Simple MPPT question.

Yzordderrex, since your battery-charging voltage is relatively fixed at any point in time, maximizing current is maximizing power, assuming the use of an efficient switch-mode converter. And of course, as pointed out, also limit the maximum current.

Seeking to minimize part area, I solved the MPPT problem in a simple way, using a SOT-23 buck converter with "precision" enable voltage, and hysteresis (the TPS54202). A divider set this to 10V. An electrolytic capacitor across the nominal 12V solar panel meant operation in the 12V region, even with heavy overcast (converter turned off when the voltage sank). The buck converter was set to 4.6 volts, to feed a Lithium charger IC, MCP73832, 4.3 volts max, on an Arduino Feather. Two SOT-23 chips, no MPPT, and only about a 15-20% penalty.

sed

ther the pwm around some and find the point of maximum current into the bat tery pack?

Just keep in mind that when you consider the charging current limit to the battery, it is not the average current, it is the peak current that is to b e limited. Also, you will only get the max power from the solar cell if th at current is relatively constant, again, measuring the average is not the way to look at it.

A DC/DC switcher should be used with capacitors on the input and an inducto r. This will give you an even enough input current and output current, reg ulated by the PWM switch in the converter. Of course the switcher should b e one designed to maximize the current rather than controlling the output v oltage.

If the solar array can generate more power than you can safely drive into t he battery, the extra power will be wasted. The current from the array wil l drop, the voltage increase and the array will not operate at the MPP.

ussed

Dither the pwm around some and find the point of maximum current into the b attery pack?

e battery, it is not the average current, it is the peak current that is to be limited. Also, you will only get the max power from the solar cell if that current is relatively constant, again, measuring the average is not th e way to look at it.

tor. This will give you an even enough input current and output current, r egulated by the PWM switch in the converter. Of course the switcher should be one designed to maximize the current rather than controlling the output voltage.

the battery, the extra power will be wasted. The current from the array w ill drop, the voltage increase and the array will not operate at the MPP.

BTW, this circuit gives you a built in dither that you can use to control t he PWM. Current through the switch is PWM. The input capacitor results in a small variation in current in the solar array. Measuring the voltage an d current of the array at the switching points will give you useful data to control the PWM for MPPT.

Just a practical comment:

If you want to test MPPT without an actual solar array or a SA emulator you can use a string of power diodes

Cheers

Klaus

Why is that? If the PWM is at even 10KHz, well, what part of the battery does anything in 100us? Genuine question here...

If it's a buck converter, the ripple current shouldn't be all that high.

If the system has local storage, the battery capacity is likely to be high enough that maximum output from the solar system under optimal conditions probably can't hurt the battery as long as you manage the voltage properly.

Yes, you can (fail to properly) design a system where that's not the case and require additional safeguards.

People I've talked with often fail to appreciate that the way you route the wires to multiple paralleled batteries can affect the amount of abuse they perceive. Current paths matter.

I never realized power diodes would output power. I always thought that was just a term for diodes that would handle lots of power.

You learn something every day.

Dither the pwm around some and find the point of maximum current into the battery pack?

the battery, it is not the average current, it is the peak current that is to be limited.

Do you really think the battery doesn't pass the full current you are feedi ng it? What do you think happens? Maybe I'm making an assumption that is not based in fact.

I know in the batteries in my car, at cold temperatures is sensitive to ove r current in charging. Seems too high a current causes lithium ions to pla te on the anode rather than be absorbed into it. Do you think this doesn't apply if the current is pulsed? Where do you think the ions go when the p ower is pulsed?

The main point is that the current from the solar array should not be pulse d. That does not operate the array at the MPP, it operates it at two very distinct points that could hardly be further from the MPP.

I think he means, stack diodes in series, inject a current source at the top = solar input. The node's voltage and current will act like a solar panel. E.g., it says near say 10-14V, but as you pull out more and more current, the voltage suddenly collapses. The MPPT point is a little ways past the top, just before it totally collapses.

That's why I used the enable cutoff-voltage trick.

Perfect MPPT is a hard goal. And not especially worth working for, IMHO, because solar output varies wildly, even at mid-day. With haze, wisps of clouds, real puffy clouds, and overcast. And then you have the solar motion, the daily cycle. When you factor in the solar-radiation and incident-angle variables, it makes little sense to eek out the last 10 to 15%.

