YamBMS JK-BMS-CAN with new Cut-Off Charging Logic (open-source)

[ChrisG]

Just wanted to report things are working great. No issues with multi-bms-ble. Multiple charge/discharge cycles (on purpose).

 

[arzaman]

In the case of an inverter with a low load (when you are away from home for several days for example), the battery voltage drops slowly and it takes a few days for the Rebulk V. condition to be true. This new criteria based on SoC makes it possible to start a new Bulk phase earlier otherwise it is possible that it will only start at an SoC of 75%.

I've been experiencing this situation recently: long sunny days with high PV production and days away from home with low consumption. As a result, the charging status remains in "FLOAT" for an extended period. The "rebulk" SoC condition could help by forcing the system back into bulk charging, thereby aiding in battery balancing.

I've now set the rebulk voltage to 52.8V. This level has been triggered only a few times in July.

 

[MrPablo]

@MrPablo @Der_Hannes @shvm @arzaman and all

What do you think of a rebulk based on two criteria :

1. Rebulk V.
2. Rebulk SoC

If one of these 2 conditions is true then a new Bulk phase can begin.

In the case of an inverter with a low load (when you are away from home for several days for example), the battery voltage drops slowly and it takes a few days for the Rebulk V. condition to be true. This new criteria based on SoC makes it possible to start a new Bulk phase earlier otherwise it is possible that it will only start at an SoC of 75%.

There's several ways we can look at triggering rebulk:

1. Voltage - can lead to long-term float and resulting drop in SOC over time
2. SOC - will resolve the above, but relies on accurate SOC
3. Time since last full charge - will resolve the issue with voltage, but won't work well in systems that rarely charge fully
4. Time since last bulk - will resolve the issue with voltage, should work regardless of SOC accuracy

SOC is likely the easiest to implement.

 

[Sleeper85]

There's several ways we can look at triggering rebulk:

1. Voltage - can lead to long-term float and resulting drop in SOC over time
2. SOC - will resolve the above, but relies on accurate SOC
3. Time since last full charge - will resolve the issue with voltage, but won't work well in systems that rarely charge fully
4. Time since last bulk - will resolve the issue with voltage, should work regardless of SOC accuracy

SOC is likely the easiest to implement.

The SoC can become inaccurate when there is very little time to fully recharge. That said, in Portugal, I was able to compare the SoC of a JK-BMS 8S against a Junctek Shunt and even without 100% recharging for several days the SoC of the JK-BMS was often equal to the shunt.

Here is the graph over a period of 1 month without 100% charging every day.



So I think that a well calibrated JK-BMS for correct measurement of a 5A current (this is my procedure) can be reliable.

Let's imagine a Rebulk with 85% SoC, what would be the problem if the real SoC is 80% or 90%? None from my point of view.

Let's not forget that in the future it will be possible to take Voltage, Current, Power and SoC information from a shunt. That would be the next step on my list.

So I add the Rebulk condition based on the SoC?

 

[silverstone]

The SoC can become inaccurate when there is very little time to fully recharge. That said, in Portugal, I was able to compare the SoC of a JK-BMS 8S against a Junctek Shunt and even without 100% recharging for several days the SoC of the JK-BMS was often equal to the shunt.

Here is the graph over a period of 1 month without 100% charging every day.

View attachment 230605

So I think that a well calibrated JK-BMS for correct measurement of a 5A current (this is my procedure) can be reliable.

Let's imagine a Rebulk with 85% SoC, what would be the problem if the real SoC is 80% or 90%? None from my point of view.

Let's not forget that in the future it will be possible to take Voltage, Current, Power and SoC information from a shunt. That would be the next step on my list.

So I add the Rebulk condition based on the SoC?

I don't think I'm directly affected by this (3 x Inverters Deye 12kW/inverter has a no-load Power Consumption of 300W-400W -> 0.3kW/0.4kW, my base/average load is somewhere in the range of 1.2-1.5kW [including Inverter losses], so in one Day I currently swing between 30-40% to 100% with 30kWh installed), but I still think it's a good idea .

However, my experience is also that a few months ago after a full charge, the maximum SOC of **ONE** Battery (with an older HW/SW Version of JK BMS, IIRC Version 10) would weirdly drift. First it would be 100%, a few days later 99%, then a few days later 98%, etc. I think for a Period it didn't go above 92% .... Then, not sure why, the last 1-2 Months it went up again to 99% or so.

The other Battery didn't have this Problem and always reached 100% [actually too early, because there was still some charging left to be done, HW/SW Version 11 for this one].

