diy solar

diy solar

Simple home made analog MPPT contoller

I'll keep you posted once things are done here. Cheers
 
Vmp moves considerably with temperature and irradiance. This is therefore not a mppt Controller it’s merely a PWM solar controller with voltage regulation. The panel being directly connected to the battery when the mosfets close. This also results in very poor power transfer for higher voltage panels.

Mppt has two features one is power conversion , the other is dynamic tracking of the Vmp This circuit has neither and show not be described at mppt. ( it’s like how some Chinese pwm tries to suggest it’s mppt when it’s not )

These days with the trend in higher voltage panels or series panels there is no use for PWM as you loose considerable panel power.
 
Last edited:
Vmp moves considerably with temperature and irradiance.

It certainly does, but that has a lot less of an effect than most people have been led to believe.
I have been thinking about all this for years, and I just had to try the idea.

Once you have a means of easily adjusting the actual operating point up and down manually with a potentiometer, you might be amazed at how little the bulk charging current varies over a very wide operating voltage range at the panels.

Sure, temperature and insolation effect the actual peak power voltage, but if you can crank that voltage through a wide range of adjustment and only see a very few percent change in bulk charging current, what is that trying to tell you ?

What really DOES matter, is controlling the loading on the panels to changing insolation, and this circuit does that brilliantly.

To put all this another way, the solar panels are a (relatively) high impedance current source.
Even a slight change in loading has a very large effect on the voltage at the panels.
If you can track that, you have the game beat.

Its basically all about impedance matching the load impedance to the source impedance.
Its of far less importance to get the solar voltage operating point exactly right, than to track a very wide loading adjustment range, to keep it "somewhere near" the peak power voltage. And then use efficient power conversion to do it.

Just wait and see the results other people are getting with this before passing final judgement, its very early days yet.

I was truly blown away by what I was actually seeing, it was totally unexpected.

The panel being directly connected to the battery when the mosfets close. This also results in very poor power transfer for higher voltage panels.

NEVER at any time is the battery directly connected to the solar panels.

You are quite wrong there, a buck regulator does nothing of the kind. The choke effectively isolates the input switched high voltage waveform from the lower voltage output dc waveform. The switched input voltage is averaged, and at very high efficiency.

Oh, by the way, I am a retired power electronics design engineer, I used to design the switching power supplies in mass produced commercial equipment among many other things.
 
Last edited:
It certainly does, but that has a lot less of an effect than most people have been led to believe.
I have been thinking about all this for years, and I just had to try the idea.

Once you have a means of easily adjusting the actual operating point up and down manually with a potentiometer, you might be amazed at how little the bulk charging current varies over a very wide operating voltage range at the panels.

Sure, temperature and insolation effect the actual peak power voltage, but if you can crank that voltage through a wide range of adjustment and only see a very few percent change in bulk charging current, what is that trying to tell you ?

What really DOES matter, is controlling the loading on the panels to changing insolation, and this circuit does that brilliantly.

To put all this another way, the solar panels are a (relatively) high impedance current source.
Even a slight change in loading has a very large effect on the voltage at the panels.
If you can track that, you have the game beat.

Its basically all about impedance matching the load impedance to the source impedance.
Its of far less importance to get the solar voltage operating point exactly right, than to track a very wide loading adjustment range, to keep it "somewhere near" the peak power voltage. And then use efficient power conversion to do it.

Just wait and see the results other people are getting with this before passing final judgement, its very early days yet.

I was truly blown away by what I was actually seeing, it was totally unexpected.



NEVER at any time is the battery directly connected to the solar panels.

You are quite wrong there, a buck regulator does nothing of the kind. The choke effectively isolates the input switched high voltage waveform from the lower voltage output dc waveform. The switched input voltage is averaged, and at very high efficiency.

