5 hours ago, plonkster said:

From what I've read the control loop is way simpler than that.

You could well be right, there may be no need for an inner loop with PI control, just the outer loop with MPPT directly affecting PWM. Except it's not always MPPT, sometimes you are charging a battery and supporting loads, and don't want to overcharge the battery. In fact, I seem to recall it's current based, i.e. the Axpert DSP will figure out a target current for the SCC, and it does its best to achieve that current if it can.

So there still may be an integral wind-up problem with the battery charger, explaining the battery voltage overshoots. But now I see that even if true, this likely is not related to the PV bird dropping from the sky problem (great meme, that one).

The discussions on the reasons why the Axpert MPPT controllers dropped output when a large load is applied has certainly enjoyed substantial coverage. More importantly perhaps is a solution for the problem and I believe that it could be solved by an additional feature of the ICC-RPi program of @Manie. In conversation with Manie I promised to formulate my proposal more comprehensively and am still working on it. In principle I proposed  a setting for a maximum load and if exceeded the Axpert would switch to grid to prevent large currents being drawn from the batteries. When the load drops below the set value and after a suitable delay to prevent switching back and forth, the Axpert would revert to solar/battery. This should be subject to the SOC being above the minimum setting if one has ICC controlling the inverters based on SOC. If you are not using ICC my question would be, why not, and this post obviously would not apply. Such a feature would protect the batteries from excessively high load currents. Perhaps other members would give it a thought and express their views or alternative proposals.

@ebrsa sounds like a quite elegant solution but my thinking on this, seeing as we are also in the software game:

What is the real impact on the batteries over time?

Does it really matter to draw a bit of current every now and then?

Does it not just stir the juices having no real effect over time?

Because the more parts / segments to a system, the more complicated, the more the potential for a FOOBAR versus KIS.

 

This "problem" is for all electricity suppliers the same, even Eskom, when huge loads are added for a short period of time.

For what I have seen with the software is that it is not instantaneous to make the decision to change so batteries are still going to have to carry the load for a few seconds maybe.

But yes, ICC would be faster than the Axpert but whilst the MPPT is searching, the load is moved, would that not extend the search time?

And then few minutes later the load is switched back to the Axpert. Now the MPPT starts searching again to adjust the increased load?

 

The questions I would ask is:
Has anybody experienced battery problems because of this? Can it be quantified?
What is the rand/cents benefit for the added complexity?

I believe the batteries can handle it.

1 hour ago, ebrsa said:

In principle I proposed  a setting for a maximum load and if exceeded the Axpert would switch to grid to prevent large currents being drawn from the batteries

Would it be able to switch fast enough?

Certainly with my modestly sized 200Ah batteries, a large load on the output for even a second (well-point pump startup for example) is enough to momentarily drop the battery voltage, and that will send the MPPT into a scan.

1 minute ago, The Terrible Triplett said:

Does it really matter to draw a bit of current every now and then?

That is a good question. If you really consider what happens on a chemical level -- and I am by no means an expert here -- you have a bunch of charge sitting on the plates that haven't yet been converted into chemical energy. If you were to monitor the terminal voltage when you switch from charge to discharge in this fashion, what you would see is that the voltage drops and then picks up slightly as you remove the charge from the plates and the chemical process gets going in order to replace it.

So I think it is a good question to ask: What really happens if you do a bit of discharge for a few seconds, are you really digging into the chemical side of the process much, or not? I don't know, really I don't. I suspect 1 or 2 seconds is nothing to worry about, you're coasting off residual charge on the plates (which you just put there a few seconds ago), sort of like a capacitor. 30 seconds would be another matter, but I still wonder how bad that really is.

Would it not be accurate to say that if such a load pulls the batteries down so much that you are alarmed about it, because the MPPT went sightseeing for 30 seconds... well then you have other problems already? :-)

The big thing with switching on high loads is that it is near impossible to gauge when to do that, I considered to have my genny "kick in" when ever we were to exceed a certain load. But then if the load is for 30 min it will be ok but what if it is the microwave and it runs for only 20 seconds or even a minute by the time the genny is hot its all over. So I binned the whole idea. 
If your loads is too high for too long rather place those on a separate circuit directly from the grid and switch to grid only on you soc.

