Jump to content
Sign in to follow this  
OomD

Axpert "back to grid" behaviour

Recommended Posts

I wonder if there is any nice document around somewhere, that describes the Axpert's thinking process.

For example, I got home yesterday and the Axpert was in bypass mode. My batteries (8 x Narada) were at 51.3V, and according to the BMV712 they were at 94% SOC. The last full charge was about 14 minutes before, according to the BMW. Now, it might be that the BMV recognised a full charge, but the Axpert did not (running 73.00b).

Anyway, I wondered why the Axpert was on bypass. In my mind there's only 2 reasons:

1. Overload condition.
2. Batteries reached the "back to grid" level.

In my case I know its not #2, and I'm not aware of any heavy load prior to me getting home. Setting 4 is on Bypass Enable, though.

What I am curious about though, is why did the Axpert not switch back to batteries again? There was no significant load (about 400W), thus no reason to stay on grid. My setting 1 is SBU, so I expect the Axpert to stay on batteries/solar until the "Back to grid" volts are reached (setting 12, mine is set to 50V).

Does anyone know what conditions makes the Axpert go back to batteries, once it has switched to grid? Maybe only the next time they are charged, or there is sufficient solar?

Share this post


Link to post
Share on other sites
1 hour ago, OomD said:

Does anyone know what conditions makes the Axpert go back to batteries, once it has switched to grid? Maybe only the next time they are charged, or there is sufficient solar?

Setting 12 triggers going to grid and setting 13 triggers going back to solar/battery.

What is your bulk/absorb voltage and what is your float voltage?

What is your charging?  CSO  OSO CUT SNU?

Share this post


Link to post
Share on other sites
4 hours ago, Chris Hobson said:

Setting 12 triggers going to grid and setting 13 triggers going back to solar/battery.

What is your bulk/absorb voltage and what is your float voltage?

What is your charging?  CSO  OSO CUT SNU?

Thanks for responding, Chris.

Absorb = 56.4
Float=55.2
Charging=OSO
12=50V
13=FUL

From your questions I deduce that the Axpert did not complete the charging (even though the BMV thought it did) and thus did not go back to batt?

Share this post


Link to post
Share on other sites

[12] = 50V is extremely high and will cause the system to switch back to grid when the batteries are still very high. I recommend a setting between 46V and 48V (48V is still quite high).

ICCout.thumb.png.bcbf800f40a1fe5006151f91d452966c.png

In the above example you can see my batteries sitting just under 55V while charging at about 40amps, then a high load came on and my batteries started discharging at 40amps causing the battery voltage to drop from 55V tp 50V - the SOC was around 80% at the time.

I use 46V as a safety net in case ICC fails, because ICC switches my system from battery to grid at 60% SOC. At this time my batteries would be at about 49V with my base load of 750watts, but if a fridge or something kicks in it will drop momentarily to 47/48V then pick up again. If I set 48V my system would switch when not yet necessary.

Share this post


Link to post
Share on other sites

Thanks, pilotfish. Yes I realise its high, but I want to try and maintain a 20% DOD max. Works out to around 50V. I know that the Axpert's voltage monitoring is causing the premature back-to-grid here, and not the actual SOC.

I am more concerned about the fact that the inverter does not go back-to-batt again, and trying to understand the inverter's behaviour.

Ultimately, I suppose, getting ICC is the better option (I simply don't get the time to finish my own SuperPV project).

Share this post


Link to post
Share on other sites
Posted (edited)
29 minutes ago, OomD said:

I suppose, getting ICC is the better option

Axpert is an excellent product when used in conjunction with ICC [edit: + a BMV], and junk without it. The cost to use it while developing your own ICC alternative will pay for itself in protecting your batteries from damage in the interim.

Edited by pilotfish

Share this post


Link to post
Share on other sites
28 minutes ago, OomD said:

I am more concerned about the fact that the inverter does not go back-to-batt again, and trying to understand the inverter's behaviour.