Except of course, my solar roof. There I want my micro-inverters to perform MPPT down to the last 3%. Hey, I paid the big bucks, and with electric credits and SRECs, 10% could work out to about $400 / year.

So you use a number of diodes so the drop across the diodes at full current is more than the compliance voltage of the current source and the voltage can drop to zero. Ok, I get that.

Not sure you really need a current source exactly. Won't a regular voltage supply do the job since the diodes are providing the knee? The hard part is getting diodes that will drop the right amount of voltage at the right c urrent. But then I don't suppose you need to emulate the actual solar arra y, just something like a solar array to test a MPPT device.

I don't follow that reasoning. How does it matter that there are factors y ou can't or don't account for when deciding how much efficiency to leave on the table in the MPPT? It seems to me it would be easy to get within sing le digits of the MPP.

Now you are talking like an engineer! Yeah!

No problem. Just install a solar tracker for your rooftop solar power plant. Besides about 20-30% annual increase in power output, you get a different view from the windows every hour, and perpetual daylight in all the south facing rooms.

Something like these homes:

"Rotating houses: homes that follow the sun"

"Rotating Homes"

More photos: Although I've never seen one, rotating homes seem to be a rather common design.

Something could probably be contrived from a circular railroad track foundation and some railroad flatcar wheels, or maybe a used railway turntable. The details will need to be worked out with an architect, planning department and local aesthetics committee. The power increase might not pay for the necessary home improvements, but you could always sell the plans or charge for tours to recover the expenses.

Should have been more clear, build a series stack of diodes, with bottom cathode to ground. Use about 20 diodes for 12V. Feed the top with a variable current source = solar input. Voltage sources not allowed.

With I_solar = full, Vout = max = 20*0.65 = 13 volts. The diodes follow the usual log I-V plots. E.g., for I_solar=full/20, Vout=13 - say 20*90mV = 11.2V. As you reduce solar current, to 1/200, 1/2000, etc., the voltage across the diode stack quickly plummets. These are the no-load voltages.

For a given solar current, as you draw load currents approaching the solar current, what's left to run the diode stack rapidly drops toward zero, and the diode-stack voltage rapidly plummets. That's exactly what you'll see with a solar panel in the sun. (I've made enough such plots to get solar-panel religion. If you want to see some actual curves, I'll post 'em.)

So, for modeling, or for bench testing, make a diode stack and feed it with a current source. Then try to run your MPPT from that, and see how well it works.

Thanks, Jeff, got it!

sed

ther the pwm around some and find the point of maximum current into the bat tery pack?

All you have to do is match the impedances for MPPT and ensure you don't ov erload it. Using PC current source as linear Z=V/I and the buck regulato r as a higher reactive impedance shunted by a real impedance when charge is transfered by increasing PWM. Use low ESR caps across the battery to abso rb the charge quickly, which then releases slower to the battery. CM Baluns may be necessary to reduce EMI.

Tony Stewart EE75

Ok, I'm with you now. It wastes a fair bit of power like a shunt regulator, but it works.

Would it be so hard to alter the current limit of a PSU design to provide a soft limit? That could be a lot more efficient.

AFAIR when we testet this, we had a voltage source, but a large resistor in series before the top of the string of diodes, so it behaved more or less as a current source

Cheers

Klaus

The solar-panel cutoff is abrupt, I'd think you shouldn't mess it up too much by taking shortcuts.

An MPPT system design is mostly a matter of an algorithm. But an MPPT algorithm doesn't have to be tested at full strength. A scaled-down test should be OK.

Dither the pwm around some and find the point of maximum current into the battery pack?

the battery, it is not the average current, it is the peak current that is to be limited.

Read up on Warburg Impedance

I did a water conductivity monitor that ran a 2kHz current through low cond uctivity water (tap water) and got up to about 150kHz on warm 2% NaOH.

Battery electrolytes are quite conductive. When there are lot of ions aroun d in solution, stuff can happen quite rapidly. The actual processes at the electrode happen very quickly, but the double layer next to the electrode s urface can get up to 10nm thick, which can slow things down a bit.

Right... but is any of that related to the charge current limit for the battery? How?