datasheets.com EBN.com EDN.com EETimes.com Embedded.com PlanetAnalog.com TechOnline.com
Events
UBM Tech
UBM Tech

Design Article

# Selecting a solar energy conversion method

## 7/3/2012 10:15 AM EDT

8. Providing MPPT for each panel
An improved method of solar system design is to use micro-inverters. A micro-inverter is a 250W inverter that is connected to each panel. MPPT is performed by the micro-inverter at the panel level. Figure 11 shows a system consisting of thirty micro-inverters, one per panel. The micro-inverters AC outputs are connected together and properly phased with the AC line.

Figure 11

Micro-inverters are fairly complex electronic products. A picture of one from STMicroelectronics for evaluation purposes, with its components clearly visible, is shown in Figure 12.

Figure 12

A simpler method of implementing a photovoltaic power system with MPPT is to use an active power optimizer such as the SPV1020. An overall system diagram is shown in Figure 13.

Figure 13

In this case one active power optimizer is connected to each panel. It boosts the panel output voltage and performs the MPPT function while doing so. Panel output voltage must be at least 6.5VDC. The SPV1020 output voltage can be as high as 40VDC. A typical value is 35VDC as shown in Figure 13. The active power optimizer utilizes the Perturb and Observe algorithm until it finds the maximum power point on the power vs. voltage curve as shown in Figure 7. This optimizer measures the input power to determine the panel’s Vmp. There are other types of MPPT converters available but they assume Vmp is a fixed percentage of Voc. This could be case under one specific operating condition and thermisitors are required to approximate the change in Vmp with temperature. The SPV1020 makes none of these assumptions. It measures input voltage and input current to determine the actual input power in setting the maximum power transfer operating point. Figure 12 shows the simple external connections. The panel is connected through the boost inductor at the Lx input and the load is connected to Vout. No other source of power is required. The resistor divider connected at the panel output senses the chip input voltage for MPPT purposes. Input power is determined by sensing the current through the main MOSFET switch and multiplying its value by the input voltage. The resistor divider connected at Vout sets the value of the output voltage.

Figure 14

The SPV1020 is an interleaved four channel converter. Figure 14 shows one of the four switching channels. Of particular interest is its 320W the power handling capability. This is accomplished in the small PowerSSO-36 package by splitting the power handling components into the four channels.

The 4-phase interleaved topology as shown in Figure 15.

There are four interleaved switching sections interleaved every 90 degrees. This diagram shows a single panel and single load, and how the four switching section are connected. Each section has its own inductor. The switch and diode pictured are both MOSFETs with low Rds on. At the default switching frequency of 100kHz, each section operates at 25kHz. The SPV1020 also integrates four zero crossing blocks, one for each branch. Their role is turn off the related synchronous rectifier to prevent reverse current flow from output to input.

In order to guarantee a correct power-up sequence, the converter initially operates in burst mode. When the input voltage is greater than 6.5 V, the converter sequentially activates each of the four phases. Initially, only phase 1 starts to work in burst mode, charging the inductor only for one cycle over 15 cycles. Then the duty cycle is progressively increased until phase 1 is switched on at every cycle at the default switching frequency of 100 kHz. After phase 1 has reached its steady-state condition, the other phases are progressively switched on in the following sequence: phase 3, phase 2 and, lastly, phase 4. If lower power than 320W is required, it may be possible to use only two of the four phases, eliminating the cost and space requirements of two inductors.

A major advantage of interleaved architecture is low ripple. Assuming a resistive load, the output voltage ripple is proportional to the output current ripple. In interleaved-4 phase architecture, total output current is the sum of the four currents flowing in each inductor. Since each phase carries one fourth the total current, for a given value of inductance, the peak to peak ripple would be one fourth the value of a single phase architecture system.

chrisrodgers

7/3/2012 4:06 PM EDT

A very interesting and article, thanks very much!

I have a question.

For a set of conditions, it seems that a given solar cell will have a finite amount of power to deliver. A fixed DC load connected directly to the solar cell will form a voltage divider from the Thevenin equiv. voltage source (E) between Rs and the load (RL), assuming Rp is very large.

Let's say RL is quite a bit less than Rs. So the loaded solar cell voltage will drop when connected directly. Ideally, you'd like a way to make RL appear to increase to accomplish Max. Power Transfer (MPP). This would result in raising the load voltage (VL) to a maximum of one half the open circuit Thevenin voltage (E/2).

