Monitor PV output in realtime under load

Thread Starter

killit2

Joined May 2, 2012
6
Hello all,

I'm working to TRY to make my own mppt style controller using an atmel MCU and an existing boost buck controller. Everything is falling into plan logically except how to monitor the availiable power from the panels. Heres an example.

Assuming the PV is putting out 10Amps at 15V there should be 150W avail of incomming power. My calculations put the buck controller at ~90% efficient on the conservative side so I should be able to kick that down to say 13.8 or so (depending on the stage of charge of couse) with 9.7Amps.

Thats all fine and dandy but how do I measure the total current avail from the panels. As these are essentially a current source across a PN junction they SHOULD give a constant current and change in voltage but the IV curve (realistic - not the ideal) due to mismatches from shading or whatever shows that at some points this is not the case - the current changes.

So other than shorting the panels across a huge load resistor for a ms or so ever few seconds and measuring the voltage, how do I actually find (in real time mind you) the total power comming from the PV? Mind you that since the boost controller is being controlled by the MCU it is essentially (spelling?) a DC/DC converter for current so if the MCU can't determine total power in, it can't do relaible calculations to setup optimal load on the PV. Its for this reason that I don't think just a simple low side sense resistor will work. If the MCU is only thinking that half the available power is there, its only going to command convertion of half the avail power.

Thanks,

Shawn
 

wayneh

Joined Sep 9, 2010
17,496
I believe the controller needs to perform tiny experiments, moving one way and the other, always climbing uphill in efficiency. It needs to learn the local slope. That said, if you have the I-V curve versus temperature and light intensity, and have measures of those two things, you're awfully close to having all you need.
 

Thread Starter

killit2

Joined May 2, 2012
6
I had thought of this method but discounted it after doing some simulations where the controller is commanded above the avail power - it goes goofy to say the least. Being that the mcu gets feedback from the output, when it goes goofy it throws a curve ball for me. I'm sure more code will over come that, but I was hopping there was something I was overlooking.
 

THE_RB

Joined Feb 11, 2008
5,438
...
Assuming the PV is putting out 10Amps at 15V there should be 150W avail of incomming power. My calculations put the buck controller at ~90% efficient on the conservative side so I should be able to kick that down to say 13.8 or so (depending on the stage of charge of couse) with 9.7Amps.
...
Well you just proved your MPPT does not work. Best case (as you said here) there is 150W input power, and 133W output. That is very poor.

Even a linear reg (which has Iout = Iin) will get 150W in, and 13.8v 10A out, or 138W out, and just connecting the solar panel to the battery will increase current a little (as panel voltage is reduced the current rises a little) and give you maybe 13.8v 10.5A or 145W.

So connecting direct to a battery will be the best power out at 145W, second would be a linear regulator at 138W out, and worse of all is your "MPPT" at 133W out. :)
 

Thread Starter

killit2

Joined May 2, 2012
6
You have a very vaild point - and I've actually considered that. Where the MPPT will actually help is that I can generate SOME power all the time as opposed to full power only for a short period. So yes, when its high noon and everything is wonderful, I would actually be at a disadvantage. But, later in the day, say when the voltage from the cell drops to 12.5 (assuming even an LDO linear regulator) I loose the ability to charge (Assuming 12.5 is the state of the batteries at that point). Thats potentially 125W that I am just throwing away. Using the MPPT approach (or at least what I am terming MPPT) I can continue to put current to the batteries. So yes, at pmax, I drop from 150w to 133w which is only 88% efficient - but later in the day Im still putting roughly 110w into the system. So 0/110 is infinite increase in efficiency in that case :).

Additionally, this system will be used in the far north in mid winter in a fixed position- so its doubtful that these cells will ever produce anything close to their pmax - in that case the MPPT approach really pays off.
 

THE_RB

Joined Feb 11, 2008
5,438
You have a very vaild point - and I've actually considered that. Where the MPPT will actually help is that I can generate SOME power all the time as opposed to full power only for a short period. So yes, when its high noon and everything is wonderful, I would actually be at a disadvantage. But, later in the day, say when the voltage from the cell drops to 12.5 ...
...
Sorry but I think you are working on an incorrect assumption here. :)

You can connect the solar panel direct to the battery, and it will still charge even in reduced light conditions.

A solar panel in reduced light will still produce the same 21v etc open circuit voltage, only its current is reduced not its output voltage. The solar panel can be modelled as basically a current source where the current is based on insolation (light level). The panel voltage is based on a "zener" curve which will top out at about 21v if no current is drawn from the panel.

So it's an incorrect assumption that you need a SMPS or MPPT etc to get power out of the panel in low light conditions.

The complete benefits from MPPT is that the panel can be run from its peak power point (voltage of about 17v) then the 17v be converted to 13v etc and hopefully increase the current into the battery. A 12v MPPT needs to run VERY high efficiencies to even be worth the bother, ie 95% eff or more and even then it may only make a few percent more power into 13.8v battery than directly connecting the panel to battery.
 

Thread Starter

killit2

Joined May 2, 2012
6
Ok, I see what you are saying. I had been misinterpreting the IV curves as the voltage plane being the base of variation with the current being fixed. After reading your post I did more research and found an animated one showing the curve move according to light and bang! - it all made sense. So you solved my issue although in an unexpected way. My whole idea was to charge the batteries as efficiently as possible without over charging and damaging the cells - it seem a much less complicated PWM approach would do that - And for as cheap as they are, Ill just buy one. Learning every day :)
 

THE_RB

Joined Feb 11, 2008
5,438
Sorry to have talked you out of making a MPPT, but that's probably the better option.

You could still make your own microcontroller circuit, but instead just put a load dump regulator in there to stop the battery from being overcharged. Your micro can drive the load dump using a FET and a 12v light bulb to draw the excess panel current.

You'll still get the benefits of the micro, ie reading battery voltages etc and maybe displaying charge history or whatever. :)
 
Top