AC Coupled PV High Charging Voltage

DIY
, , , , ,
1

Hi,

I’m having trouble charging my battery from 99% to 100% because my Quattro’s aren’t maintaining my requested absorption voltage when my Enphase PV microinverters are running (I’m using the PV assistant). Basically the battery voltage keeps rising until I hit a maximum cell voltage within the battery (the internal balancer can’t keep up) and the BMS then disables charging and then my Quattro freaks out with High DC Voltage alarms (the noises for which sound very not good for my victrons…).

My understanding is that the Quattro should start raising the output frequency on AC-Out once the battery hits 55.2v, thus lowering the Amps of AC from the microinverters, and allowing the battery to absorb the dwingling amps and balance itself. This isn’t happening. For example today I saw 56.81v while monitoring VRM and the output frequency was still 60.1Hz when in my understanding the Quattro’s should be telling the microinverters to slow down at least 1v earlier (my grid profile for the microinverters will shut off at 61.2Hz and theoretically limit after 60.1Hz). If I switch off the PV Inverters (via the circuit breaker) and turn on AC-In via circuit breaker (I have a 240V 20A connection to the grid for emergency charging) then it charges perfectly at 55.2V exactly as I would expect.

What am I misunderstanding in the situation?

My system details below:

I’ve got 2 Quattro 10k’s in split phase (USA) off grid setup. I’ve got a mix of DC and AC charging. DC with Smart Solar MPPT and AC with Enphase IQ8+ on AC Out of the Quattro’s. The battery system is DIY with a JK BMS and 48v LiFePO4 battery setup. Everything is connected with a rasbpi running VenusOS.

I’m using DVCC and BMS controlled charging so the requested charging voltage from the BMS is 55.2v (3.45v per cell) for 1 hour and then float at 53.6 (3.35v per cell). Now I’m not sure if the Quattro Charger settings from VE.Configure adhere to the BMS/DVCC settings but I also have that set to 55.2 for absorption.

2

I found 55.2V was too low for mine to balance correctly. There didn’t seem to be enough energy in the cells to balance properly. I run mine at 56.5V and I dont get any overvoltage cells and I only have the 0.6A balancer, not the 2A. I did do a full top balance at 3.65V when I first built the battery as well to see if there were any problem cells.

You can try this Node-RED flow. This flow adjusts DVCC when the voltage rises. I have a function to scale the charge current when the voltage starts to rise. I find my battery is 99% full at 54.4V. You can adjust it how you like. I use this on my 12V system, which allows it to balance and charge nicely. The JSON code is in the text file.

My battery is full and balancing. At 55.2V I found cells would hit 3.6V but now I barely get a disconnect because the cells that run away do it much earlier in the charge cycle and brought down by the balancer. At 55.2V there would be many that are near 3.55V and hitting 3.6V. The balancer can only pull down one runaway cell at time.

image
image750×1333 214 KB

ADJUST DVCC ON VOLTAGE.txt (5.2 KB)

3

Hi, Thank you for your response. I would like to limit my use of NodeRED although I do have some flows. I have a concern about the solution you shared, namely that I believe the PV Inverters do not seem to listen to DVCC charge current limits (based on previous experience) probably for the same reason they do not slow down when the battery absorption voltage is reached (either in VE.Configure settings or in DVCC). Are you using DC or AC charging? If AC, is it connected to AC-In or AC-Out and what are the inverters?

Considering the charge algorithm works perfectly when the grid is connected to AC-In, and my batteries do balance just fine at 3.45/55.2V (the issue is that the charger ramps the battery voltage up to 57V+ which makes it very hard to balance when as you say the battery is basically 99.9% full at that point and the energy to go from 3.5 or 3.55V to 3.65V is miniscule) I want to have the Quattro’s slow down the PV Microinverters and charge closer to 55.2V (not sure if the PV can ramp down enough hold that without shutting off but that’s a different problem).

Does anyone know why this could be a problem? Why does the Quattro not increase frequency and reduce charging in my scenario?

I’m looking at my PV Inverter Assistant and the settings are:
Installed on L1 (the only assistant currrently) with these settings:
VE.Bus BMS: No
Restart PV Inverters: 52.8
Inverter Start Reducing: 60.1Hz
Inverter @ Minimum: 61.1Hz
Inverter Disco: 61.2Hz
Total Installed PV Power: 3000
Total Installed Panel Power: 3000

4

My PV solar system is grid-tied, and my MPPT solar panels are east-facing. The setup was done by a highly experienced Victron installer, who also helped me understand the system. I’ve managed to make some adjustments using Node-Red. It’s unclear whether you are running ESS or not. If you have exports enabled, then the DVCC feature is ignored, and you will receive higher charge currents and voltages.