Oh, by the way, I am a retired power electronics design engineer, I used to design the switching power supplies in mass produced commercial equipment among many other things.
Sorry I missed the fact the choke and it’s diode were external

It’s not mppt and shouldn’t be so described
 
It does track output power though, and amazingly well.
But I agree with you, MPPT should be reserved for a true perturb and observe software driven controller.
So what shall we call this ?
Constant voltage tracking maybe, a CVT controller ?
The Warp Solar Controller, or something else ?
 
Per Wikipedia it is an MPP setup. You just don't have tracking. For fixed panel and battery setup does it really matter? Whenever I look at my SCC's the panel voltage is only varying a few volts.
 
Per Wikipedia it is an MPP setup. You just don't have tracking. For fixed panel and battery setup does it really matter? Whenever I look at my SCC's the panel voltage is only varying a few volts.
None is suggesting Vmp is the greatest thing on the planet. Mppt delivers small incremental gains. Far more gains over conventional PWM are as as pointed out by “ warspeed “ to be had by power conversion technology ie boost conversion and manual operating point selection.

The next gain of tracking Vmp is incremental not Revolutionary
 
No disagreement from me. Was just pointing out there us a term for fixed voltage and my own observations. Most aspects of the system are a compromise based on the individuals situation. It's good to see other ways of looking at things, after all it is a DIY forum. :)
 
I do thermal tracking on my control which gives acceptable results without a lot of inconvenience. A 4x4 piece of sheet metal painted black which experiences the same sun conditions is the closest. This eliminates long wire runs. Just ambient air temperature gives close enough results with seasonal adjustment of the central power point. This is considered a MPPT method. It can be confusing when several chip manufacturers call their chips MPPT when they are actually fixed voltage with no IV calculations. An external thermistor adjusts the chips panel voltage divider.
 
The pyranometer principle is pretty much the industry standard for measuring solar radiation, and its really just a fancy piece of black sheet metal.
https://en.wikipedia.org/wiki/Pyranometer
Some cheaper pyranometers just use an ordinary (single) photovoltaic cell, and measure short circuit current through a low ohm resistive load.

I think I have finally figured out a way to do some back to back testing between my own analog solar controller, and a Make Sky Blue perturb and observe solar controller.
I now just need to pull my finger out and set it all up.

There are series strings of four 24v panels here, and I have two identical strings of those facing north, running in parallel with zero shading.
Also have a pair of Turnigy dc power meters.
I can hook up one string with its own power meter through my own controller, and the other string through the other power meter through the Make Sky Blue controller. Both feeding into the same battery at the same voltage.

The Turnigy power meters only measure up to 60v max voltage unfortunately, and my system runs at well over 100v.
But I can still measure accumulated amp hours on each with excellent resolution.

It will need some very careful adjustment of charging voltage on each controller to get them to be exactly the same, for it to be a fair test, but current during the very final stages of absorb should give a pretty fair indication of any voltage unbalance.
Most of the amp hours accumulated should be during serious bulk charging anyway, but I will watch how this all goes.

I could run it for say a week, and compare the total accumulated amp hour contribution of each controller over the full week.
The next week I swap over just the controllers, leaving the same panels connected to the same power meters.
Any difference in controller efficiency should begin to show up over time, regardless of variable weather conditions.
Any suggestions on my methodology gratefully received.

I have so many other projects on the go right now, but will eventually do this.
 
Hooked up the test as described above. Four north facing panels into my constant voltage controller, and four identical north facing panels into a Make Sky Blue mppt controller. A separate Turnigy power meter connected to the output of each controller, set to measure amps and accumulated amp hours going into the battery.

First I compared the solar panel open circuit voltages, they were within a volt. Short circuit currents were different by just under five percent.
The wiring lengths are a bit different too, not sure what effect that might have, but we shall see...

Late yesterday with a fully charged battery, I struck the anticipated problem of not being able to get both controllers set to identical output voltages.
With even just a very few millivolts difference (at 100v) one controller would seize control and trickle charge at say half an amp or less, where the other controller would register zero current. Even the tiniest most sensitive voltage adjustment caused the controllers to toggle, one charging and the other off.