Interposing the SPV1020 boost converter between the solar cell and RL will make the load impedance appear matched to the solar cell (RL = Rs), boosting VL and hence the load current.

Maybe I'm not seeing something here, but doesn't this violate conservation of energy? Or is MPP accomplished only "piece-wise", during each PWM pulse of the boost converter?

Thanks for any light you can shed on this.

sudhakar_amberroot

7/3/2012 11:54 PM EDT

Thevenin cannot be applied here because of the nonlinear device being present which is the parallel diode array. So the maximum point will not be E/2 but much higher since the current through the diode will come down with reduction of the terminal diode.

chrisrodgers

7/4/2012 7:22 PM EDT

In the article it states "Since the equivalent circuit of a solar panel is represented by a current source with parallel and series resistances, the Thevenin equivalent circuit can be shown as a voltage source with a single series res." This seems to say that Thevenin is indeed applicable here. But my point is not about Thevenin per se, nor about the maximum achievable load voltage, but rather the question of conservation of energy. As you say, as the load voltage rises, "the current through the diode will come down". This is a result of P=EI. You can change the voltage, but the power from the solar cell does not change. So I still wonder how, for a fixed DC load resistance, it's possible to increase the load voltage using a boost converter. More load voltage will result in more load current, and therefore more load power. Any comment on this, Ed?

twarner250

7/5/2012 10:19 AM EDT

The power delivered by the cell (or array) is function of its output voltage, it is not fixed. As the cell voltage rises, the current in the internal diode rises, leaving less of the photo current for the load.

Any power not taken by the load will be converted to heat.

HarrisMarcus

7/9/2012 4:36 PM EDT

You have stated “As the cell voltage rises, the current in the internal diode rises, leaving less of the photo current for the load”. This is not so. If you follow the panel I-V curve shown in Figure 6 when the cell voltage increases the cell current decreases. The maximum cell voltage is the open circuit voltage, Voc, where the current is zero.

jg_

7/5/2012 6:30 PM EDT

Thevenin is a poor approach, a better model is a current source, with a parallel clamp diode, and better curves would include constant power profiles (V*I = constants) which makes the MPPT seeking easier to visualize.

HarrisMarcus

7/9/2012 4:35 PM EDT

The maximum power available from the panel does not change but the power that is actually extracted from the panel does change and it is a function of the load resistance RL. This maximum power point occurs when RL = Rs. The boost converter decreases the value of RL by adjusting its duty cycle. The input resistance seen by the panel is RL x (1-du)^2 where du is the duty cycle internally set by the SPV1020. The SPV1020 duty cycle is internally adjusted so that RL x (1-du)^2 equals the panel output resistance Rs.

HarrisMarcus

7/12/2012 1:38 PM EDT

The maximum power available from the panel does not change but the power that is actually extracted from the panel does change and it is a function of the load resistance RL. This maximum power point occurs when RL = Rs. The boost converter decreases the value of RL by adjusting its duty cycle. The input resistance seen by the panel is RL x (1-du)^2 where du is the duty cycle internally set by the SPV1020. The SPV1020 duty cycle is internally adjusted so that RL x (1-du)^2 equals the panel output resistance Rs.

Brakeshoe

7/6/2012 4:14 PM EDT

Yes, the Thevenin model applies in figure 2, as you are looking back into the terminals of the source. What's more, if you apply a variable resistive load, you will get a linear graph of voltage vs current.

chrisrodgers

7/7/2012 10:40 AM EDT

I see that my understanding is slowly beginning to clear. I found the found Fig. 3.7 on pg. 24 of the following article to be helpful:

https://digital.library.txstate.edu/bitstream/handle/10877/3171/fulltext.pdf

I think what was bothering me was that the boost converter cannot increase the load voltage to any arbitrarily high level. It is necessarily limited by the PV maximum power curve for a given set of conditions (illumination, temp., etc.) This Fig. (3.7) seems to bear that out that.

HarrisMarcus

7/9/2012 4:37 PM EDT

The article you provided is very thorough and detailed. Thank you. You cannot get more power out than you have coming in.

HarrisMarcus

7/9/2012 4:34 PM EDT

You are correct. The intent of using the Thevenin equivalent is to show that whatever the value Rs even though non-linear, RL must match it to provide maximum power transfer.

