Friday, 20 October 2017

The Cheapie controller is MPPT - not PWM

Well I've had a better chance to fiddle with this now, and I'm quite sure that this little UEIUA CPS-2420 controller is properly an MPPT controller (unlike the views of others)

So here is my evidence for that. PWM will essentially pull the panel voltage down to very close to the battery charging voltage. This is because it can't do voltage to voltage conversion. Given that people normally run a "nominal" 12V panel, which is often open circuit at 21V this is less of an issue than it may seem. Panels usually  have its maximum rated power at a voltage something like 17V ... in the case of my black 100W panel is actually 17.8V (... which is of course at the maximum obtained measuring at Standard Conditions , namely full sunlight but with a panel temperature of 25°C). So the drop from there to 13 or so optimal for charging isn't a significant loss. Such would be a case of a panel at more or less optimal voltage match for the purpose, and PWM just has to groom it a little.

But if you run panels in series not in parallel then things change and PWM is at a disadvantage.

So in the above experiment we see that with two panels in series we are clearly getting MPPT because:

  • circuit voltage of the two series panels was brought to 38.6V
  • the panel was delivering to the MPPT controller 1.5Amps
  • the battery was being held at a charge voltage of 13.5V
  • with an observable 3.56 Amps going into it
were this system a PWM system this would simply not happen.

So the youtube reviews you may watch on this controller make what I believe to be the error of solely determining the capacity of the controller based on a panel which is already closely matched to the usage (battery charging) and measure what it puts out into the battery.

I've already shown (in previous posts) that the system is capable of delivering a few extra amps to the load without taking all that power from the charging. In a previous experiment (with one panel) I was able to deliver 3.5Amps into my load (from the load port) while still pumping 0.56Amps into the battery (3.5 + 0.56 = 4Amps) yet it failed to deliver 4 amps to the battery when simply charging the battery. This leads me to see that it can (and does) produce more amps, but only if the entire system requires it (in its view).

Clearly this new test demonstrates that the controller is able to deliver the full panel capacity to the "system"(given as also observed the panel temperature pulling down its maximum power delivery capacity). 

This word SYSTEM is the critical point. The controller is intended to be a part of a system, not just a stand alone battery charger. As such if we take that viewpoint and explore its capacity as a system we can see that it is delivering the capacity of the panels into the system, and not just pumping it into the battery.

I went and did a second test (because the sun came out as I was writing this) and found the following:
  • two panels in series was loaded by the controller to produce 6.48 Amps @ 20 Volts input into THE SYSTEM.
  • the controller then poked 6 Amps into the battery @ 13.2 Volts (you need to recall that battery at rest unloaded is a different voltage to the battery under charge and the controller has NO WAY to know what the resting state of the battery is until perhaps night)
  • the controller still fed the load with 3.5 Amps @ 12.8 Volts
so if this was PWM we would still only get a maximum Amps of 5.5 (the short circuit amp rating and ignoring panel temperature) yet it was able to feed 6 + 3.5 = 9.5 Amps into the system

Considering that the combination of the two cells was in theory only able to deliver 11amps,  that this controller pulled 9.5Amps to feed to the other two demands on the system is pretty darn good.

Clearly MPPT, not PWM.

If your goal is to shove the most power from your panel into a battery (from which you draw load in parallel to charging) then (as observed) the CPS-2420 MPPT controller is perhaps not your best bet, especially if you have only one panel producing (an open circuit) voltage of around 20V.

For that role I've found the little T20 (which I bought earlier because it was cheap) to be at an advantage there as it puts out a few more amps into the battery (with nothing on its load) than the CPS-2420 does.

Its a nicer looking unit in some ways but perhaps more confusing because it has more options, allows you to configure more and the menu is not straight forward to navigate ... this is them side by side,

With two solar panels in series the T20 it was still only delivering the same amount of amps as with one panel (and pulling the whole rig down to the lowest common denominator voltage) which is what PWM does. Here it is on the two panels (which in series produce the same amps, but double the volts).

as you can see its not able to take advantage of the potential 40V available to it. Because its PWM.

So where does this leave us?

It means to me that the choice of which controller needs to be a decision based on "what are you goals".

In fact I still like this little white controller (even though it lied about being MPPT and is really PWM) because it provides some additional features which I find useful. I mainly use it with a small panel (10W) to keep the battery conditioned on my ride on lawn mower.
I like that I can set it up for myself:
  • battery max voltage (so it won't over charge and will cope with a variety of battery types)
  • battery min voltage (where it will kill the load were I running one, and should it go below a voltage value I set)
  • a display of the current battery voltage (so I don't need to pull out my multimeter to test it)
  • I can leave it alone and know its doing its sole job
Keeping my mower battery trickle topped up (in a good manner and voltage sensitive) is an ideal usage. The mower sits idle in winter and can result in dead battery (from self discharge and sulfation). This controller and a 10W panel prevents that.

Even if you wanted to a run a short term load off your battery then this little controller does well as long as you manage that load yourself.
If you have timed operations in mind then this little controller does that too ... if you want more amps out of it then you'll have to run your panels in parallel (to minimise the losses of voltage, while doubling your amps). Just make sure you've got your diodes set up right with that ...

As I've read elsewhere PWM isn't simply worse than MPPT, it provides things differently and at a lower cost (although at this price point that's moot).

As it happens these two controllers are amazingly low cost and both do things differently, so picking one will be determined by your needs, your setup and your system.

