential on the shaded string than previously, during the time of partial shading. Then weirdly a difference in peak output in favour of the previously shaded string.

vs this previously

 

 

Not sure if you mean you still have the shading and output is higher. 

My one string that got shade starting at 12h00 before gave 2.5-3kWh a day and now with less shade due to sun position is now yielding 6kWh per day. 

String without shade pointing North is up from 7.5 to about 9.8kWh per day. Longer day and better angle of the sun. 

On 2023/05/21 at 12:21 PM, zak687 said:

Update: Still waiting on a response from both Sunsynk and the installer on a remediation from either, but I did try the approach of the OP (rfl) of cycling the DC power and sure enough the offending string went from 200w to 2000w with no change to the shading, which does kind of point to it being an issue on the inverter optimization side. 

Is there a safe way to cycle the DC without powering down the inverter - weary of cutting breakers under load?

Did you get a reply from Sunsynk about this? Mine is suffering the same fate. I have to drop the DC breaker and switch it on again and the PV output immediately jumps up to where it's supposed to be. Oddly enough if I drop the PV DC breaker connected to my one Sunsynk the PV connected to the second Sunsynk (parallel installation) drops as well. Surely this must be some kind of bug in the firmware?

⬇️

Edited October 15, 20232 yr by zak687

On 2023/10/09 at 10:11 PM, Scorp007 said:

Not sure if you mean you still have the shading and output is higher. 

My one string that got shade starting at 12h00 before gave 2.5-3kWh a day and now with less shade due to sun position is now yielding 6kWh per day. 

String without shade pointing North is up from 7.5 to about 9.8kWh per day. Longer day and better angle of the sun. 

Meaning I still have the shading, and output is both higher but also closer to what is expected in spite of the shading. Previously the string with the shading would produce 10x less power until the shading disappeared completely later in the day, whereas it is now closer to expected (proportionately less power in line with some bypassing due to shading).

On 2023/10/10 at 12:03 PM, p_i said:

Did you get a reply from Sunsynk about this? Mine is suffering the same fate. I have to drop the DC breaker and switch it on again and the PV output immediately jumps up to where it's supposed to be. Oddly enough if I drop the PV DC breaker connected to my one Sunsynk the PV connected to the second Sunsynk (parallel installation) drops as well. Surely this must be some kind of bug in the firmware?

Sunsynk asked me to check twice following firmware upgrades and both times did not see an improvement (this was some weeks prior to observing the change), but otherwise claim to be still working on it.

Hi,

I noticed the same thing with my Deye SUN-6K-SG03LP1-EU.

I spoke with the Deye service about this and they advised me to add optimizers 😐

But, since the latest firmware version:

HMI: 0000-C369
MAIN: 0-4385-1515

A new “MPPT multi-point scanning” option has appeared.

I haven't tested it yet but I hope it will help to have better behavior.