HarrisMarcus

7/9/2012 4:33 PM EDT

RL will normally be much greater than Rs.

Interposing the SPV1020 boost converter between the panel and RL decreases the load resistance seen by the panel. The load resistance seen by the panel Rin = RL x (1-du)^2 where du is the duty cycle internally set by SPV1020. The duty cycle is set by the SPV1020 so that RL x (1-du)^2 = Rs, the source resistance of the panel.

HarrisMarcus

7/12/2012 1:37 PM EDT

RL will normally be much greater than Rs.

Interposing the SPV1020 boost converter between the panel and RL decreases the load resistance seen by the panel. The load resistance seen by the panel Rin = RL x (1-du)^2 where du is the duty cycle internally set by SPV1020. The duty cycle is set by the SPV1020 so that RL x (1-du)^2 = Rs, the source resistance of the panel.

dafi

7/3/2012 8:29 PM EDT

Yes an interesting article indeed. Thank you.

I have two questions:
1. what level of monitoring is needed to detect a failure in one panel/optimizer? It would seem that without some monitoring a failure could occur in one optimizer and you would not know about it at teh central inverter.
2. how applicable is the use of optimizers for an off grid (battery) system? I am assuming they would still serve their purpose by delivering more power to a DC-DC inverter/battery charger.

HarrisMarcus

7/9/2012 4:39 PM EDT

You would need to monitor the panel output voltage and input voltage.. This can be done through the SPI bus in the SPV1020. The SPI bus provides a direct measurement of input voltage, input current and duty cycle. Output voltage can be computed by Vout = Vin/(1-du).

HarrisMarcus

7/9/2012 4:40 PM EDT

Optimizers can be used for batteries as well as AC grid systems. STMicroelectronics has an evaluation board with P/N STEVAL-ISV005V1 that uses the SPV1020 and SEA05 constant voltage constant current controller to charge a 240W lead acid battery.

dave_dne

7/3/2012 9:16 PM EDT

The enphase M215 data sheet shows 208/240VAC output. But in this article "Power loss due to wiring" analysis indicates 115VAC output from the M215. Ed are we looking at the same thing here?

And the M215 has remote monitoring features.

On the 'cons' side - the M215 is does have a narrow min/max start voltage of 22/45VDC. I don't work for enphase but I did give their micro inverters a detailed review for a home PV system.

And lastly, what about potential problems with DC arc faults using a 350VDC voltage to the central inverter? Thought I read somewhere that NFPA/NEC was now addressing this issue? Anyway, I'd like to see something mentioned about any future NEC code requirements for high (~300VDC+) DC voltages to a central inverter.

HarrisMarcus

7/9/2012 5:45 PM EDT

We are looking at the same thing. The article compared a microinverter delivering 115VAC, not necessarily Enphase which delivers a higher voltage, to a higher voltage, lower current DC system.

The SPV1020 has performance monitoring features available through its SPI bus that include input voltage, input current and duty cycle. Output voltage can be computed from input voltage and duty cycle.

dave_dne

7/10/2012 12:33 PM EDT

twarner250

7/5/2012 10:18 AM EDT

The power delivered by the cell (or array) is function of its output voltage, it is not fixed. As the cell voltage rises, the current in the internal diode rises, leaving less of the photo current for the load.

Any power not taken by the load will be converted to heat.

HarrisMarcus

7/12/2012 1:42 PM EDT

You have stated “As the cell voltage rises, the current in the internal diode rises, leaving less of the photo current for the load”. This is not so. If you follow the panel I-V curve shown in Figure 6 when the cell voltage increases the cell current decreases. The maximum cell voltage is the open circuit voltage, Voc, where the current is zero.

green_is_now

1/23/2013 4:44 PM EST

you misinterpreted him you are both saying the same thing.

rroy22520

7/5/2012 11:08 AM EDT

What about delivering solar power not to the AC grid but to the local loads? MPPT absolutely must be considered for moving and non moving cells and reflector. But also the infrastructure for intelligent load switching/engagement to optimize the local loads so they put less demand on the grid. Solar in the desert is great if you can build a factory (a suitable load) underneath the arrays so the transmission distance is not the issue, getting the raw materials in and the finished product out.

http://www.lulu.com/spotlight/poconoarmchairreview

7/6/2012 1:12 AM EDT

Doesn't maximum power output fall drastically over the first few months? That means, it seems to me, that the control circuits have to be designed to handle outputs that are achievable for only a small percentage of the array's life, and only early on. After that, isn't it true that the control circuits will never see that level of output again?

http://www.lulu.com/spotlight/poconoarmchairreview

7/6/2012 1:14 AM EDT

It would be nice if solar panels could be sold "broken in" so that their initial degradation would not complicate array designs.