It wont go back to batt until [full] again.

Share this post


Link to post
Share on other sites

Hi OomD

OK

 50.9V is 100% SOC for a rested battery and 80% SOC is about 49.7 V for a rested battery. So attached to the Axpert I would say about 49V. As @pilotfish has mentioned a large load can artificially cause a drop in voltage and as soon as the load is removed the voltage recovers.  Thus I have always recommended 48V and 47V if you bank is on the small side. @pilotfish has a large battery bank and he has lowered it to 47/46V  so perhaps my recommendation is too high. 

I set program 13 to 54 V since at that voltage I know that real charging is happening. One could have the situation early in the morning where there is insufficient sun to do real charging and then your inverter is going to switch back and forth  between grid and solar.  I would not want to set program 13 to FUL as that might take hours. If you have a small bank I would perhaps raise it to 55V but definitely not FUL as in my case I would not be using the solar array and that defeats the objective in my mind. Yes use grid because the battery voltage is dropping but don't charge  fully using grid.

Are you using vanilla firmware?

Share this post


Link to post
Share on other sites
2 hours ago, OomD said:

maintain a 20% DOD max. Works out to around 50V.

Voltage control is tricky. Though 50V might equate roughly to 80% SoC, that is when the battery is open circuit and left standing for an hour... or at least with no load on if you cannot do that. So the challenge is to pick a voltage that is low enough to avoid unnecessary switching, but high enough so that it will "bounce back" to 50V a few minutes after it switched (and therefore equate to roughly 80% SoC). If you know your loads well and you know more or less what kind of load you have late afternoon/early evening, then you can usually determine a good switching voltage through empirical research and successive adjustments.

Share this post


Link to post
Share on other sites

Thanks for all the replies, guys, and also the great detail Chris.

I'll lower the voltage to 49V, I understand 50V is a bit high given the Axpert's (erroneous) use of it to determine SoC. @Chris Hobson I'm using 73.00b currently, not going to 73.00c yet until I decide finally what to do re parallel use of 4KW/5KW inverters (different thread). I'll adjust the back-to-batt setting to 54V, thanks for the advice. But then, should I not be lowering the float too? I mean, if float is set to 55.2V as it currently is, it will technically not get down to 54V? Or does the setting of 54V for back-to-grid mean 54V and greater?

Once again, I appreciate the feedback, guys. I have purchased ICC in the meantime and busy getting my Raspi up and running, and will probably eventually hook it to the BMV as well. Just need to get the VE.Direct cable.

Share this post


Link to post
Share on other sites

Hi OomD,

 

  In my case, the parameters are 49 and 54. 50V is too high given the size of my battery bank (250 Ah at 48V). As a piece of advice, to monitoring my installation remotely, I use watchpower in an Intel Compute Stick in Windows (Also in the beggining I used an old laptop), and connected to internet via Anydesk. I can manage any parameter from my Iphone, Ipad or any computer.

 

Regards,

Share this post


Link to post
Share on other sites
3 hours ago, OomD said:

I mean, if float is set to 55.2V as it currently is, it will technically not get down to 54V? Or does the setting of 54V for back-to-grid mean 54V and greater?

Yes, putting setting 13 at 54 V means 54 V or greater. It just needs to be at 54 V or greater for about a minute. But there is a minimum of 2 minutes before it will switch (patched firmware; I think it's 10 minutes for factory firmware).

It must have gotten below the float voltage to have gone to bypass mode (as far as its measurements are concerned; you might not see all the measurements externally).

As others have noted, your main problem is leaving setting 13 at FUL. That means don't go back to battery mode unless the Axpert judges the battery to be full. Then, it's not looking at battery voltage at all; I think it waits till the next time you go to float mode (maybe if you're already in float mode it doesn't count).

