diy solar

diy solar

Low-power bad weather solar harvesting, an attempt to build a better SCC...

rin67630

Solar Enthusiast
Joined
Apr 29, 2020
Messages
1,066
Location
Nort-Rhine-Westphlia Germany
Edit: i have given up with this experiment: it works well, but the tiny buck converters are too sensitive to work outside.
I could achieve performance, but not reliability.


Let us face reality: in the power range 10W to 100W there is no real MPPT SCC, to do the job well.
An expensive Victron might be technically at the top, it needs however 30mA for its own operation and that continuous drag is jeopardizing every benefit of the MPPT operation.
You have a few hundreds of cheap SCCs makes on the market that just pretend to be MPPT.
Buy them, open them and you will just find plain PWM inside, you cannot even find a buck coil for the power conversion in them. :(

Beside that you have a few better buck conversion chargers, based on CN3767 or BQ24650 chips, that also advertise -even in the manufacturer's data sheet- with MPPT, but do only provide a constant voltage regulation on the primary side. It is better than nothing, but is literally
"Manual Power Point Trimming":rolleyes:

I am using such modules and have provided a networked monitoring for them, and am currently consuming only 15mA including my ESP8266 that does other tasks beside...

I think I could get those modules to work better and perform a real MPPT by software upon injecting a controlled PWM signal mixed into their "MPPT" potentiometer.

But before I begin this development I would like to ask if you know better existing modules in the power range 10W-100W.

My current choice is to build up on that module:
BQ24650 10A Solar Panel Controller
that seems to have a decent 600Khz synchronous buck conversion, flowback protection, and can be adjusted to different chemistries.

I am open to every suggestion, if you know something better.
Regards
 
Last edited:
You say "bad weather solar harvesting"
Are you trying to get the most out of 10W production from a 100W panel?
Or are you thinking of 500W, 1000W or more array on a very cloudy day, producing 100W and less?
 
I know that more panels are the easiest solution, but I am after an affordable off-grid instrumentation solution to be placed on the public domain or on agrarian land and more solar panels mean more mechanics, vandalism etc... so I try to get the best out of a single panel.
The solution must be roughed, frost-tolerant, maintenance free, with a lot of telemetry built-in.
 
Reason for my question was thinking maybe the MPPT which produces power for multi-hundred watt days was the best way to go for overcast days. Could be the PV voltage has dropped to where Vmp can't be maintained, so a longer series string would work much better?

Panels are getting down to $0.15 to $0.50 per watt for large sizes. If small panels had similar cost (and they probably do as components integrated in a product), then 10W to 100W spans $1.50 to $50, which doesn't leave much budget for charge controller.

Newer panels are 18% to 20% efficient, vs. 12% to 14% for crystaline panels of some years back. Can put out 50% more power in same size.

I think skipping MPPT and just a buck converter delivering voltage to battery would be the way to go. With lithium, don't need temperature correction like lead-acid. Feed a temperature sensor into the shutdown pin. Isc of panels varies a lot with illumination. Voc/Vmp doesn't vary so much.

The bigger bang for the buck it to operate load according to available power. Communicate data less often when less power available.
 
Reason for my question was thinking maybe the MPPT which produces power for multi-hundred watt days was the best way to go for overcast days. Could be the PV voltage has dropped to where Vmp can't be maintained, so a longer series string would work much better?

Panels are getting down to $0.15 to $0.50 per watt for large sizes. If small panels had similar cost (and they probably do as components integrated in a product), then 10W to 100W spans $1.50 to $50, which doesn't leave much budget for charge controller.

Newer panels are 18% to 20% efficient, vs. 12% to 14% for crystaline panels of some years back. Can put out 50% more power in same size.

I think skipping MPPT and just a buck converter delivering voltage to battery would be the way to go. With lithium, don't need temperature correction like lead-acid. Feed a temperature sensor into the shutdown pin. Isc of panels varies a lot with illumination. Voc/Vmp doesn't vary so much.

The bigger bang for the buck it to operate load according to available power. Communicate data less often when less power available.
...and more (bigger) solar panels mean more mechanics, vandalism etc...
The panel cost is not the main factor: The street lamp-pole and the basement cost to put everything out of reach are more costly than the panels.
Then the conversion from MPPT (Manual Power Point Trimming) to MPPT (Maximum Power Point Tracking) will practically be at no cost: it will just be software, and I like the challenge...
 