The rating plate on the solar panels say 30.7 volts max power voltage, so I set my constant voltage controller charging voltage to four times that.

So today at sunrise the fun started.
Initial solar output was very low with the sun rising almost due east, completely edge on to my north facing panels used for this testing.

The east facing half of my passive tracker was on the other hand going flat out right from sunrise, and had brought the battery right up to full charging voltage well before 9am, before the north facing panels could even really get started.

Constant voltage charger. Make Sky Blue mppt.
1.04 amps 1.03 amps. 7:30 am slight haze. Did not think to measure battery voltage.
solar 122.8v solar 119.2v
1.705 Ah accumulated. 1.598 Ah

2.48 amps 2.48 amps 8:30 am slight haze. Battery 3.397 v/cell, bulk charging.
solar 122.7v solar 115.6v
3.080 Ah 3.052 Ah

3.07 amps 0.73 amps 8:45 am slight haze. Battery 3.45v/cell reached full charging voltage.
solar 122.8v solar 137.7v (absorb mode)
4.109 Ah 3.896 Ah

3.55 amps 0.26 amps 8:50 am.
At this point my east facing panels were putting out about 9.5 amps before the north panels could really even get started.
Should have realized this would happen.

I will make a few changes to all this, and try again tomorrow using just the north facing panels.
 
Last edited:
Can you put a big enough load so the battery doesn't get fully charged so you don't have the chargers stepping on each other problem?
 
What I have just done, is move the battery very close to the two controllers (within a couple of feet), and fit series resistors into the output of each controller. These resistors are each 0.1 ohm 50 watts. Also I switched off all my other solar controllers leaving only the two controllers under test to charge the battery.

The resistors have helped a lot with load sharing at low charging current, and as you suggest, switching in some extra battery loading, causes them to track much better than before at low current. They do still fight each other a bit, but nowhere near as much as before.
I don't wish to go too extreme with this load sharing strategy, as that really defeats the purpose of the test to some extent.

If I can get this setup working well enough to gain some confidence in it, I will run it for several days, then swap over the two controllers, so each runs off the "other" solar panels and the "other" amp hour meter.

I think the results are either going to be too close to call, or one controller will show some consistent advantage.
Either way I don't think we are going to be able to get figures right down to the last 1% accuracy or repeatability.
But over time, and hopefully some variable weather, it might be possible to draw some conclusions.
 
Last edited:
Where final charge points are being reached you really don’t care which controllers is supplying the juice. There’s no point having load sharing control where you’re at 10-15% of controller charge ability
 
You are quite right of course, but what I am trying to do is compare the relative efficiency of two very different solar controllers.

If both keep in exact lock step during bulk charging, then if one quits charging early because its set output voltage is set a fraction low, and the other keeps going supplying a slowly diminishing absorb charge current, very slowly ramping right down to zero.

At the end of the day, charger two shows more ampere hours accumulated , simply because it was working over a much longer time period.
Charger one may actually have a higher functional efficiency if it was running by itself, but the accumulated amp hours of a second charger may mask that....

Its really difficult to seperate two different charge controllers that operate with different modes of operation, or algorithms when charging the same battery. But there is really no other practical way to make a comparison that I can see.
 
Last edited:
Its been truly horrible weather here, cloudy, rain showers, thunderstorms, very little sun.
That is actually good, it means both controllers are attempting to bulk charge all of the time, without the absorb unbalance problem.
I have three quartes of my solar disconnected, only running two small 1kW banks of four panels for the test. My battery is at minimum voltage most of the time now, drawing just enough grid power to maintain minimum safe voltage.

The Make Sky Blue mppt is always slightly ahead, but never by very much, regardless of conditions. Its probably about four to five percent better.
Then disaster ! Both power meters stopped accumulating at exactly 99.999 amp hours.
I was going to just let it run unattended for about a week to get some decently high figures, enough to make a reasonable comparison.
The spec sheet of the power meters does not mention a 100Ah limit, which seems low for a 130 amp instrument.