Following are some sections from pages on the web which I have found salient (and their URLs for reference):
From this link

The preceding discussion of PWM vs. MPPT may cause some to wonder why a PWM controller would ever be chosen in favor of an MPPT controller. There are indeed instances where a PWM controller can be a better choice than MPPT and there are factors which will reduce or negate the advantages the MPPT may provide. The most obvious consideration is cost. MPPT controllers tend to cost more than their PWM counterparts. When deciding on a controller, the extra cost of MPPT should be analyzed with respect to the following factors:
1. Low power (specifically low current) charging applications may have equal or better energy harvest with a PWM controller. PWM controllers will operate at a relatively constant harvesting efficiency regardless of the size of the system (all things being equal, efficiency will be the same whether using a 30W array or a 300W array). MPPT regulators commonly have noticeably reduced harvesting efficiencies (relative to their peak efficiency) when used in low power applications. Efficiency curves for MPPT controller are printed in their corresponding manuals and should be reviewed when making a regulator decision.
2. The greatest benefit of an MPPT regulator will be observed in colder climates (Vmp is higher). Conversely, in hotter climates Vmp is reduced. A decrease in Vmp will reduce MPPT harvest relative to PWM. Average ambient temperature at the installation site may be high enough to negate any charging advantages the MPPT has over the PWM. It would not be economical to use MPPT in such a situation. Average temperature at the site should be a factor considered when making a regulator choice.
3. Systems in which array power output is significantly larger than the power draw of the system loads would indicate that the batteries will spend most of their time at full or near full charge. Such a system may not benefit from the increased harvesting capability of an MPPT regulator. When the system batteries are full, excess solar energy goes unused. The harvesting advantage of MPPT may be unnecessary in this situation especially if autonomy is not a factor.

and this link suggests too:
The Solar input nominal voltage must match the battery bank nominal voltage if you’re going to use PWM
So if you do have a single "nominal 20V" panel (which probably puts out its max power point at 17, but amps at 14 will still be close to max), then you are matching the battery to the nominal voltage (especially when its a hot climate like Australia) and so you'll see less advantage to MPPT. Which I think is the other testers issue; I don't think he applied sufficient voltage advantage for MPPT to give its best run.

So know your needs and pick you animal :-)


I like to do things in a scientific manner, if you have any issues with my conclusions then proper scientific method would suggest that you examine my methods and identify what I did wrong. Try and replicate the results if you can before just slamming me.


bloggermouse said...

I've ordered one but haven't received it yet.

I am convinced the unit does DC-DC conversion (explaining current gains). I am less convinced it is doing MPPT in the normal sense. A much cruder algorithm that would line up with both your and Adam's observations might look like this:

1. detect Voc
2. guess at Vmp based on a common correlation: Voc x .8. This is a common starting point in more refined MPPT algos, which use it as a starting point. A crude MPPT algo would use it as a "good enough" ~Vmp.
3. run the panels at ~Vmp and DC/DC downconvert until battery reaches a given setpoint
4. then run the panels at Vbatt + 2v and and DC/DC downconvert. PWM or shunt to stay at the setpoint.

I don't have a problem with that kind of setup, particularly if they A) reliably restart the algo when battery voltage falls, B) keep the price down to about $25 shipped from CN and 3) be honest about what it's doing. Heck, I'd market it as value added feature. "Simpler algo saves you money and still gives more power than PWM!"

I have no explanation about why the unit responds to loads on the LOAD output but not on the battery itself, but if it's happening then it's happening. Hard to disagree with empirical testing. When I get mine I'll test that, and also test if it acts the same when the LOAD draw is light, as when using a relay from LOAD to control inverters and other non-trivial loads.

obakesan said...

bloggermouse, thanks for your comment.
I believe that Adam's assumptions are flawed, and that actual MEASURED maximum powerpoint is dependent on a few more factors. For instance temperature being away from the "ideal" of 25C. Adam also fails to factor in efficiencies (what is the max theoretical efficiency of a toroid, the switching and circuit parasitic losses (which will be higher with the tiddly voltages and amps he's talking about) and even the losses engendered by actually measuring the current).

I agree that there is DC-DC going on (how else can one explain the panel side being 30V with two in series and the battery side being 13V ... it simply can't be PWM

As to why it can't detect load how could it without a load measuring shunt ... this is what all load types do. The fundamental assumption of the circuit is that its charging the battery (of which it has no way to know if its 5Ah or 120Ah type and no way to be told that). This much is clear: the design concept of this simple machine is to have load on the load.

Should one (like me) prefer to put an inverter on the load directly (for quick peak high draws) then with a sufficient battery that won't upset it.

I've found that the controller does shut the load off when the voltage falls below (I forgot to write it down, but if I recall right was about) 11.9V

Please post your testing and I'll link to it :-)

bloggermouse said...

I think Adam's concern was that panel voltage was falling off drastically from 40v to less than 20v, even when loads were added. Then not responding to loads (indirectly) on the battery terminal. I agree about the losses and that cell temp higher than lab conditions moves Vmp downward.

What I meant about load was I haven't seen a controller that treated, say, 15A of demand in bulk charging and 15A on the load output differently. One would think PPT would adjust Vpanel to match that 15A demand, regardless where it is going (BATT or LOAD terminals). If I am reading your tests correctly, the PPT algo would respond to loads on LOAD terminal and not loads on the battery. That just seems weird to me. I hope I can duplicate it and try to figure out why it happens.

Just ordered mine today and it's on the slow boat so it will be a while....

Thanks for experimenting and posting. It's valuable stuff.