1
\$\begingroup\$

I'm designing a small solar charger with these characteristics:

Solar panel with 19% monocrystalline cells

  • Voc = 7,50V
  • Vmp = 6,00V (Voc -20%)
  • Isc = 1,07A
  • Wp = 6,42W

IC

Microchip MCP73871 linear 1A lipoly/li-on charger with VPPC capabilities (input voltage regulation loop)

Battery

Li-ion 3,7V 6600mAh 24,42Wh

-

This configuration works fine, but I decided to move to SunPower cells because the solar panels fabrication, when it's needed to cut the cells, in China is made better. Mainly, due to the fact that Sunpower cells have the "back-contact" technology. So, given that the Sunpower cells are more efficient (23%), the solar panel will output more total watts, that an IC such as MCP73871 will actually waste because of its 1A current limit.

SunPower solar panel

  • Voc = 7,125V
  • Vp = 6,10V
  • Vmp = 5,7V
  • Isc = 1,35A ◁◁◁
  • Wp = ~8,25W

So, I moved my attention to another IC, this time LT3652 (rated 2A). At a first glance it looked fine but then I discovered that it didn't work with a 6V solar panel. It has a VIN_START value of at least 7,50V to start the IC, after that the actual normal nominal operation voltage can be 5,95V.

A solution to this problem could be to cut the SunPower cells smaller to get an higher voltage in the same surface, given that this IC is actually a buck-converter. But someone told me I could really lose efficiency because of the big voltage difference with the input and the battery.

Or perhaps the solution could be to use a different approach with another IC. For example, the ST SPV1040 is a very cool chip with a MPPT perturb and observe embedded algorithm and it looks simple to design. It's designed as a boost converter, so almost any low voltage solar input source is desirable. My idea could be to use two uncut Sunpower cells to reach 1,22V. To build a panel with 2 full cells of 125mm x 125mm should be much easier to design and to fabricate, because it has no waste and there's no need to cut the cells. I will boost this voltage to charge the li-ion battery pack. However I can't understand yet how much current this IC can handle and if it's capable alone of a li-ion charging profile.

Within SPV1040's datasheet I found ILx, a value for the current limit.

enter image description here

Is ILx referred to the also called "switch current" of the inductor?

This project has a limited space available for the solar panel surface, so the whole design could be more complicated. I'm talking about 257mm x 175mm.

The very ultimate goal for this charger is to harvest the most possible amount of watts/time, but any solar-specialized technology such as an active MPPT should be taken into account without to fall in an excessive electric circuit complexity, because I'm not an engineer.

Any effort with this question will be very appreciated. Thank you.

\$\endgroup\$
  • \$\begingroup\$ No, it's the current into the LX pin. \$\endgroup\$ – Ignacio Vazquez-Abrams Jul 15 '16 at 15:30
  • 1
    \$\begingroup\$ @pjc50 Obviously I intended that if I slice the cells thinner I could connect more of them in series to get an higher voltage in the same surface. \$\endgroup\$ – Nic1337 Jul 15 '16 at 15:36
  • 1
    \$\begingroup\$ @Nic1337 Well asked question largely. Meaning of comments re China may not be clear to some. It would be interesting to know your application if you are able to say. (I have substantial experience in in-China manufacture of small solar lighting products). | It is easy to have the "little things" eat up the efficiency gains you aim to get from eg Sunpower cells. Looking both widely and closely at all aspects is necessary to get best overall result. \$\endgroup\$ – Russell McMahon Jul 15 '16 at 17:07
  • 1
    \$\begingroup\$ @Nic1337 "Full time maker" sounds 'interesting :-) - and will sound different to various people. Some people who have been involved with electronics for a long while (and some others as well) hate the term "maker" with a vengeance. They are about equally enthusiastic about terms such as sketch, shield and similar new and non-essential terms which seek to replace existing ones that have a long history. New is not bad per se, but the objection is often that the new way [tm] is 'sold' as fun/easy/trivial/thought free/non-rigorous/unthinking... . That is of course an unfair over-generalisation ... \$\endgroup\$ – Russell McMahon Jul 18 '16 at 13:58
  • 1
    \$\begingroup\$ ... but there is also some truth and merit in the perspective. I am an old fogey, but I try to see the new approaches and new terms as things that may help some younguns achieve more things than otherwise. I actually dislike the term "maker" but it makes no difference to how I deal with people and not much difference to how I view their input. It makes SOME difference as when a person uses the term 'maker' it often gives an indication of their relative inexperience and knowledge. This of course is not always the case - even with very young people - and I know people almost as ancient as me ... \$\endgroup\$ – Russell McMahon Jul 18 '16 at 14:02
1
\$\begingroup\$

You may be best off using 'conventional' monocrystalline silicon PV cells, arranging Vout to best suit your needand maximising packing density and area occupied on your device. Many manufacturers waste substantial area in intercell spaces (> 10% common, 20% not uncommon) and this can usually be reduced to maybe 5% of area with care.