Share this post


Link to post
Share on other sites
7 hours ago, Coulomb said:

Yes, putting setting 13 at 54 V means 54 V or greater. It just needs to be at 54 V or greater for about a minute. But there is a minimum of 2 minutes before it will switch (patched firmware; I think it's 10 minutes for factory firmware).

It must have gotten below the float voltage to have gone to bypass mode (as far as its measurements are concerned; you might not see all the measurements externally).

As others have noted, your main problem is leaving setting 13 at FUL. That means don't go back to battery mode unless the Axpert judges the battery to be full. Then, it's not looking at battery voltage at all; I think it waits till the next time you go to float mode (maybe if you're already in float mode it doesn't count).

OK that clears it up a bit.

But... sorry to be nitpicky, I just need to understand the consequences of using it like this. If set to 54, with bulk set to 56.4 and float set to 55.2, this will cause the inverter to switch to batteries while the batteries are still in bulk phase. As the battery voltage ramps up, and crosses over 54 for about a minute, then the load is switched to battery. This creates the possibility of the  batteries never really being fully charged, depending on actual load and solar availability.

In the long run the batteries crystalise more with this scenario, don't they? Unless, off course, a periodic equalising charge is used to offset this.

Note please, I am not arguing the correctness of your info, I really appreciate the feedback. Just trying to understand it.

Share this post


Link to post
Share on other sites
Posted (edited)
34 minutes ago, OomD said:

If set to 54, with bulk set to 56.4 and float set to 55.2, this will cause the inverter to switch to batteries while the batteries are still in bulk phase. As the battery voltage ramps up, and crosses over 54 for about a minute, then the load is switched to battery. This creates the possibility of the  batteries never really being fully charged, depending on actual load and solar availability.

That's true. Most installations have enough solar panels that you'd be fully charging your battery most days anyway. So you may as well not use grid power while charging.

I have a lithium battery, so it doesn't matter if it doesn't get fully charged every day. Even with lead acid, unless your PV input is marginal, I don't think it's worth worrying about.

Edited by Coulomb
Added "fully"

Share this post


Link to post
Share on other sites
3 hours ago, Coulomb said:

That's true. Most installations have enough solar panels that you'd be fully charging your battery most days anyway. So you may as well not use grid power while charging.

I have a lithium battery, so it doesn't matter if it doesn't get fully charged every day. Even with lead acid, unless your PV input is marginal, I don't think it's worth worrying about.

Makes sense, thanks!

Share this post


Link to post
Share on other sites

Adding to Couloumb's response, if my lead acid batteries are fully charge, in winter, at least once a week, I am happy.

Batteries are going to last X time, either use them in that time, or throw away "unused batteries" due to the age.

The only golden rule, try not to go below 20% DOD (SOC of 80%).

Having said that, once every few months, on lead acid's, I have no issue going down to 60% SOC. Let the batteries work a bit say I.

Share this post


Link to post
Share on other sites

@OomD setting up the Axpert voltages has been a topic of much discussion over time. I was advised by Jaco Fourie who originally released AICC, developed by him and @Manie and the forerunner of ICC, to set the parameters concerned at low voltages to prevent clashing with AICC. So my parameter 12 (battery to grid) is at 44V and parameter 13 (grid to battery) at 48V. Parameter 26 (absorb voltage) is set to 58.4, the highest the Axpert is capable of as it should be 59.3V for my Trojan T105 batteries. I compensate for this by setting parameter 32 to 120 minutes so as to force the maximum charge voltage for some time as suggested by Trojan. Parameter 27 (float voltage) is set to 54V which is the float voltage for the T105. Parameters 26 and 27 should be set to the voltages specified for your particular brand of battery. However the user manual for Rolls batteries, for which @pilotfish posted a link on another thread, contains most useful information. Just Google it and download.