I utilise both DC charging and grid export in the morning if the battery is too full. I aim to have the battery fully charged by around 3-4 PM and usually perform the absorption phase for about an hour. There have been times when I haven’t used enough energy from the battery overnight, causing it to be full by lunchtime, but I haven’t encountered any issues with this.

Are you grid-connected? It’s important to note that you can’t shift the frequency unless you are in island mode. The output frequency will always match the incoming frequency, and the charge voltage will be slightly higher to accommodate exporting.

For a cleaner setup, consider connecting the Enphase inverters to the input rather than the output. If you were using a Fronius inverter, you could take advantage of its zero feed-in capability.

You might also want to try setting the charge voltage to 54.2V to account for the increase due to excess PV output.

5

Hi, Thanks again for your reply and more details about your setup. Our setups are quite different. My system is not grid-tied, no ESS. I’m using the PV Inverter Assistant. Frequency shifting occurs but as far as I can see this only happens when the Quattros experience an error (or charging is disabled, etc). I’m not aware of any benefit in connecting to AC-In instead of AC-Out for my system but if someone can explain how that would affect the problem I would be curious. As I mentioned before, although I’m not grid-tied, I do use AC-In to charge in an emergency (from the grid) but using ESS and being in parallel with the grid is not allowed in my use case.

I guess I can try changing the charge voltage lower to rule out all things but as the system makes no attempt to frequency shift over the 2V from 55.2 to 57.x, I would be shocked if this worked (and annoyed!).

FYI Victron supports SunSpec in VenusOS 3.60 and above so zero feed in is theoretically supported in many products although I’m not sure which are confirmed in the USA.

6

You could probably optimise this with careful configuration. DC-DC charging is excellent for battery charging, while AC-coupled PV is ideal for supplying AC loads directly.

My setup uses east and west-facing arrays connected to MPPT DC-DC chargers, so the batteries get charged in the morning and receive additional charging in the afternoon when needed. My AC PV inverters face north, so during midday I use that power directly for AC loads like hot water and air conditioning in summer and other ESS that use their AC-DC chargers. Any excess goes to the grid. You might be able to do something similar by strategically timing your AC loads to match the AC PV production.

Let me explain the issue I had with 55.2V as the charge voltage. I have a screenshot of the battery data showing 10-12 cells in the 3.45V range. What used to happen with the lower voltage was that 3 or 4 of these cells would exceed 3.45V simultaneously and quickly hit 3.6V, which would disable charging. However, this battery still had at least an hour of charging remaining at that current level.

With the higher voltage, there’s enough energy to push cells above 3.45V, and the balancer pulls these higher cell voltages down while bringing up the lower cells, and this balancing happens much faster. I know it seems counterintuitive, but the slightly higher voltage improved the balancing process

IMG_3965
IMG_3965750×1334 151 KB

7

In my understanding the assistant should be running all devices the same as ess assistant would be.

You could probably adjust your OVP protection on the jk bms to tame its sensitivity.
The other way is to adjust the grid charge amps on the inverters so they can’t charge the bank as fast from AC.

Works with the ess assistant and if the ac pv has comms and modbus control with the GX. It has worked with fronius for ages.
Enphase is still in testing.

8

Sorry, yes it is also installed on the L2 device (but requires no settings to be input)

I’ve already pushed the BMS to its spec limits hoping this would give the Quattro room enough to adjust but it seems the issue is that the Quattro is not doing any frequency shifting. Or??

Then I would run into over-charge protection (and generally lose power) when I’m charging before this – and to my understanding adjusting the charge limit is not available within Node-RED. Again, the Quattro should do this natively, correctly(ish), as it does with AC-In from the grid? There must be people still charging their battery with PV AC and the PV Inverter Assistant

Indeed there is a guide for setting up enphase with zero feed in wit european IQ microinverters but when I asked Enphase to enable modbus on my IQ Gateway they just told me Victron is not supported in the US. Anyways zero feed not working is fine – I’m off grid.

9

Correct. I have some set ups like this. For some batteries with slow balancing we increas absorption time but have lower charger amps set in the quattros as well for the bulk so there is no overshoot.

In the battery monitor on the inverters what is your bulk finish percentage? Sometimes adjusting that helps.

