Searching \ for '[EE]:: Paralleling PV panels with differing illumi' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=paralleling+panels
Search entire site for: ': Paralleling PV panels with differing illumi'.

Exact match. Not showing close matches.
PICList Thread
'[EE]:: Paralleling PV panels with differing illumi'
2016\04\27@203224 by RussellMc

face picon face
part 1 1853 bytes content-type:text/plain; charset="utf-8" (decoded base64)

To: Whoever wants to comment.

Answer to at least 1. and 2. would be interesting.

______________


Q: If you had PV panels which *each were evenly illuminated *but the
illumination level of each varied (eg 100%, 80%, 50% of 1 sun) and you
placed them in parallel, what would you expect the output of the low light
panels to be compared to what it would be if they were optimally loaded?
This could happen with eg panels pointing at different sky areas.

Note that, within one panel,  no cells are shaded more than any other - ie
all series strings have ~ equally illuminated strings. Only whole panels
have differing illumination levels.

Assume to make it 1/2 answerable that  the fully illuminated panel is
working at its max power point so Vout suits it perfectly and other panels
are added and the voltage does not change.
(This could happen with eg a stiff battery load or an inverter input with
feedback control etc). The arrangement is artificial but does not overly
change the answer , probably.

Assume panels behave like this. This is not meant to set specific V or I
curves but to just show general shape.
Lines are for insolation (sun energy) decreasing in 5% steps from top
(100%, 95%, 90% ...)
Blue dots are maximum power points at that insolation level.

[image: Inline images 1]
Questions:

For question marked Q: at top of page:

1. What you would have said off the cuff before I asked the question.

2. What you would say off the cuff now I have asked the question.

Only if excessively keen:

3. What you would say after having done some figuring.
Here is my lonnnnnnnng answer (with summary at start).
Don't look at it until you have answered 1. & 2. above (if then).


        electronics.stackexchange.com/a/230889/3288


    Russell

part 2 8683 bytes content-type:image/jpeg; name="PV cell VI curve energies_crop1_W235_j70.jpg" (decode)


part 3 197 bytes content-type:text/plain; name="ATT00001.txt"
(decoded base64)

--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

2016\04\27@213939 by Justin Richards

face picon face
Did you mean "... would be if they were optimally loaded? ..." or "... would
be if they were optimally illuminated?..."

I have considered the question of series string performance when one panel
is not as optimally illuminated as the others but never in parallel.

Assuming I have understood the question

Q1 Off the cuff ... The less optimally illuminated

at 100% will deliver 1/2 or 50% the total power

at 80% will deliver perhaps 5%

at 50% will be seen as a load by the other panel and become a hindrance and
will deliver -10%

Q2 Very hard for me to separate because before off the cuff in the before
time I would still consider the question.

I am very curious and will now read the answer

Justin


On 28 April 2016 at 08:31, RussellMc <spam_OUTapptechnzTakeThisOuTspamgmail.com> wrote:

{Quote hidden}

-- http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist
.

2016\04\27@215247 by RussellMc

face picon face

On 28 April 2016 at 13:39, Justin Richards <.....justin.richardsKILLspamspam@spam@gmail.com>
wrote:

> Did you mean "... would be if they were optimally loaded? ..." or "...
> would
> be if they were optimally illuminated?..."
>
> ​I meant optimally loaded (as written).
ie a 50% illuminated panel when optimally LOADED will make ABOUT 50% output.
But, as V_50%_optimal <>Vmp , when connected to a stiff bus at Vmp some
reduction in power is expected.
​


> I have considered the question of series string performance when one panel
> is not as optimally illuminated as the others but never in parallel.
>
> ​Yes. Series is easier to conceptualise.

​


{Quote hidden}

​Somewhat like std understanding and somewhat like my prior - although
panels draw far less back current than may be expected in most cases. ​

​But​



> Q2 Very hard for me to separate because before off the cuff in the before
> time I would still consider the question.
>
> I am very curious and will now read the answer
>
> ​Hopefully it makes sense and also changes your understanding.


   R
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

2016\04\27@223910 by Harold Hallikainen

flavicon
face
Because the maximum power point appears at about the same output voltage
independent of illumination, it appears you'd do fine connecting the
panels in parallel, since this would force them all to this MPP voltage.
See http://www.linear.com/solutions/4445 for the relationship between MPP
and loaded voltage.

Harold

-- FCC Rules Updated Daily at http://www.hallikainen.com
Not sent from an iPhone.
-- http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist
.

2016\04\27@235251 by RussellMc

face picon face

On 28 April 2016 at 14:19, Harold Hallikainen <haroldspamKILLspammai.hallikainen.org>
wrote:

> Because the maximum power point appears at about the same output voltage
> independent of illumination, it appears you'd do fine connecting the
> panels in parallel, since this would force them all to this MPP voltage.
> See http://www.linear.com/solutions/4445 for the relationship between MPP
> and loaded voltage.
>
>
​Yes.
I had included a comment on that concept ​near the bottom of my answer.