Of course all this assumes that you control your Axperts with ICC, either Raspberry Pi or Windows version. Mine is the RPI with a BMV700 connected. The BMV is essential and if I have to buy again it will be a BMV702 with its midpoint measuring feature. I am still battling to find the right settings for the BMV700 so that it won't be necessary to set parameter 32 but leave that to the BMV. Maybe Victron experts like @plonkster will have some advice, for all of us who are using ICC, on the correct settings for the BMV 700 or 702. 

The question also arises that if ICC takes care of proper battery charging, would we still be affected by the shortcomings in the Axpert charging firmware that the Aussie firmware addresses. Perhaps @Coulomb or @weber would be so kind as to enlighten us. Nevertheless I run the Aussie firmware as it is clearly better than the Voltronics effort. I am considering replacing my 2 Axpert 5Kva/4Kw units with 2 5Kva/5Kw units and then it will be back to Voltronics firmware. Just need to find out what the repair cost of my faulty unit will be to make an informed decision. So it would be good to know that ICC will address the charging problems, assuming that is the case. Learning about this stuff is just never ending but at least also fun.

Share this post


Link to post
Share on other sites
4 minutes ago, ebrsa said:

correct settings for the BMV 700 or 702

Truth be told, I don't think there's just one set of settings that works for everyone. You have to tune it until it works well for your circumstances. The rest of your equipment plays a role too.

For example, a Victron inverter has a variable absorption phase, or if you don't want to read all that, essentially absorption is 5 times bulk. This means it spends ample time at higher voltages, and you get better results by setting  the "charged voltage" setting on the BMV to something that's about 0.3V lower than the absorption voltage. This tends to give good reset behaviour, because a low tail current at a higher voltage is a much better indication of a full battery.

But not all equipment is created equally and some manufacturers can't fix their float bugs. So you have to compensate, and the way you do that is to pick a voltage that's about 0.3V lower than float, and use a really really low tail current... and you hope you don't get too many false positives.

Then you pretty much play with the tail current and the charged detection time until you find something that sort-of works for you.

Share this post


Link to post
Share on other sites
16 hours ago, ebrsa said:

The question also arises that if ICC takes care of proper battery charging, would we still be affected by the shortcomings in the Axpert charging firmware that the Aussie firmware addresses. 

Neither Weber or I use ICC, we have our own monitoring software, based on Node Red. So it's hard for us to know what ICC is really doing. But I can't see it affecting when the Axpert goes from bulk/absorb stage to float stage, unless it uses Dynamic Charge Control, and I'm pretty sure it doesn't. And you need the patched firmware for DCC anyway.

The basic problem remains that Voltronic Power don't recognise that there are charging bugs, and they don't provide any mechanism to work around them.

Share this post


Link to post
Share on other sites
Posted (edited)

Thanks very much for your explanation @Coulomb. My understanding of charging by the Axpert is improving and clearly learning never ends. I have not seen any difference on ICC graphs in the time the charge voltage remains at absorb voltage, parameter 26, even with the BMV700 detection time set to 50 minutes and with parameter 32, time at absorb voltage, set to AUT. So it would seem that ICC does not control switching between absorb and float voltages but uses the BMV battery voltage information only to control state of charge switching between battery/solar and grid. The only conclusion I could come to is that all the settings related to charging are ignored by ICC and fiddling with them is meaningless. I will enquire from @Manie about this. It would also mean that should I switch to the 5Kw Axperts, for which you and Weber's patched firmware is not available, it will be a huge step backwards to inferior Voltronics firmware with serious charging flaws.

@plonkster thank you for your reply too, including the link to the document about charging. Perhaps you could give your views on the matter of the BMV charging settings having an influence on battery charging if it is connected to a Raspberry Pi running ICC. I guess we will only know the true facts if @Manie enlightens us.

And yet another question. If I run only one 4Kw Axpert and it is on battery supply, will it switch to bypass if the load exceeds 4Kw and what would be a safe maximum load it would be capable of supplying in bypass mode.