10

There’s a bit of voodoo there :smiley: But reality sets the way forward… Totally understand using the settings that are working best in your system. I’ve also charged at higher voltages, especially when I wanted to be sure they were top balanced, and was charging at 3.5V per cell recently (before I backed off because of this issue) but in normal usage I don’t care that they are perfectly balanced or 100% full so raising the voltage any higher than is necessary is something I avoid. Who knows how much it will affect the lifetime of the cells, maybe insignificantly or negatively! But I generally follow what the off grid garage guy / Andy has determined in his testing and research. I just want the system to charge correctly!

Yeah that’s cool and sounds well designed. I also have a lot of East-Northeast (I’m in Northern Hemi!) panels (AC) and they perform really well despite the orientation, and then I have a bunch of South facing DC. It’s NOT optimal but I have serious shading issues so microinverters are the best option even if I over-produce for the load and lose plenty of efficiency. But yeah it’s cool to try and get as much efficiency as possible.

11

95%

It seems I would have to significantly lower the charge amps for it to absorb correctly considering I see no frequency shifting. Again, I’m not seeing any attempt to slow down the PV AC once absorption voltage is reached – to me that seems like the root issue. What other good tests are there to see if the Quattros are doing any limiting? I could set the charge amps to half of what I expect and see if it maintains 60.7Hz or whatever. I could turn on the prioritize renewable energy setting and see if the charger will hold 52V constant voltage instead of constant current charging.

*I do see frequency shifting (to the disconnect frequency) when there is an error from the BMS or Charging is set to off on the inverters

This raises another question I’m not certain of. Doesn’t DVCC take precedence as it does in the MPPTs for everything other than max charging current for the device? My BMS defines my absorption time and when charging with AC-In + Grid this works perfectly. Obviously I have some false assumption about something here

12

Honestly if you are using the can, the jk bms doesn’t work great its too simplistic.

13

All you need to do is stop the cells from hitting the high voltage disconnect. The problem I had with 3.45V per cell was that some cells were still at 3.3V, which allowed the cells at a higher voltage to creep towards 3.6V and it was not one but 3-4 at the same time. I would have 200-300mV deviation when they were near fully charged. Now I only get 60-30mv. There are many factors that go into charging a large battery. 3.52V per cell is only 14.1 volts when compared to a 12V battery, and I charge my 12V system to 14.2V and my 24V system to 28.4V. Everything is working well, and some of my batteries are over 4 years old. My 12V system does a cycle a day.

If you are getting high cell voltages, then you need to figure out what to change. I dont know if it’s worth looking at something like the Tigo TS4-A-O and their tools to disconnect solar. You could use the relays on quatro to shut some panels down.

14

I’m not sure what you mean – it works perfectly outside of this specific issue I’m having (i.e., with MPPT and Grid Charger). What feature is it missing for correctly charging lithium iron phosphate batteries compared to the Victron features? Honestly a bit of a non-critical side topic because charging is only not working correctly with PV AC so my question remains what a good test here is to troubleshoot the frequency shifting. I guess I could set a really low absorption voltage and see if it maintains that

15

This is the job of the Victron charging ecosystem – PV Inverter Assistant, DVCC, etc. The BMS is sending the correct information – the Quattros are not doing anything with it (probably because I’m doing something wrong!). I cannot stop a high voltage disconnect if I’m getting unlimited amps pumped into my battery – that’s a guaranteed outcome. The Quattros should be saying “the battery voltage is over 55.2V, slow down!!”, i.e., raise the AC frequency. Instead the Quattros only respond to BMS alarms at which point the BMS is not accepting charging.

16

Yeah, I understand. When I commissioned a Victron installer to take me off-grid, he said stay on there and do not AC couple, you will be better off with a 450/100 and leave the 5000w PV arrays exporting to the grid and use the AC charger if you need it. AC-coupling is inherently difficult, and there are many who have the same problem as you. I was lucky.

Did you look at the Tigo optimisers and see if they are something that you can use to disconnect some panels?

17

I have tigo for the DC array’s (for emergency disco) but as the MPPT correctly modulate voltage I don’t need to use them. I think I would buy another battery and keep my system from ever reaching 100% outside of a controlled charge than get on the roof and touch the panels with microinverters! Haha

So is this a known issue? Like frequency shifting takes xyz minutes to start or anything documented (that doesn’t make sense to me since it is immediate when there is a BMS Error so it should be immediate when there is a specific BMS voltage).