Anyway that what I would do if I had eskom. 

 

TTT thanks for your views. This is exactly what I am looking for as battery technology seems more like witchcraft than science to me. I have seen battery load currents as high as 96A from my 450Ah bank and found that a bit alarming. What a safe load from a bank would be remains a mystery and I have not found any information on the websites of manufacturers. Agreed, there could be all kinds of problems but if the maximum load was set very high on the feature I propose, the inverter will never switch based on battery load so it is under control of the owner and his choice. I sure hope some more information with references to the source will be forthcoming. Of course you may wish to consider a similar feature for your SolWeb of which I have to admit I have zero experience. But sharing is what the forum is all about or should be and by and large it succeeds admirably.

@PaulF007 you have a point. My problem is that the electrician who wired my house made that impossible with the way it is wired. So in desperation to avoid potential overload of the first Axpert I installed, the easiest solution was to add a second one in parallel. One can also add a programmable time delay before switching to grid by the control program takes place. Thanks for highlighting the issue, I have to confess that I did not think about a high load of short duration which will most likely not cause any harm to the batteries. You live and learn with the emphasis on live at my age.

16 minutes ago, ebrsa said:

... witchcraft than science to me ...

O no, it is quite scientific - mathematical, tried and tested. But yes, some marketing material tend to elevate it to a magical level. 

Your T105's, can take it. It says they can take a draw of @75 Amps for 115 minutes, so 96A is fine for a few seconds. 

http://www.trojanbattery.com/product/t-105/

@ebrsa here is a very informative site on all things battery: http://batteryuniversity.com/

If there is space in your board you could also consider a ECU . We used it on our geaser where if the load from the Kitchen were to exceed 10a it would cut out the geaser for the duration of the load.It is a digitla unit so it works instand but you also get a LCU that works off bimetal strips so if you exseed a certain amp lever of longer that 60 sec it will cut the geaser out and once the stips cools down (ie no more load) the swicthes on again.

Worth a look at.

@The Terrible Triplett thanks for the URL's, I have been there before. Cannot help but question whether the manufacturers are always telling the truth, the whole truth and nothing but the truth so help me.........Consider a situation where one regularly discharges to say 80% SOC at 75A with one at 20A. Which bank would last longest. I would put money on the 20A one. At present I run on battery/solar power from 2:00 am until 17:00 pm just to avoid large loads at supper time. The batteries usually do not discharge much below about 78% SOC. @PaulF007 fortunately the geyser is not a problem the element is 2KW but it has not been on since winter as I use a 16 evacuated tube heater for domestic hot water and right now the top of the tank is at 57C. It is a sunny day but we are never short of hot water. As my installation is not a year old yet, I hope we will have enough sun in winter so that I can boost the temperature a bit with solar electricity if need be on cloudy days.

12 minutes ago, ebrsa said:

... manufacturers are always telling the truth, the whole truth and nothing but the truth so help me ...

Very good question.

Edmund once got info that Trojan have something akin to an expiry date. If they say X cycles at Y% DOD, that is their "expiry" date. Batteries could very well, and tend to, last longer so you are assured you will get your cycles if you as the user have all the T's crossed and I's dotted as per their documentation. And they respond to emails too.

So for me what Edmund found out made sense for forklift and golf cart users would not accept  / tolerate anything less than the truth.

FWIW: Profits on batteries are extremely low (if you shop right) as well as on Trojans so suppliers need to very make sure they work out of the box, as there is no fat in for sorting problems afterwards.

And I chose Trojan over many other batteries because they are worldwide and for SA circumstances, with good local supplier backing, I could not ask for more, for as you may know, I was behind SA Trojan batteries here on the forum, till I handed it all over to PowerForum Shop.

22 minutes ago, ebrsa said:

fortunately the geyser is not a problem the element is 2KW but it has not been on since winter

Those tubes make the world of difference! But you might be able to use the relays to cut "non essential" loads should you have a spike for a short while. 

20 minutes ago, ebrsa said:

Consider a situation where one regularly discharges to say 80% SOC at 75A with one at 20A. Which bank would last longest. I would put money on the 20A one.