Note that they claim the idea is patent pending.
I'd be surprised if that would stick, but maybe.

Note also that it is less good than it could be.
What is a better match is to use an "mx+c " curve fit.
Or rather

Vtarget = Vx + Iload x k

Vx and k are chosen to place the reference voltage on a lowest error line
in their fig 3
Something like Vref = 15 + 2 x Iload in their fig 3.

It would be easy to fit a non linear line if desired.
Especially easy when implemented in software.

This is not the first time I've written this anywhere. but note that this
message is public domain.



                Russell
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

2016\04\28@001449 by James Cameron

flavicon
face

On Thu, Apr 28, 2016 at 03:51:30PM +1200, RussellMc wrote:
> On 28 April 2016 at 14:19, Harold Hallikainen <.....haroldKILLspamspam.....mai.hallikainen.org>
> wrote:
>
> > Because the maximum power point appears at about the same output voltage
> > independent of illumination, it appears you'd do fine connecting the
> > panels in parallel, since this would force them all to this MPP voltage.
> > See http://www.linear.com/solutions/4445 for the relationship between MPP
> > and loaded voltage.
> >
> >
> ​Yes.
> I had included a comment on that concept ​near the bottom of my answer.
>
> Note that they claim the idea is patent pending.

Position of "patent pending" was just after "input voltage
regulation", so take it to mean their implementation of IVR rather
than the outcome of the regulation itself.  C1 acting to smooth
whatever it is they are doing in the chip, which from datasheet block
diagram looks to be something like PWM of Vin to SW transistor
switching.

http://cds.linear.com/docs/en/datasheet/3652fe.pdf

> I'd be surprised if that would stick, but maybe.

Maybe the way the PWM is derived, IANA(P)L.

{Quote hidden}

--
James Cameron
http://quozl.netrek.org/
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

2016\04\28@033345 by Justin Richards

face picon face
Does this imply that two separate arrays, one facing East the other West
(due to limited roof realestate) could be connected in parrallel without
the need for a dual tracking inverter with only a small performance hit.
-- http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist
.

2016\04\28@045335 by RussellMc

face picon face

On 28 April 2016 at 19:33, Justin Richards <EraseMEjustin.richardsspam_OUTspamTakeThisOuTgmail.com>
wrote:

> Does this imply that two separate arrays, one facing East the other West
> (due to limited roof realestate) could be connected in parrallel without
> the need for a dual tracking inverter with only a small performance hit.
>
> ​Sort of, maybe.
Close to "yes in many cases"

If part of a panel becomes shaded ​then either
- the max current for all cells in the same series string is the current
that the shaded cell generates
- or if the shaded cell has protection diodes then for N cells in series
and 1 shaded cells thyen
current max is as before but
Vpanel_now  = Vpanel x n/(n-1) - 1_diode_drop

For panels illuminated evenly but at 2 different levels.

Working through my stack exchange answer, for 100% and xx% illuminations,
down to about xx >= 50% it looks fairly benign.
For ery low xx  it can still be remarkably good.
In my 2nd examples, for 20% insolation the 20% panel makes 79% of the
current it would at optimum but at aboyt 43/39ths the voltage so power drop
is
79% x 43/39 = 87% of the power it would otherwise make.

This is if the 100% panel still works at the old MPP.
Odds are the combination has a different Vmp and the end result will be
BETTER than calculated above.

As xx insolation falls there comes a point that Voc is <= the operating
voltage of the 1st panel and you get nothing.
In my SE answer that occurs at about 5% insolation (bottom line shown is
10%) so you don't lose much.


Russell
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

2016\04\28@121150 by RussellMc

face picon face
part 1 3536 bytes content-type:text/plain; charset="utf-8" (decoded base64)

Here's a (rough) worked example on a real product

Panel here


www.mitsubishielectricsolar.com/images/uploads/documents/specs/MLU_spec_sheet_250W_255W.pdf

Graphed result here


dl.dropboxusercontent.com/u/30808964/PV%20panel%20mitsubishi%20V43.jpg
         1362 x 1544 resolution.

____________________

200 x 227 version

[image: Inline images 1]

____________________________

This is only at 100 90 80 70% full sun but shows what happens.

Green circles show optimum and paralleled current at the selected light
leve.
V is set to Vmp at full power as before.

IF I did it right then results are "interesting".
900 W/m^2 loses little
800 loses rather more - about say 0.2/6.3 or about a minimal 3 %
BUT 700 W/m^2 loses LESS than 800 W/m^2.
Their lines or mine may be wrong.

