Thanks. Let's see if I understand correctly:
So the goal here, if I understand it correctly, is to prevent the Panels from climbing too high in voltage and thus having the current fall to 0?
That is the basic functionality, although target isn't current > 0, target is to maximizecurrent x voltage = power.
en.wikipedia.org
That curve shows current decreases (on an inverse exponential curve), reaching 0A at Voc.
The curve you need is W = V x I vs. V
community.victronenergy.com
Note that maximum power occurs at around 85% of maximum voltage.
MPPT starts at Voc, pulls current lower while calculating wattage, finds the peak, dithers back and forth to stay at the peak.
The method the MPPT uses to do that is to adjust impedance until it finds a maximum?
Typically a buck converter, increases current by varying switching duty ratio to find that maximum.
Some MPPTs halt adjustment once a maximum is found so they will find local maxima instead of global?
Exactly. Draw more current until power starts to decrease again, dither back and forth to stay at the (local) maxima.
If at lower voltage there is a dip to lower power, then climbs again to an even higher power, it never finds that.
In my system I'm intending to put 10 panels in series onto MPPT1. I'm intending to put a different 10 panels in series into MPPT2. The current design has 50.1% of MPPT1's panels in shade, and 50.2% of MPPT2's panels in shade.
That works, 10 panels MPPT 1 sometimes full sun, sometimes 5 panels shaded.
10 panels MPPT 2 similar situation.
So long as 5 x Vmp is within MPPT operating voltage, should work.
But, make sure no panels get shaded while others in series get full sun. That will push full current through bypass diodes, can cause diodes to fail and panel to melt or crack. Try to have shadows only fall on panels with light is at about 45 degree angle or beyond.
I do not have any panels in parallel in my design (although I might do that in the future). I haven't done the math yet, but I can see how trying to set impedance for two things in parallel could be difficult if the things in parallel are not close together in their impedance.
Two strings in parallel is OK if all panels are unshaded, even though sun is off-angle for panels of one string.
Note in curves that greatly reduced sun only reduces Voc and Vmp a percentage.
Ok for partial shade too, probably best if no more than 15% difference in shading so Vmp remains similar.
This is where you can try to arrange for shade to start impacting panels on both strings at same time.
Also note arrangement of cells and bypass diodes in panels. Shadow on end (short side) of panel would take out all 2 or 3 sub strings. Shadow on long side would only take out 1/3 or 1/2 of its production. If multiple rows of panels can shade each other, "landscape" orientation should reduce loss due to partial shading.
I think what was proposed to me earlier in this forum is that I should do my best to group all shaded panels into MPPT1, and all full-sun panels into MPPT2- is that correct? Or is it the opposite?
Having one string unshaded means even a dumb MPPT should work OK. In this case better to have all shaded panels in one MPPT.
If 5s panels causes voltage to drop below where MPPT works, then with all shade encroaching on one string, when 25% of your 20 panels are shaded, one entire string stops producing and 10 panels remain.
If shade encroaches on on both strings equally, then when 25% are shaded, that leaves 14 or 16 panels producing. Only when 50% of panels are shaded does production drop to 10 panels.
Sun good, shade bad.
Something I don't understand: If the MPPT is adjusting impedance, wouldn't it simply adjust to maximize the highest output voltages in your string (sunny)? In the cell IV curve, current is flat at lower voltages so it seems like an MPPT's local maxima should also be the global maxima for series-strings that include shaded (lower voltage) panels?
See above. Maximum W = V x I, not maximum V (which occurs at I = 0A, W = 0)
Essentially for a series-string, the MPPT is going to see a ONE voltage input, and a current that goes with that after impedance is set. As the impedance is adjusted, current will vary. Current will exponentially fall off (according to series IV chart above) if impedance is too low. If impedance is too high, current will slowly/linearly fall off according to series IV chart. Seems relatively easy to find a global maximum under those conditions, no?
Yes, current falls of as voltage increases.
If MPPT sweeps voltage from Voc all the way down to zero, it will pass through local and global maxima. Then it can return to global maxima and sit there for a while.
It will dither and stay on top of the maxima it is at. But if lighting changes that may become a local, not global maxima. And there it sits.
So, after a while, a good MPPT will sweep again to see if there is a new global maxima.
Wouldn't bypass diodes eliminate the problem of having some shaded panels in a string?
Bypass diodes make it possible for a 10s string to put out the power of 9s string when one panel is shaded.
Without the diode, current would drop to zero.
With some light falling on the panel, it may produce 1A instead of 10A, and MPPT might not pull voltage low enough to benefit. Requires decent MPPT algorithm with global sweep.
I think (but haven't yet used) half-cut panels which are internally wired 2p3s. If shade falls on a single cell, current drops from 10A to 5A. Good solid power being produced, but instead of 1/2 power it could produce 2/3 power at lower voltage. A 10s string would drop to half current (and somewhat higher voltage). This would reduce power to 1/2 (or a bit more) when it could have been 29/30th. I don't know for sure, but think a dump MPPT would be easily confused.