Edited by ebrsa

Share this post


Link to post
Share on other sites
1 hour ago, ebrsa said:

And yet another question. If I run only one 4Kw Axpert and it is on battery supply, will it switch to bypass if the load exceeds 4Kw and what would be a safe maximum load it would be capable of supplying in bypass mode.

Yes, the Axpert will switch to bypass mode on overload, as has been mentioned elsewhere, subject to enabling that facility (setting 23, "overload bypass"). I have no idea why you'd want to turn that off, or why it's off by default.

In bypass mode, the limit is the rating of the relays (40 A) and the AC input breaker (also 40 A). So as long as your cabling can handle it, 40 A is the safe limit. If your mains is 220 VAC, that's 8.8 kVA, likely some 7 kW depending on the power factor of the load.

Share this post


Link to post
Share on other sites
1 hour ago, ebrsa said:

Perhaps you could give your views on the matter of the BMV charging settings having an influence on battery charging if it is connected to a Raspberry Pi running ICC.

As I understand it ICC is the CCGX of the Axpert world. From what I've seen, it's essentially a combination of open source components (such as mysql and emoncms) and some proprietary components (written in FreePascal/Delphi). When the code leaked back in the day I did inspect the components it uses (yes, I was a bit naughty). At least the one that was leaked used many free components inside the proprietary bit, but no license agreements were broken (ie all of them allows for commercial use, no GPL tie-downs). So that is essentially my understanding of it.

What it does is twofold, it monitors certain parameters, draws nice pictures for you, sends them upstream to other services, but mostly what people want it for is the fact that it can read the Victron BMV and that it can modify selected configuration items on the inverter to fake better "go to grid" behaviour. In this respect it isn't really rocket science: It gets a number from the BMV, and if that is lower than some other configured number, it changes the config on the inverter. Done. Not too different from what we do with the Multi, except we don't change config, our changes are more volatile (goes away with a power cycle) and we have a safety catch where the inverter goes to passthru if it gets no external comms.

Back to the BMV though: It needs to reset occasionally. What you want to avoid is a premature reset, but you still want it to reset or rather synchronise with a battery that is full (for some value of full). That is because the battery is not a fixed quantity. The efficiency is variable (on charge rate, temperature), degradation means the capacity recedes, essentially the "full" point isn't in the same place every time. Unless you reset once in a while (once a week preferably), you will slowly drift. In the short term this doesn't matter so much, if you're off by a few percent it's of no real consequence. It is still significantly better than using voltage.

The importance of a periodic sync-at-full event is why we're always talking about the parameters on the BMV that influence that, namely setting 2 (Charged Voltage), setting 3 (tail current), and setting 4 (charged detection time). If the voltage is above setting 2 and the current is below setting 3 for a time period as per setting 4, the BMV resets.

So what causes premature resets? Essentially false positives on the tail current. On days with low sun, or in conditions where the available PV balances out with the loads so that the current going into the battery is below the tail current. The battery is usually at a fairly good state of charge already, sometimes 85% or more, which we know because the voltage also manages to rise above setting 2 even with such low current, but it isn't full yet. If this condition persists long enough, you get a premature reset.

The ways to get around it is:

1. Increase the charged detection voltage. The higher it is, the less likely it is that you'll find yourself up there at lower states of charge.

2. Decrease the tail current setting. Again, that makes false positives less likely.

3. Increase the charged detection time. The logic behind this is that the conditions that produce false positives usually don't last more than a few minutes.

The setup that gives the least false positives is where you do all three. For a system that is cycled daily that means setting the charged detection voltage to 0.3V less than the absorption voltage, the tail current to 1% of the capacity, and the detection time to 15 minutes. The chances of having a PV/load balance that produces such a high voltage at only 1% current for 15 minutes is vanishingly small and when it does occur would almost certainly happen above 95%.

The trouble with your equipment is it doesn't hold absorption for long enough. So you have to decrease the charged voltage, and that significantly increases the probability of a premature reset. So your only option is to set the charged detection time to something very long.

