# How to implement mppt algorithm

Discussion in 'The Projects Forum' started by ali zeeshan, Mar 30, 2014.

1. ### ali zeeshan Thread Starter New Member

Feb 7, 2014
18
0
hi there everybody i want to make a solar panel based system implementing mppt i am a student of electronics and have a bit of knowledge about different algorithms such as P&O, Inc conductance etc... the problem is am not clear on how to physically implement this... have a little idea that need controller to control switching of mosfet but the overall design is not clear nor the component list please help me with this...
one more thing i am interested in testing the design on matlab (simulink) first and have found a model for solar panel but i don't have idea that what components are required..... i would gladly accept all kind of advice and support i can get from here.. Thanks in advance

2. ### Papabravo Expert

Feb 24, 2006
10,338
1,850
Based on reading the wiki on MPPT, I suggest you start with the problem of measuring voltages and currents accurately over relatively small ranges.

http://en.wikipedia.org/wiki/File:Solar-Cell-IV-curve-with-MPP.png

Ideally, for a single cell, you want at least 8 bit resolution over the voltage range {0.3 V:0.5 V} which is 781 microvolts per step; a tall order. You also want at least 8 bit resolution over the current range of {5 mA: 40 mA} or 136 microamps per step; also a tall order. If you can't measure these things accurately then no matter what control algorithm you pick, it won't be very effective.

Once you get past the challenge of measuring voltage and current you have to understand how to vary the load so that you can move voltage and current to where you want them to be. It may involve MOSFETS or it might involve other things.

You might find the following paper of interest.

http://lib.tkk.fi/Dipl/2010/urn100399.pdf

And this one

http://ww1.microchip.com/downloads/en/AppNotes/00001521A.pdf

Those were the top 3 hits. I get paid big bucks because google is my friend; you can too.

Last edited: Mar 30, 2014
3. ### ErnieM AAC Fanatic!

Apr 24, 2011
7,435
1,625
Measuring each and every cell is a fruitless task, literally! Your measurements will consume a significant portion of the power produced.

Just read the total output voltage and the output series current for all cells.

There's no need for extreme accuracy in these measurements: as long as any two given measurement A reading A' and B reading B': if A' > B' only when A > B then the measurements are acceptable.

4. ### THE_RB AAC Fanatic!

Feb 11, 2008
5,435
1,305
It's not worth doing 12v->12v MPPT unless you can build a SMPS buck converter with >95% efficiency. Regardless of the control algorithm.

Even with 95% efficiency you will barely get more power into the battery than a direct connection of solar panel to battery.

Bernard likes this.
5. ### Papabravo Expert

Feb 24, 2006
10,338
1,850
If each cell has it's own unique MPPT then anything you do on the average of series and parallel connected cells will imply that some are operating where they should and some (most) are not. How is this an improvement? BTW I did not read the thesis in great detail.

6. ### ErnieM AAC Fanatic!

Apr 24, 2011
7,435
1,625
It is an improvement over suggesting a measurement that simple is not possible. A "solar panel" is not a solar cell, it is an array of such cells, with internal solar cells wired in parallel and series to build up it's rated voltage and current at the resulting two terminals.

Hence an individual cell is simply not accessible.

Additionally, exactingly accurate measurements are not necessary as you are not measuring the power, you are maximizing it. So relative measurements are just fine, as "more" is still "more."

Last edited: Apr 1, 2014
7. ### Papabravo Expert

Feb 24, 2006
10,338
1,850
Yes, but the fundamental question is:
Is optimization applied to a collection of cells arranged in a particular way, the same as optimizing the individual cells. The applicability of this hypothesis is not self evident to me.

As always, extraordinary claims require extraordinary evidence.

8. ### ErnieM AAC Fanatic!

Apr 24, 2011
7,435
1,625
Perhaps you are unfamiliar with the term "industry standard." It is the way that people who actually do things do these things.

Go look at the specification of *any* solar panel. Count the number of leads on the output connector.

That number is two.

Next check out the sealing characteristics and note such terms as "waterproof" or perhaps "hermetic."

Either term indication "NO USER ACCESSIBLE COMPONENTS INSIDE."

Beyond this, I refuse to explain the bleeding obvious to those who cannot see.

9. ### Papabravo Expert

Feb 24, 2006
10,338
1,850
For all your bluster you did not answer the question. The basic idea of the MPPT algorithm is to optimize the system characteristics so that maximum power transfer is achieved. In the wiki article the algorithm is explained in terms of an individual cell. I understand how panels are packaged and I am familiar with the term "industry standard" having written a few of them myself. So I repeat: Is MPPT a valid algorithm when applied to a collection of cells where you OBVIOYSLY cannot optimise each cell to operate at the MPPT point. The answer is not obvious, so I think you should answer the question or hold your peace.

10. ### alfacliff Well-Known Member

Dec 13, 2013
2,449
428
wouldnt it be better to track the solar panel to the sun? maximum out put that way.

11. ### ErnieM AAC Fanatic!

Apr 24, 2011
7,435
1,625
A tracking mount increases the output for a price: one very expensive motion tracker and increased insurance to replace the entire panel array next time it gets windy. While an aimed panel will then be producing more potential power one still has to harvest it in an efficient manor.

MPPT is designed to extract the maximum power at any given instant. A solar cell will produce a relatively constant voltage for any current up to some (very variable) maximum point, then the voltage drops right down to zero as you continue to increase the current to the max (which is the short circuit current).

As power is voltage time current there exists one point where it is maximum, and this point can and does change. The algorithms all vary the current to find that max power point.

12. ### THE_RB AAC Fanatic!

Feb 11, 2008
5,435
1,305
The answer is yes, the MPPT algorithm can be valid for the entire collection of cells.

I designed a commercial MPPT controller back in the days when the technology was rare, before you could buy them for peanuts on Ebay.

Solar cells operate as current sources with the current totally based on insolation (sunlight strength). The noload voltage would be infinite but there is an inherent zener effect in the silicon so a 12v panle noload voltage is about 22v. That is the voltage where the internal zener effect current drain is matching the cells current source. Into a short circuit you can measure the cells sourced current.

Assuming you have a high efficiency DC->DC converter the MPPT algorithm can operate the panel up near the zener "knee" at the point where V*I result is highest, before the zener starts consuming too much of the power. Usually around the 17-18v mark.

Like I said before the big challenge is not the MPPT algorithm but the ability to convert 17v to 14v at an efficiency significantly higher than a direct connection 14v to 14v.

Hope that helps.

Papabravo likes this.
13. ### Papabravo Expert

Feb 24, 2006
10,338
1,850
Thanks. That was all I was looking for.