Either way I will do some more testing and see what I can figure out. Just a lot of time and effort and was hoping someone else already experienced this

18

I am not sure. It’s at the hardware level, it may be due to reference voltage, as the MultiPlus has no idea the battery is full and maybe has no idea of the voltage. These details are all locked away. You really don’t want the frequency of the power you use to be changed, as it messes with all sorts of connected devices that rely on a stable clock fixed by frequency.

You can never have enough battery. I am sure if you want to charge to something like 54.4V and have 4-6 hours to absorb, you will be able to reach 100%. But you will still need to use that energy during the night.

19

From what i have seen, usually it happens at the end of bulk charge, so can be around the when bulk finishes percentile set for some, others, when the cells stop taking a large amperage (which is more difficult with lithium hence the need fo current and voltage control). When set up at lower voltages there seems to more of an issue, running the cells slightly higher as mentioned by Owen seems to work better.

With the JKbms, they don’t do both voltage and current control. Personally i have found modded drivers work better rather than their can control (from 2017).

20

After doing further testing, I have even less confidence that I understand what is happening with AC Coupled charging here but I have a guess that something in my udnerstanding / usage of the grid profile is wrong? I ran 2 different absorption voltage tests, described below and with VRM screenshots at the end.

First test: I set the absorption voltage very low on the BMS (53.2V), I wanted to see if there was a delay / response from the PV AC charging that I wasn’t seeing when the battery was full because the BMS was shutting off.

Result: immediately upon hitting this voltage, the PV AC power dropped to it’s minimum (I guess) of about 120W (that’s about 8% of what it was previously doing). DC Charging seemed to mostly stay at its maximum. This happened 3 times in succession because there wasn’t enough PV DC power to hold 53.2V and as soon as the PV AC ramped back up (5 min delay) then the battery would get to 53.2V and PV AC would ramp down again. This seems the safest charging behavior! Great!

BUT, during this time period I do not see any frequency shifting recorded in VRM (1 min logging interval). Strange. Perhaps it’s so quick that it does not record the shift?

Second test: I then set the absorption voltage to 54.4 (3.4V per cell) which is considerably lower than I would like but I already know that my higher absorption voltages are not working.

Result: AC PV ramps down 3 times, with the 3rd being overvoltage BMS shutdown.

First ramp down: 54.41V in VRM, PV AC shuts down completely (no frequency shift recorded)

Second ramp down: 54.47V in VRM, frequency shifting begins to 61.2Hz (maximum before disco) but no PV ramp down until 6 minutes later when it drops to 0. Battery now at 54.98V

Third ramp down: Within 3 minutes of PV AC ramping back up, battery hits 57.22V, frequency shiftinf in VRM at 55.31V, BMS overvoltage disconnect, frequency shifting stays at 61.2Hz until battery voltage drops to to overvoltage protection recovery voltage.

Thoughts: Obviously the battery is getting super full at 54.4V even, as the voltage spikes super fast by the third event. Which means I want my PV AC charging to a minimum at higher voltages. However, I don’t see the charging logic working here. Even though I dont’ see frequency shifting in my first test at 53.2V, it does seem to be happening because the PV AC ramps way down (immediately!) at exactly that voltage 3 times. However, later I see that PV AC is running full blast even when the VRM is recording frequencies of 61.2Hz (but in the 2nd event it does eventually ramp down to 0 before a BMS disco error forces it to; in the 3rd event the BMS disco comes error causes ramp down to 0). Another difference I can’t square is why the PV AC drops to 8% power in the 53.2V test but 0% power in the 54.4V test.

My working theory is that something in the frequency shifting profile of my microinverters could be the cause? Maybe Victron frequency shifted to 60.2Hz (briefly) in the 53.2V test (not recorded) and the PV AC slowed down to 8% but in the 54.4V test maybe it shifted to 61.2Hz and the grid profile interprets this as “run full blast” because it’s out of scope? I’m using ISO-NE 2018 and from the data sheets I was looking at, it seemed it should have between 60.1 and 61.2Hz to limit power production and then shut off at 61.3Hz (or maybe at 61.2Hz).

Here’s 54.4V absorption

Screenshot 2025-08-04 at 11.22.31 AM
Screenshot 2025-08-04 at 11.22.31 AM2162×1360 251 KB

Here’s 53.2V absorption

Screenshot 2025-08-04 at 11.22.51 AM
Screenshot 2025-08-04 at 11.22.51 AM2132×1334 243 KB