That's why I've said to @Coulomb before that even the extended time-to-float in the firmware is still too short for lead acid. @pilotfish recommends that people raise their float voltages on systems that are cycled daily to get better absorption. I tend to agree with that, and will then recommend that you use my three-step recipe, and set the charged detection voltage to 0.3V below whatever you picked as float.

Share this post


Link to post
Share on other sites

Thank you for your reply @Coulomb. I feel a lot more confidant to run one inverter while the faulty one is being repaired. @Mike installed an 8Kw breaker in my output circuit and that has only tripped once as I keep an eye on the ICC display and run the system time based so that I am on grid when demand is high such as evening meal being cooked. All your answers to my questions should be valuable to all Axpert users and also addresses the matters raised by @OomD ,who started this thread. With  a lot of luck Voltronics may even read all about your posts on the shortcomings in their firmware and purchase the changes from you and @weber. We should be so lucky but one can only hope.

Thank you also @plonkster for the detailed explanation you posted above. At last I have a decent understanding of the way the BMV works. Since the bulk charge voltage of my T105's is 59.3V (no absorption voltage is given on the data sheet) and the Axpert maximum charge voltage is 58.4V, would it not be preferable to set the charged voltage on the BMV to 58.4V-0.3V=58.1V and the tail current and detection time  to 1% and 15 minutes as per your suggestions. The user manual for Rolls batteries recommends the following for flooded lead acid batteries using a smart 3 phase charger.

As stated previously, the Bulk charge phase brings the bank to approximately
80% state of charge. The remaining 20% charge is a function of time and current.
The charger will maintain current level until the Bulk set point has been reached.
The charger will then switch to the programmed Absorption voltage and timer. As the
battery charge nears completion, the internal resistance in the battery increases and
charge current begins to decrease. It is assumed that over the time of the Absorption
charge that 50% of your maximum charge current will be available (this is factored
into the equation). 0.42 = (20%/50%) + 5%. 5% is added due to losses.
ABSORPTION CHARGE TIME
Where : T = 0.42 x C /I
T = ABSORPTION CHARGE TIME
C = 20 hr RATED CAPACITY (total battery bank)
I = Charging Current (Amps) (recommended 10% to 20% of C20 discharge rate)
0.42 = ( 20%/50%) + 5% (5% is added due to losses)

My bank is 450AH and I charge mostly with solar but have set parameter 17 (charge priority) on the Axpert with Aussie firmware to SOL which is solar first. I am trying to minimize or avoid charging from grid but when no solar is available, mostly during nighttime,  would like the grid to top up the batteries to 100% charge. The system run on a time schedule to achieve this and not state of charge. I also try to avoid discharge below 80% SOC by adjusting the schedule depending on the season and manually switch to grid on cloudy days with insufficient solar generation. 

Using the Rolls formula above and assuming C10 current, I would have to set parameter 32 (bulk charge time) to 2.1 hours which would keep the bulk/absorb voltage at 58.4 for that time. Mostly I don't see 45A charge current as the house runs off solar during the day and my 2300W panels has their limitations. I am really just trying to see how to maximize the life of my batteries as I guess we all do. Perhaps I should read up on @Coulomb's postings on how the Aussie firmware manages battery charging to get a better understanding of the optimal settings. But any views on the issue will be much appreciated.

Share this post


Link to post
Share on other sites
6 minutes ago, ebrsa said:

would it not be preferable to set the charged voltage on the BMV to 58.4V-0.3V=58.1V

Yes. 0.3V under the value you configured on the inverter.

Share this post


Link to post
Share on other sites
Posted (edited)

Thank you. Just set it to 58.1V, tail current to 1% and detection time to 15 minutes. At present I do not have the grid connected to the inverters and am charging the batteries, which are at 95.3% SOC from solar but current is down to about 1.3A with voltage at 58.5V according to ICC.

Edited by ebrsa

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×