So I have now reset both power meters to zero and will try again, but this time make sure I take readings before the power meters fill right up.
With some decent sun, that should happen fairly quickly with about 8 theoretical amps of solar available to each controller.

Two observations, my controller adjusts far more quickly in rapidly changing conditions, showing a very brief advantage in charging current.
I am only running one fairly crappy mosfet in my controller that has a fairly high rdson (58 milliohms) and my choke runs warm. That inefficiency could be costing me. Also, I have no idea how equal the two seperate banks of solar panels are. They are pretty old, bought second hand about five years ago, but they are a matched set. Until I swap the two banks over, no way to really know.

I will assemble a second constant voltage charger, and I might be able to use mosfets and choke out of a dead Make Sky Blue, and run it at the same switching frequency as the Make Sky Blue. That should make the power losses in the electronics in each about the same.

Constantly trying to make this as fair a test as possible, but that might involve a few different measurement runs to get things as even as possible.
The closer all the hardware is in each system the better, leaving only the mode of operation being the difference would be ideal.
 
I noticed during all the previous testing that my choke was running pretty hot, so I decided to fit a huge oversized 30 amp toroidal dimmer choke instead of the original 15 amp toroidal dimmer choke. My circuit board has provision for two parallel mosfets, but I have only been running it with one so far.
So its now running with both mosfets and the much larger choke.

Re zeroed the two amp hour meters and set up for another test, still with 0.1 ohm series resistors between each controller and the battery.
A nice sunny day today for a change, both controllers bulk charging very strongly. My controller ran stone cold at around 7.5 amps, the Make sky Blue controller ran quite warm at slightly lower current. At the end of the day, still bulk charging, my contoller was ahead in amp hours into the battery by +3.7%.

Previously it was behind by 4 to 5%. So it seems that when bulk charging, power conversion efficiency becomes far more significant than mppt versus constant voltage mode of operation. Hardly surprising really, as all day, the solar voltages were always within a few volts of each other.

When it was dark, I re zeroed both amp hour meters, and swapped over the two banks of solar panels. Another sunny day forecast for tomorrow.
 
Last edited:
The system ran for another day with everything set the same, except the solar panels have been swapped over between controllers.

Another beautiful spring day here, clear blue sky all day, and the mppt controller ended up being ahead by +3.9% at the end of the day.
I had to run some extra continuous inverter load during the afternoon, to stay bulk charging right through the day.

So the small difference has turned out to be between the solar panel banks (and maybe the different wiring gauges and lengths), not the controllers.
I had to run two extra fairly long wires up to the roof using odd scraps of cable I could scrounge, to rig this test up.

To be completely objective, I did notice that the mppt controller gained most around sunrise and sunset, but my controller showed a very slightly higher charging current during the mid day peak. Overall there is nothing between the two.

The mppt is highly likely going to be better in very marginal solar conditions, I was seeing things like 600mA mppt, versus 500mA constant voltage controller, a significant 20% difference. So 60 watts, versus 50 watts from a 1Kw string of panels.
That extra ten watts is not going to make a huge difference in the overall big picture.

When it really comes down to it, adding one extra solar panel to the series string would gain me far more than fretting about a very few percent power loss in the controller or the wiring.

I never tried to optimize the solar voltage setting on my controller, I just set it to the max power printed on the rating plate, and that appears to work quite well.

Next, I will run just my controller with the two amp hour meters connected in series for comparison, and I will try to plot a curve of solar voltage versus charging current.

I have a sneaking suspicion that the rating plate data is all measured at 1kW per square metre "standard" solar illumination, which is higher than we can actually get.
The actual measured power peak in real sun might not be at the exact voltage that the rating plate says it is. The published curves do show that the voltage peak falls slightly with reducing insolation.
 
Back
Top