Optimisation of front sheet optical loss can help make up losses elsewhere. Worst case should be 10% and around 1%-2% is possible if cost is not an issue.

The MCP73871, data sheet here acts as a linear regulator for charging purposes. Max Iin = 1.8A so at 4.2 V out max power = V x I = 4.2 x 1.8 = 7.6 Watts. However, unless you use a preceding MPPT converter the proposed 8V Vmp PV panel will have a maximum efficiency of 4.2/65 = 70%. LiIon cell mean voltage across charging range is closer to 3V7 so mean efficiency ~~ 3.7/6 = 62%.


The LT3652 data sheet here is a nice device if the 7,5V Vstart can be accommodated. Note that the application note tends to focus on battery chemistries which can be floated when 100% SOC (stae of charge) is reached. Lithium Ion cells MUST NEVER be "floated at end of charge. Charge voltage MUST be removed at the end of charging, that their days may be long on the face of the land. The LT3652 can accommodate this need with suitable design.


A simple buck converter / charger such as the LTC4002 may better meet your needs. This allows 5V to 22V input and uses and external MOSFET switch & external flywheel diode and external current sense resistor. fficincy of about 85% can be obtained at Vin = 6V and adding a synchronous rectifier FET in place of the flywheel diode may increase efficiency slightly.

LTC4001 is a 2 A buck regulator LiIon charger BUT Vin max of 5.5V is an annoying limitation.


Your LiIon cell will need at least 4.2V to charge fully - say 4.5V minimum available.

Sunpower cells are difficult to join when cut into fractions of a cell. Most PV panel manufacturers appear unable to do this. Using whole cells should present no difficulty.

Boosting from low voltages is usually less efficient end to end than either boosting from a voltage closer to FVout or buck converting from above Vout.


The SPV1040 is capable of 3 Watt output maximum (data sheet table page 6).
The inductor maximum maximum current (really the switch max current) is shown as 1.8A as you say BUT you must use the worst case = minimum-maximum current for design purposes.

The graphs in the SPV1040 data sheet on pages 8 & 9 show efficiency at various Vin/number of cells combinations.
At full power with 3 cells they show 80% at 4.5V out at 2 Watts without MPPT and 89% with MPPT.
Note that with 3 cells they show 2W max.
Sanity check:
3 cells ~= 1.5V full sun.
Duty cycle at 1.5V in and 4.5V out at 100% efficiency
= 4.5/(1.5+4.5) = 75%.
Max power at 100% = 1.5V x 1.8A x 75% = 2.0 Watts IN.
= what they said. Pout =~ 2W x 90% = 1.8W out MAX = far short of your desired power level.

For the SPV1040 ILX appears to be the inductor current, & the switch current & the LX pin current.

\$\endgroup\$
  • \$\begingroup\$ Dear Russel, just a correction: MCP73871 has a 1.8A limit for IN and OUT pins, but its maximum charge current for VBAT remains at 1.0A. That 0.8A surplus can only be used to power the system circuit. In summary: IN(max) = 1.8A VBAT(max) = 1.0A OUT(max) = 1.8A, when OUT asks for 1.8A, VBAT = 0A \$\endgroup\$ – Nic1337 Jul 18 '16 at 9:36
  • 1
    \$\begingroup\$ @Nic1337 Yes - you are correct re max charge current from MCP73871. I had missed that point. I did not consider it suitable overall given your spec as it's action as a linear regulator makes it extremely inefficient in your application. I included it for completeness. In other circumstances if it was otherwise suitable you could increase Ichgmax with a very little external circuitry* but in this case that would not help its basic inefficiency. (* eg if charging and Vbat < CC/CV changeover voltage of 4.2V turn on a transistor from 'OUT' to 'VBAT'. (eg TL431, jellbean transistor, few resistors. \$\endgroup\$ – Russell McMahon Jul 18 '16 at 13:00
  • 1
    \$\begingroup\$ @Nic1337 The above charge capacity extension method is just a 'thought experiment' in what is possible. There are enough LiIon charager ICs available that the far better alternative is to select an IC that directly meets the need. \$\endgroup\$ – Russell McMahon Jul 18 '16 at 13:02
  • \$\begingroup\$ Yes, this week I'm gonna have a consultancy with an electronics company :) \$\endgroup\$ – Nic1337 Jul 18 '16 at 13:59
  • \$\begingroup\$ @Nic1337 You have about as good an "electronics company" available here as you'll find anywhere on earth. Really. There are probably tens of people here who are as close to the leading edge of most aspects of all the things you need to do and know. Your task, here or in your other meetings, is to fully and clearly convey your actual requirements. \$\endgroup\$ – Russell McMahon Jul 18 '16 at 14:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.