And I would place my money on not going below 80% DOD, to get my cycles as per the specs, irrespective of the amps drawn till 80%.

There is also a good argument to be made that if you do not use the batteries, they will maybe last as per specs, but not give you the ROI you could have gotten if you used them to say 50% DOD.

Problem is with us all being "new" to batteries with the huge costs involved to replace the bank if we get it wrong, no-one has any real life experiences using a number of banks.

Nor do we carefully measure the batteries cells every day, nor do we keep a log of daily temps of the batteries. So we are close to flying blind.

@PaulF007 asked questions about the lifespan to users of Trojan batteries. Turns out the batteries will last X years, whether you use them or not, he found.

The guys having to use batteries for off-grid applications, the real "harde baarde", check their batteries every day and they have banks lasting +-15 years.

So it all goes to how much effort we want to put in to check each cell.

15 hours ago, Don said:

Assuming, they are aware of the problem. 

I'll report it - the more people that report it the more likely they are to react

Here's a thought, 

On 1/18/2017 at 10:14 AM, ebrsa said:

TTT thanks for your views. This is exactly what I am looking for as battery technology seems more like witchcraft than science to me. I have seen battery load currents as high as 96A from my 450Ah bank and found that a bit alarming. What a safe load from a bank would be remains a mystery and I have not found any information on the websites of manufacturers. Agreed, there could be all kinds of problems but if the maximum load was set very high on the feature I propose, the inverter will never switch based on battery load so it is under control of the owner and his choice. I sure hope some more information with references to the source will be forthcoming. Of course you may wish to consider a similar feature for your SolWeb of which I have to admit I have zero experience. But sharing is what the forum is all about or should be and by and large it succeeds admirably.

A battery bank can safely deliver far more than that, if the load demands it. And, then, off cause, the cables, fuses, etc need to handle the load as well. 4KW @ 48V equals 83A. 6Kw at 48v = 125A. 10Kw at 48V = 208A!

On 1/18/2017 at 10:00 AM, plonkster said:

Would it be able to switch fast enough?

Certainly with my modestly sized 200Ah batteries, a large load on the output for even a second (well-point pump startup for example) is enough to momentarily drop the battery voltage, and that will send the MPPT into a scan.

So, how about using capacitors to absorb these spikes, instead of switching loads forward and backward? I haven't researched this, on a large scale such as on 5Kw inverters, but it works well for car sound systems on 12V circuits. 

37 minutes ago, SilverNodashi said:

Here's a thought, 

A battery bank can safely deliver far more than that, if the load demands it. And, then, off cause, the cables, fuses, etc need to handle the load as well. 4KW @ 48V equals 83A. 6Kw at 48v = 125A. 10Kw at 48V = 208A!

Hi Silver

You are in the industry so I am going to pose a question. My batteries have a max discharge current of 2600A and a maximum charge of 78A. Not that I approach either of these two values on discharge or charge.What rate of discharge would you recommend folk do not exceed. I know that this is going to be affected by the size of the battery bank (either through multiple battery strings or big 1000Ah 2V cells that can handle big discharges). Let us have a hypothetical 8 battery 6V 225 Ah Trojan RE battery bank. The data sheet does not even mention a maximum discharge. Folk can then have a ball park figure based on how their own battery bank stacks up to a Trojan 48V 225Ah  battery bank.

I am sure Mike and many others will be able to make a valuable contribution.

3 hours ago, SilverNodashi said:

So, how about using capacitors to absorb these spikes, instead of switching loads forward and backward? I haven't researched this, on a large scale such as on 5Kw inverters, but it works well for car sound systems on 12V circuits. 

Given the thick cables and the relatively small amounts you can store in a capacitor, something tells me it's not going to work. I think it works for car audio systems because you only have to "back up" a couple hundred milliamps in most cases.

One thing that might work, and this might sound a little odd, but if you have a separate sense wire for the MPPT (some of them have it, obviously not the axpert), you could hack a small resistance (insignificant compared to the input impedance of the measuring circuit) and then put a capacitor across that. A sudden drop on the battery will then be filtered out (you may have to experiment) as the cap has to discharge via the resistor to be pulled down to the battery voltage. In other words, you cheat the MPPT into not seeing those low events by smoothing the voltage waveform :-)