I have a DYI 16s bank that has been running now for about 2.5 years. Fully understanding what I was getting into, I am monitoring the bank, through Grafana, almost by the second.
Last couple of weeks I will get my Victron system crying that the battery is low and even had two or three times where the system even shut down and then startup again.
Looking at the SOC it was around 50 - 60% so it was not like bank was under pressure and it is also it was not under load .
Now looking at the data this is where I get confused.
Comparing the BMV vs the BMS bank voltage there has alwys been a diference , I am assuming it is as a result of the point of measurement.
The blue line is where the system shut down so there is no noticeable deviation
If you compare the highest cell vs the lowest cell in the bank (delta) you can see that the there were a small deviation but nothing to get excited about. |
There was a bigger difference when I boiled the kettle.
But the rest of the data does not support the Midpoint deviation.
I don’t have an idea how to approach this, but it is annoying having the system restarting and then getting 5 emails from VRM that the battery is low.
Any thoughts?
The difference is probably calibration. I would not worry about that. The sudden drop in voltage and appearance of a delta in the midpoint monitoring says bad connection or bad cell. Bad connection should create a midpoint deviation proportional to the load, so this feels more like bad cell.
Victron inverter is probably complaining because of low voltage. What VE.Bus alarms show up on VRM?
Question, would the cell be consistent or can it be intermittent?
In other words, works for weeks then one day an error. Works again for weeks with no issue.
No big draws either at the time.
Methinks, a suspicion, that the BMS is overloaded trying to supply the data via the RS485 port.
Hence my question.
If the BMS switches off, the BMV shunt is after the BMS, BMV volts will drop off a cliff, inverter will also switch off.
If the BMV shunt is before the BMS, then the BMV will not have an issue, although the inverter will still switch off.
Apologies , I was out of town for two day’s.
To answer question 1 ,
The BMV is AFTER the BMS. I decided in the beginning that I would prefer the BMS to have the least amount of “disturbance”.
I also expected that there will be a difference in the voltages of the two devices but they should but fairly linear as the point of measurement just changed.
I will have to recheck the connections also I don’t recall ever calibrating the BMV since I bought it waayyyy back. So definitely look into that.
Considering the bad cell , there could be a chance but it is highly unlikely , my delta would then be all over the place and so far it has been pretty constant.
Maximum delta over 30 day was only 46mv.
No no no, BMV must be first in the “queue”, the very first device after the negative pole.
Then the BMS.
For accuracy too. Now you are not measuring the bank ah in/out, you are measuring that with the BMV’s draw in between.
Also, methought, BMS SOC is iffie, being volts based. BMV SOC more “stable” for the system, also why I put the BMV first.
And then the fact that the BMS’es don’t have shunts that are very accurate compared to the BMV. My BMS shows zero watts going into the batt, the BMS shows at times up to <100w still going in, at 100% SOC.
If the two SOC’s are just about spot on, and mine is month after month, I KNOW all is fine.
And IF they differ big time I also KNOW there is an issue with the bank/cells.
Learned that lesson in a very expensive way.
And I vaguely recall I read it on one of the Victron docs/saw it in a video, how to connect the BMV and where.
The data you posted on the BMV, was the BMS switching off the BMV.
Bad cells. Knew you had copious data on each. That you control the charging based on the Delta/cell shooting out.
It would be consistent. The way he describes it sounds like it might fit. The issue seems to appear around 60% SOC, or thereabouts. The midpoint suddenly deviates by about 2 volts, and the delta becomes 1.4V. This is very typical of one cell that has degraded (so it now has half of its original capacity left), so the battery behaves normally (due to the flat voltage curve of LFP cells) until one of the cells is “empty”, at which point it drops out, skewing everything else.
Anyway, since it is a DIY pack, it is fairly simple to check. Discharge… stand ready with a multimeter. Wiggle connections to make sure it is not that, and I am sure you will find your culprit.
Paul can see each cell’s volts, as he records it all. He would have seen a cell drop off.
He has gone all out ito getting BMV data into graphs.
I say it is the BMS that had an issue (it was replaced since), that switched off resulting in the BMV also having a recorded drop in volts as indicated. 100% expected.
If it walks like a duck, quacks like a duck come to mind.
If his BMS was before the BMV, the BMV would not have had recorded anything other than normal state of affairs, and he would have known for a fact it was the BMS, the inverter going off.
Why? I speculate it is the BMS software being “overloaded” with all the constant data pulling. I wonder how a developer would have pulled the data, to mitigate traffic?
I now realise I don’t actually know what your point of reference is. For me, “Before” would mean “closer to the battery”. Anyway, I’m too lazy to analyse this to death. BMVs generally don’t just fail randomly, it is extremely rare, and since it is a DIY Bank, I still say do the easiest and cheapest thing first: Get out the DMM and get a second opinion. I mean, if it was a commercial battery in a case that is hard to open, voids warranty, and all that nonsense, sure, then I’d waste time coming up with scenarios that don’t require opening the box
Battery Negative pole —> BMS —> BMV - when BMS switches off, then the BMV is also off.
vs
Battery Negative pole —> BMV —> BMS - then BMV will never switch off unless it is done on purpose.
Well I never see a zero measurement, so I am somewhat puzzled about why we’re barking up this tree. So I tend to just fall back to the question asked: Could it be the BMV? Sure… anything is possible, but since it is not something I’ve seen often (I’ve never seen such a thing), it strikes me as the kind of hope you might cling on to because the alternative is too terrible to comprehend (the battery really has a problem!). And if it is then easy to check said battery… why would you not do that?
You don’t see the massive drop at around 2:25 of the lowest cell voltage? He said there wasn’t particularly high loads at the time. At 60% SOC, I would not expect the voltage to drop that low.
But I see the highest cell is also dipping, and that argues against me.
Anyway, it seems we are at least agreed: There is probably a reason why the BMV came up with that weird value… it isn’t wrong, it actually measured it.
Apologies for missing the robust discussion, was on the road for most of the day but hey it pays for the Solar Habit .
Looking at the discussion I first disagreed with the BMV location but considering everything it does make sense.
So as the facts stand, I have a system that switched off and on by itself and apart from the BMV showing a single mid voltage spike (one data point) no other reasons as to why it happened and since then everything is running smoothly without issues.
I would not be able to confirm if the BMS went and switched off and on again, but thinking about it, it could explain the BMV complaining.
If I move the BMV so that it is always powered I will pick up the error. I am also leaning towards the BMS that might not be too happy in being polled for data all the time but there is no evidence other than having this “new” problem.
Also need to check all the connections, just good maintenance any way.
I don’t think that it is a bad cell. I would have picked up a problem every day as the bank gets pulled down to 50% daily and sometimes as low as 40% without problem on the delta anywhere.
So first step is going to be to move the BMV shunt connection so that I can confirm who switches what off.
I have also started looking at Arduino Voltage monitor. The standard ADC give reasonable resolution, but I might get some 16bit ADC’s. Then I can pull the cell voltages independently without any problems. Just not when I will have time to build it :).
