I have a MP2 in ESS mode, connected to a 16cell/16kWh LiFePO4 battery.
Absorption voltage is set to 56.8V and Float to 54.0V.
Yesterday, very sunny day, battery was being charged (solar is connected via ACin).
At SoC 95%-100%, I got a high voltage alarms. I found below Battery DC voltage, with a considerable spike towards 60V, just before absorption voltage was reached. Spike started at 56.6V and peaked immediately to 60V.
Is this normal behavior? Or can anyone help to explain what is happening?
My (wrong?) assumption is that MP2 will not charge above absorption voltage.
If this is normal behavior, what would be sensible settings for high voltage alarm?
Probably a battery disconnection decided by the BMS because one of the cells went beyond the programmed protect voltage.
Therefore, the output of the MP2, remaining open circuit, spiked.
Also: Use the ālimit managed battery charge voltageā in the settings>System setup>Charge control menu to limit the battery voltage to a slightly lower voltage to avoid disconnections.
Drop the absorption to 56V (3.5V) or even 55.68V (3.48V) and see if itās happening again.
Even 3.48V is enough for a LiFePO4 cell to reach a full 100%.
Thanks, will give it a try and at same time I am also looking at the JK-BMS settings; if these are more strict that Victron side, I will get the unexpected shutdown by BMS.
One nagging questionā¦we are here specifying e.g. absorption to 0.01V accuracyā¦but I do see already a 0.02V difference between MP2 voltage measurement and Lync measurement (if I normalize it to per cellā¦the total difference at accu level is 0.3V).
Which one can I trust most?
The JK-BMS was also off with 0.02V per cell, compared to Lynxā¦.but I calibrated that one to Lynx readingā¦.but again, not sure which one to trust
To the mV, for the sake of conformity, the way engineers doā¦
I, for one, I am measuring the voltage when current is zero (or near zero), i.e. idling, because could be cable loses when under load.
And, indeed, between batteryās BMS and Victron inverter is a difference of about 20mV, which is more than negligible, 0.04%.
Who to trust?.. Neither. Just your trusty, calibrated, multimeterā¦
Youāre right about DVCCā¦
Butā¦
I have a VTAC battery connected to a MP2 on one of my systems on which Iāve also decided to not use DVCCā¦
Why?
Because even if I follow the max cell voltage and try to reduce the charging voltage to prevent cellsā overvoltages, the MP2 will execute the command and adjust the voltage only after 20 to 30 seconds.
Tooooo long and sometimes inefficient, because the max cell voltage will overshot.
So, up until they will speed up the process, in both Venus scripts - where the update is at 3 seconds - and inside firmware - where they use a too slow voltage filter - the DVCC will stay off, the voltages and current set conservatively and let the BMS do the job.
Victron donāt recommend that BMS systems update the DVCC parameters more frequently than 20s.
There is something wrong in your system OR BMS design if you batteries can overshoot the max voltage in 20s.
Either your charge current to battery capacity ratio is too high, or the BMS is not starting to curtail the charge current until the cells reach max voltage.
I find that if the current is ramped down over the last 100mV, then the system works fine.
Battery 100Ah/51.2V
CCL, on last 100mV, is set by the BMS at C/5.
Still, I have limited the current to 15A for all charging period.
So 15% of battery capacity I donāt think itās a too high currentā¦
Example for overshotā¦
Voltage: 55.4V - not so high, right? You would say 3.46V per cellā¦
But itās comprised by 1 cell at 3.65V and the rest 15 cells at 3.45Vā¦ Overvoltageā¦
The CCL is not recommended for control. Victron recommend using CVL. The Multiplus is not too good at following CCL, but not too bad at following CVL.
The 100mV I use is PER CELL not on the battery voltage as a whole. So as the highest voltage cell approaches 3.45V the charge is started to be restricted, and by 3.55V the battery charge is zero. So if the battery is out of balance, itās the highest voltage cell that controls the charge, together with the balancing current. No overshoot - ever.
I compute the CVL by using the calculated CCL and the estimated battery impedance. Gets a little complicated.
Still, I have limited the current to 15A for all charging period.
You need to restrict the charge current to a value less than the balancing current once the highest cell voltage gets to 3.55V. This can be anywhere from 50mA to 3A depending on your balancer. If you donāt control this, then you will get overshoot on an out of balance battery.
Do you have some reference for this?
Because exactly this is the problemā¦ The charger doesnāt respond quickly than thisā¦ And 20s is a long time.
For example the same period is considered when injecting to grid and 20s is too much for a proper power regulation.
Quite a few complains here on this forum for this slow actingā¦
Yes I do have the reference for this, but it is a controlled document and I cannot share this.
If you do the control on the cell voltage, then 20s is ok. My own system where I charge from MPPTās I do use the CCL, and the refresh is about 3.5Seconds. This is not using Venus code.
For example the same period is considered when injecting to grid and 20s is too much for a proper power regulation.
Quite a few complains here on this forum for this slow actingā¦
Ok, but most grid codes require a slow rate of change of power, to aid in grid stability. Personally I think that they get some of this wrong, because it can leave the grid subject to sudden loads turning on or off, which could be more easily compensated for by inverter/battery systems rather than the gridās inertia. Some Grid standards require something like 5 - 6 minutes for 0 - 100% power change.
So itās not really Victronās fault that the system is slow acting, this rate of change of power is subject to international standards, and Victron need to comply with this for certification.
During my tests, back on June, this is what Iāve obtained, which was pretty OK, but again, later Iāve decided against, because the below behavior was after some tweaks on the MP2 firmware.
As there is not feasible to use modified firmware for others, Iāve dropped this and used it without DVCC.
āneeds to very stable and slow to prevent oscillations. It is probably not helpful to update the parameters more than once every 20 seconds, but do test intensively to determine your own timing.ā
most of the problem is that the BMS that pylontech use only has 50mA of balancing current for a fairly large battery (100Ah). This means that in solar charging the BMS can only balance out ~200mAh of charge difference (0.2%). In my opinion, Balancing current needs to be at least 1%, up to 5% for Solar charged systems.
Also: NOT basing the CVL on the highest single cell WILL lead to oscillations and high individual cell voltage alarms, particularly if balancing currents are low.
Iāve spent 5 years tweaking my BMS software and hardware design, and it is now stable. Iām still thinking that using PID control on the CVL may help speed up the charge - but only slightly.
Solar charged systems may get from 2 - 6 hours of charging time, and the BATTERY MUST balance within this time in order to get to full charge.
Iāve talked about 2 batteries here, first VTAC and later Iāve pointed out to the actual implemented algo inside DVCC for preventing high voltage alarms on Pylontech, where oscillations are present.
Although both BMSes, VTAC and Pylontech have passive balancers with small currents like youāve said.
Anyhow, a lot of room for improvements, too bad that, like someone pointed out the other days, Victron has a very small engineering team and the world moves a bit faster than Victron can move.
Still, good hardware and firmware design on later devices, especially RS series.