I am currently working on my ESP8266 controller program to retrofit real MPPT functionality to many <10A cheap solar controller modules advertising MPPT but only providing manual power point trimming, or even only a fixed point at 17V.

These modules are still interesting for low power solutions, because they require much less power than full fledged MPPT controllers.

My current ESP8266 DIY based system provides a full graphical battery monitoring, tracking, statistics. I just need to extend the software with a tracking point injection and a few convenience nice-to-have's.
While I am on the way and have a few digital outputs left over, I could provide control for two additional buck converters with enable function,
so I could also remote/programm-switch these power outputs to various voltages and reduce power when the connected devices are not in use.

The current bill of material will be:
- your small pseudo-MPPT solar controler (cannot be a PWM-module) e.g.
MPPT solar controller battery charging board or BQ24650 10A Solar Panel Controller
- a Wemos ESP8266 WiFi enabled microcontroller
- an INA226 I2C bidirectional current/power monitoring sensor
- 1 to 3 or more buck converters with enable input
- alternatively, if you prefer to switch AC a 2 Relays board
- 6x4cm or bigger prototyping board
- a few resistors, wires, screw-terminals and a small industrial junction box.

n.b: I am not affiliated with Aliexpress and do not even recommend them, but their shop provides almost every device's pictures and details to describe what's requested.

Stay tuned, I'm on my way...
 
Last edited:
Here is the schematic:
1603998406679.png
The mother module with wired power lines:
1603977275758.png
The mother module without ESP / INA 226
1603977351173.png
The complete device without enclosure/connections:
1603977455306.png
 

Attachments

  • 1603977114823.png
    1603977114823.png
    126.2 KB · Views: 17
Last edited:
Finally, while tinkering, the expectation grew.
Here is the schematic of what I realized:
1604270133069.png

So currently the project will provide:
- injection of MPPT correction information to modules falsely claiming MPPT and based on following chips:
* -CN3722 (Li-Ion and LiFePo)
* -CN3767 (Lead Acid with absorption)
* -BQ24650 (Synchronous mode 600Khz, wo absorption)
* -XL4015 blue module with 4 trimmers, wo absoption)

n.b. the numerous modules/ cheap blackboxes on the market, just providing PWM and falsely claiming MPPT cannot be improved!
(the ESP8266 and the INA226 can however provide to them plain monitoring functionality and the software will consider that case as well)

- injection of voltage correction information (to force lead-acid absorption stage for modules that do not consider it)
- battery voltage and current monitoring (with trending in the cloud)
- panel voltage monitoring
- statistics and Coulomb integration
- logging of events (low/high voltage, cycles, begin / end of full charge)
- energy not harvested (after battery full)
- computation of internal resistance
- 5V 1A output ( permanent )
- two independently adjustable voltage outputs 1A (2-12V) switchable by software
- two relays switchable by software (unloading strategies, alarming...)

I began soldering on a 6cm * 4cm prototyping board, that was definitely too tiny! the cabling at the backside became quite a mess.
I will probably redo it later on a bigger board. For the time being, it will do the job.
I also had buck converters, without an enable pad, at hand. Hence I had to wire directly the enabling signal to SMD resistors on the board.
The next version will use better converters.

Tomorrow I will post photographs and begin the debugging and the trimming of the software.
 
Last edited:
Finally, instead of trying to fix the crappy behaviour of analog circuits with pseudo MPPT, I found it to be more effective to start from a cheaper plain synchronous buck converter without power point control and do the complete control (MPPT and Battery charge control) by software directly into the CV / CC potentiometers.
That way I will have only one type of hardware and provide the different charging profiles by software as required.
I have re-ordered a LM25116 based 300W Buck converter (I have fried mine with an incidental solder bridge :-(
One should ALWAYS keep the workbench clean!

The schematic will remain practically identical, excepted the PWM controls are now CC and CV.
 
The basic software and the dashboard are ready:
1604497551513.png
The voltage displays and the current / power displays, the controls for DC_outs and relays are operational.

Now i have to find out the best resistor settings to inject the current/voltage changes, it seems that the high 220K values must be reduced for the buck converter that i have for the tests...
 
Last edited:
Now the final schematic:
1604571599733.png

and the corresponding dashboard:
1604568866071.png

I am still experimenting with a spare 10W panel that does not get much sun.
But everything will be full scalable!
I will consider several common buck converters on the market, but the device works already as a passive reporting device for many SCCs on the market.
The only limitation is for PWM devices that mostly switches on the negative poles: I will not be able to report their panel voltage.