Method.
Drop vertical from peak power point on power-V curve to relevant V-I curve.
This is mpp for that % insolation.
Draw line horizontal left to show optimum I loaded.

For 100% curve draw line (red) vertically downward to x axis.
This is Vmp at 100% light.

>From intersection of red line and white VI lines draw horizontal lines
(thin red) to Y axis to get Ixx at Vmp100.

Compare differnces of related black and red lines.

HOWEVER - just realised - just looking where the vertical red line
intersects the CYANish power-V curves shows how much loss you get - you can
see peak power at x% insolation and off-peak power when paralleled.
Clear and easy.

AND you can see that the 800 W/m^2 curve loses more power than the 700
W/m^2 one does (!)



SO

Simple method (Agh!)

Draw line vertical from power-V curve for 100% sun to X axis.

Intercepts with other power-V curves show power loss in this case.

QED.

E&OE.

___________________




On 28 April 2016 at 20:52, RussellMc <apptechnzspamspam_OUTgmail.com> wrote:

> On 28 April 2016 at 19:33, Justin Richards <@spam@justin.richardsKILLspamspamgmail.com>
> wrote:
>
>> Does this imply that two separate arrays, one facing East the other West
>> (due to limited roof realestate) could be connected in parrallel without
>> the need for a dual tracking inverter with only a small performance hit.
>>
>> ​Sort of, maybe.
> Close to "yes in many cases"
>
> If part of a panel becomes shaded ​then either
> - the max current for all cells in the same series string is the current
> that the shaded cell generates
> - or if the shaded cell has protection diodes then for N cells in series
> and 1 shaded cells thyen
> current max is as before but
> Vpanel_now  = Vpanel x n/(n-1) - 1_diode_drop
>
> For panels illuminated evenly but at 2 different levels.
>
> Working through my stack exchange answer, for 100% and xx% illuminations,
> down to about xx >= 50% it looks fairly benign.
> For ery low xx  it can still be remarkably good.
> In my 2nd examples, for 20% insolation the 20% panel makes 79% of the
> current it would at optimum but at aboyt 43/39ths the voltage so power drop
> is
> 79% x 43/39 = 87% of the power it would otherwise make.
>
> This is if the 100% panel still works at the old MPP.
> Odds are the combination has a different Vmp and the end result will be
> BETTER than calculated above.
>
> As xx insolation falls there comes a point that Voc is <= the operating
> voltage of the 1st panel and you get nothing.
> In my SE answer that occurs at about 5% insolation (bottom line shown is
> 10%) so you don't lose much.
>
>
>  Russell
>
>

part 2 12313 bytes content-type:image/jpeg; name="PV panel mitsubishi V21z_w200.jpg" (decode)


part 3 197 bytes content-type:text/plain; name="ATT00001.txt"
(decoded base64)

--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

2016\04\28@143240 by embedded systems

face picon face

Russel you've missed the best question here:
At maximum illumination how will influence the cell efficiency it's
temperature?
maximum illumination = maximum heat...

Vasile

On Thu, Apr 28, 2016 at 7:11 PM, RussellMc <KILLspamapptechnzKILLspamspamgmail.com> wrote:

{Quote hidden}

--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

2016\04\29@064456 by RussellMc

face picon face

On 29 April 2016 at 06:32, embedded systems <TakeThisOuTpiclist9EraseMEspamspam_OUTgmail.com> wrote:

> Russel you've missed the best question here:
> At maximum illumination how will influence the cell efficiency it's
> temperature?
> maximum illumination = maximum heat...
>
> ​I'd not missed it (believe it or not) but I decided not to mention it as
it added extra variables which clouded the main point, which is the ability
of off optimum cells to rise in voltage to the initial Vmp with minimal
power loss​.
I looked (again) at temperature versus power outputs as I skimmed through
data sheets the other day looking for ones with both V-I and V-Power curves
on the same graph.

As you correctly note, temperature has a significant effect. Maybe 5%-10%
loss from the nominal values which are almost always specified at 25C and
the actual.

Long ago (hmmm - 2007)  I achieved an agreeable gain in output (from a then
30 year old! 50 W BP panel) by running a very thin film of water down the
front face. Water flow was reduced to the point just above where a film
could not be maintained. Presumably-slight losses due to water optical
losses (and possibly anti-reflective gains) + cooling gave a net gain.


In most cases this would not be practical but it was interesting. It's
uncertain whether attempts at passive rear air cooling would be
economically justifiable. With continuing low panel costs and implications
for mounting forces I'd guess probably not, but ... .


      Russell
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
mailman.mit.edu/mailman/listinfo/piclist

More... (looser matching)
- Last day of these posts
- In 2016 , 2017 only
- Today
- New search...