Just to say: don't put the Rebulk SOC "too high", otherwise they might never go to float. Of course your situation might be different, but this is my Experience. I wouldn't recommend to set it above 85% (worst case), probably even lower (75% or so). And I agree that while the SOC is not that accurate with the JK BMS built-in sensor, it's probably "good enough" for our purposes. I don't routinely go below 20% so I have some headroom there (plus a charger in parallel that "forces" a minimum Voltage of say 51 VDC).

Are there any "Risk" if it keeps switching between Rebulk-Float-Rebulk-... all the time there is a small transient Cloud above your PV Array ? My Situation is maybe a bit extreme but with those high delta voltage values it will take a few hours to rebalance the Pack I guess. But is it really neeed ?

@Sleeper85: are you trying to force Bulk to do a "rebalance" of the Cells, to make sure that they get 100% Charged from time to time, minimize the voltage/SOC swing (-> increase Lifespan) or mainly optimize your economic yield in terms of Power (if there is excess Power on the PV side, take the chance and "boost" charge the battery before it's "too late") ? If you are "away for a few Days", i.e. the load aerage/peak is lower, thus shouldn't the unbalance also be lower (assuming it's caused by internal losses / difference in ESR, then ESR * I^2 will also be lower when you are away), wouldn't it ?

Maybe all of them ... Just trying to understand better the reasoning there.


On a separate Note
I think I just threw away the idea of installing a shunt on these new Batteries I'm currently building. Would have been 2x Victron Shunt but not much Space (or Money) for it. And I'm not in the Position, unlike many of you with maybe only 1 Inverter, to just install one: I have 3 Inverters, 5+ Chargers, 4+ Batteries all on 2 DC Distribution Busbars. Unless of course doing some very crazy stuff with 8+ Cable lugs on top of each other or something like that (Bad Idea !).

I have a few DC Current Transducers (https://eu.mouser.com/ProductDetail/Tamura/L34S400D15?qs=EyUr%2BTmQJGcqRQyWuoPhyw==&countryCode=DE&currencyCode=EUR, assuming their Datasheet is correct max. Offset would be ~ 0.02V/4V * 400A ~ 2A). DC Current Transducers (+ ADC Chain etc), from Experience, have quite high DC offset though

Do you think something like that could be an Alternative to a Smart Shunt ? IMHO a Current Shunt will have much better (lower) Offset though ...

EDIT 1: saw @MrPablo Post, so you force to Rebulk in order to "reset" (recalibrate) the SOC Counter at 100% at the High Battery Voltage & Low Tail Current Condition (if I understood correctly).

 

[MrPablo]

The SoC can become inaccurate when there is very little time to fully recharge. That said, in Portugal, I was able to compare the SoC of a JK-BMS 8S against a Junctek Shunt and even without 100% recharging for several days the SoC of the JK-BMS was often equal to the shunt.

Here is the graph over a period of 1 month without 100% charging every day.

View attachment 230605

So I think that a well calibrated JK-BMS for correct measurement of a 5A current (this is my procedure) can be reliable.

Let's imagine a Rebulk with 85% SoC, what would be the problem if the real SoC is 80% or 90%? None from my point of view.

Let's not forget that in the future it will be possible to take Voltage, Current, Power and SoC information from a shunt. That would be the next step on my list.

So I add the Rebulk condition based on the SoC?

Yes, SOC based is by far the easiest to implement. I mention the alternatives for reference, but they all require additional logic and testing.
This is definitely a cost / benefit moment - it's a low risk problem so dev effort should be kept low.

 

[Sleeper85]

@silverstone



Here is the graph of version 1.17.3 with an inverter with low load (max 150W) while we were in Portugal. I just tell myself that it might be interesting to recharge more often and not leave the SoC below 80% which would be a conservative default value.

In the case of a large load on the inverter the Rebulk V. would probably be reached more quickly so these two conditions would be complementary.

I added it to the version being tested at home. This will be part of version 1.4.2.

 

[silverstone]

@silverstone

View attachment 230706

Here is the graph of version 1.17.3 with an inverter with low load (max 150W) while we were in Portugal. I just tell myself that it might be interesting to recharge more often and not leave the SoC below 80% which would be a conservative default value.

In the case of a large load on the inverter the Rebulk V. would probably be reached more quickly so these two conditions would be complementary.

I added it to the version being tested at home. This will be part of version 1.4.2.

Sure, it is a good idea as I also said . Just IMHO (especially if you have the same issues that I do) the SOC threshold shouldn't be set too low.

80% would be fine. If you do NOT have that weird SOC drift that I had, then 90% would also probably work. Keep at 80% for default though .