Today, I will put the current software on GitHub. It is fully parametrizable and will get many options.

All passive reporting functions are working excepted the optional OLED display (but i have code for it from another project), now I need to work on the active part:
- MPPT injection and charging profiles for various chemistries
- Parametrizable controls for DC Out1&2 and Relay 1&2
- On the long term, I plan to interpret the weather forecasts to calculate the future radiance expectation and try to anticipate the risk for empty battery...
You are welcome to participate!
 
Last edited:
Nothing wrong with manual MPPT control. I have used modified buck converters to hold the panel at fixed power point and a $5 PWM charge controller and 60V array. It acts as a linear current booster. Even thermal tracking wouldn't be that hard to add.
 
Nothing wrong with manual MPPT control. I have used modified buck converters to hold the panel at fixed power point and a $5 PWM charge controller and 60V array. It acts as a linear current booster. Even thermal tracking wouldn't be that hard to add.
I was not happy with the behaviour of the fixed power point modules.
In the morning they begin one hour too late and stop an hour too early in the evening.
OK that are not the best hours, but I don't want to lose that little power neither.
Under casted clouds they are frankly suboptimal, and that is an important point for me, with my "english" weather in North Rhine Westphalia.
(by the way, BIG, BIG WIN for TRUMP in North Rhine Westphalia as well, good that they stopped to counts the votes early to prevent Democrats fraud) :p

Additionally I will provide full charging profiles for several chemistries, what many modules don't provide.
And, last but not least, I will optimize the quiescent current to a minimum.

Anyhow the powerful reporting abilities and the dashboards will be available for your solution as well.
 
The hardware is up and running, I still need to solder definitely the two control resistors which value i needed to find out.
Photo will follow once it's done.
The good thing is that my low-power 15W buck converters have practically no own consumption.
The software is operative with a manual CV slider control of the buck converter that charges the battery.
It is pretty touchy and the power really varies heavily on the panel voltage.
I will surely need to run with two INA226 to segregate the charge/load currents from the battery, else potential load current variations may perturb the MPPT loop.
I have rerouted the 2nd PWM output to control the voltage of one of the convenience DC outputs, so I can set-up the voltage between 5V and 12V remotely/programmatically.

1604655987437.png
I have coloured the dashboard to reflect the dependencies between the different elements.
Stay tuned...
 
Here is the built SCC
1604682369129.png
it fits easily in a 10cm x 10cm box with all connections.
Here the detailed view:
1604682476104.png
It features:
-Software MPPT 12/24V 2A controller
-2 convenience adjustable power supplies switchable by software, one of them with software controlled voltage.
-Option for 2 relays module up to 250V AC 10A.
-5V output on USB connector.
BOM cost: <15$.
Power consumption 1mA + 7-20mA for the ESP8266 WiFi module.
 
Edit: i have given up with this experiment: it works well, but the tiny buck converters are too sensitive to work outside.
I could achieve performance,
but not reliability.


Rebuilt the whole system with 3 d-sun buck converters, and added an OLED display.
I can feed 8W into the battery without problem: the miniature buck converter (without any heat sink) does not get warm (36°C measured)
I made a pic with my hand to give the scale:
1605903177421.png
The 3 buck converters are doing the following:
1) panel to battery charger with software MPPT (still under scrutiny, needs some adjustments.)
2) battery to 5V for the ESP8266 (hidden under the ESP8266 module, itself under the display)
3) Battery to 6-12V convenience output, software controlled in voltage and on-off.
The software can control a 4th buck converter on-off + 2 relays (or 3 relays).
The web display:
1605903765939.png
  • The blue widgets show the panel voltage and the control slider,
  • The green widget shows the battery voltage
  • The purple widget shows the battery current (to=positive -from=negative)
  • The pink widget shows the battery power
  • The light grey widgets show the weather (from openweather.org)
  • The yellow widgets control the convenience DC output (ON-OFF and voltage slider)
  • The orange widget could control a 2nd convenience DC output (n/a)
  • The darker grey widgets could control the relays (n/a)
  • The white widgets at the bottom show the day balance in Ah (for today and yesterday) the battery current and the battery internal resistance.
  • The plot shows the panel voltage, battery voltage and battery power.
Enjoy!
 
Last edited:
Back
Top