HarrisMarcus

7/9/2012 5:46 PM EDT

The power output of a solar panel changes a slight amount with time and this is specified in the panel manufacturer’s data sheet. This case the power optimizer can be used to very good advantage because as the Vmp and Imp points change with time, the SPV1020 dynamically tracks these changes and enables the maximum power available to be extracted from the panel.

Trinity51

7/6/2012 5:50 AM EDT

Rich - as far as I am aware the panel output is surprisingly constant with time. Most panels are guaranteed more than 90% rated output after 10 years and more than 80% output after 25 years (approx 100,000 hours of daylight use). Typical results are better than that, so degradation is just a fraction of 1% per year.

Mineyes

7/6/2012 3:39 PM EDT

"Southern Exposure is Obvious" That is true only in the Northern Hemisphere. A better statement, might be "Equatorial Exposure is required"

HarrisMarcus

7/9/2012 5:46 PM EDT

Yes you are correct. I should not be thinking only in terms of the northern hemisphere.

Robotics Developer

7/7/2012 2:29 PM EDT

A very nice article with good technical details! Very much appreciated, thank you! I am wondering if this level of complexity is needed when using solar power both locally and not at normal line voltages (ie. 12V or 24V DC lighting/systems)?

HarrisMarcus

7/9/2012 5:47 PM EDT

Using a power optimizer is not complicated and system design is straightforward. For a low voltage lighting application, the power optimizer will allow the maximum amount of power available to be extracted from the panel.

selinz

7/9/2012 4:46 PM EDT

Interestingly, a solar panel sales guy came by on the day that this article was published. I was astonished to find out that I could actually save money immediately. It turns out that we use double the "average" user and for anything beyond the norm, the 15 cents/kwh goes to about 30cents/kwh. Thank you PGE. This makes getting enough panels to knock our electricity useage in half has a quick payback period. (7-8 years).
If you are a do it yourselfer, it can be even less. see
http://www.wholesalesolar.com/enphase-solar-power-system.html
for example pricing. And no, I am in no way affiliated with these folks. They popped out of a simple google search. YMMV.

GREAT-Terry

7/11/2012 11:32 AM EDT

Very interesting MPPT control method. I like the micro-inverter idea as it is the most efficient way to dump energy to the grid. Managing multiple 200-300W power is much easier than a huge 10kW stuff. Great article!

HarrisMarcus

7/12/2012 1:35 PM EDT

Actually the microinverter method is less efficient than power optimizers and a large central inverter because 1) the voltage is higher and current lower which reduces I square R losses and 2) due to economies of scale there is less overhead and fewer number of components along with control circuits. The efficiency of the SPV1020 power optimizer is 98% and the efficiency of one large central inverter will be more than multiple microinverters.

green_is_now

1/23/2013 4:49 PM EST

until the cells get unbalanced and your 1% advatage quickly disapeers and micro inverters win on efficiency for both momentary cell inbalance (think clouds, bird droppings...) Or if a cell degrades or is mismatche power can actually drop in a cell in a series string.

green_is_now

1/23/2013 4:49 PM EST

Also there is no reason the 120 v cant be interleaved to give balanced 220Vac.

green_is_now

1/23/2013 4:51 PM EST

Large voltage ratio power coversion has lower and lower efficiencies as voltage ratios go up.
So its not just IR losses must also consider switching losses with respect to voltage ratio boost levels.

karan91

1/30/2013 5:05 AM EST

hi,any suggestion Solar Energy Conversion method for solar boat project??

LindaSF

4/1/2013 12:14 PM EDT

Your article is a little over my head but from the financial stand point there are 3 points that I always consider when debating between central inverters and micro inverters. Micro inverters are preferred when you need module level monitoring (monitoring the performance of each individual solar module), when partial array shading occurs, or when the design is constrained by locating modules sections of the roof with different orientation. If non of the above are true I usually recommend a central inverter. You can compare prices anywhere online: http://webosolar.com/store/en/80-microinverters