JK BMS SOC Drift - How to (not) calibrate the current sensor and will it actually help?

[@Shellyfi]

JK need to add calibration curve option. Set reference amp to different points

[@SkypowerwithKarl]

Yes this!

[@Drillmechanic]

This is such an expensive option that not all solar inverters have it, and you want an inexpensive BMS to have this calibration CURVE option!

[@svenmueller]

@@Drillmechanic it's not actually expensive. The BMS already has a microcontroller and software. And curve calibration can be implemented to great precision by a lookup table and linear interpolation (the best option would be polynomial fitting, but that needs more computational power). Heck, even adjusting slope and offset via just two measurements (high and low current) would improve what they have a lot.

[@DanBurgaud]

@@svenmueller Your idea is doable with the current hardware. Only need software. Just allow us to calibrate at multiple points and let the controller intrapolate/extrapolate the curb.

[@Drillmechanic]

@@svenmueller How BMS can control the charging-discharging curve(V-A*h) - only by some communication with charging invertors which have already precise SOC calculation components!

[@videostar555]

This is what I was saying in your last JK video. Great this is brought up.. I use JK in small cars and if the customer is not charging fully every time, the SOC will be high but voltage will be low.. In most cases the SoC remains at 75% and then to 0 because the voltage is already 3.2V.. The JBD BMS is clever because it has SoC for 100%, 80%, 60%, 40%, 20% and we can assign value for each.. it seems to work even though SoC doesn't match at times, as soon as we discharge, it callibrates itself to correct SoC after sometime.

[@RichardSteigmann]

Hi Andy, on a sunny day set the charging current on the inverter to a safely achievable value, check it with a current clamp and put it in BMS 👍😎☀️

[@OffGridGarageAustralia]

I thought about limiting the CCL to say 80A and wait for the MPPTs reaching that. It should be fairly constant. But how accurate is the reading at say 10A when calibrated at 80A?

[@RichardSteigmann]

@@OffGridGarageAustralia Yes, the calibration only seems to work selectively. The measuring sensor in the bms is probably not really linear

[@michaelfleming8454]

Hi Andy, nice video and yes, this is an issue in many electrical and electronic devices. Prior to smarter circuitry, an analogue signal was used to control servo motor speeds or even high speed spindles in the CNC world. Spindles that ran up to 40,000rpm were controlled using a -5v to +5v analogue signal (10v). This was calibrated quite successfully using a base voltage parameter and a scale parameter. We used a high quality power supply for the base voltage, and then adjusted the scale while checking the spindle speed with a tachometer at various different speeds to get the closest calibration across the board. These days the spindles are equipped with ethercat or some other similar communication protocol, and you simply request the desired speed. Perhaps JK can add a scale parameter to enable the user to set a high amperage baseline with a power supply, and then lower the amperage and adjust the scale parameter until the reading is accurate across the board. What do you think? Perhaps you could ask Andy #2, or the engineer that JK has assigned to you if that is even possible.

[@TrevorFraserAU]

Always fun to watch you Andy, thank you for your continued hard work and great entertainment! The education component goes without saying.

[@OffGridGarageAustralia]

Thanks a lot Trevor!

[@miklkr]

Hi! Andy, you could use a ballast resistor, such as one from a wind turbine, as a load.

[@tyweb9264]

Excellent Video again Andy, and one that I will embark on my self in about 4 weeks. As an instrument tech that has been calibrating equipment for the past 30yrs, a single point calibration will never work for exactly the reasons you have shown. In general terms, You always need a min of 2 point calibration. Typically a low and a high value and from these two points the instrument plots a slop and match's its reading to the slope. Be good if JK could implement a low value cal point and a high value cal point. Maybe 5amps for the low and 20 A for the high might be enough but with a 200amp shunt you really need to be significantly higher (60% to 80%) to get a really good representation. The lower 5 to 20 may be just fine if you are always working in this low range. I have set my BMS to a max charge of 70amps. And from my 12kw solar I typically can see 140 to 150 amps in these shoulder months where the day is cool 15-17 deg and sun is good. Currently this push's into 4 batteries, 2 more should arrive by the end of the month. When I had planned to address this for the same reasons.

[@OffGridGarageAustralia]

Thanks a lot for sharing. I'll discuss with JK.

[@bill2960]

This is the way that I would use a two point hi and low calibration point. Another point that is being missed. Is when you have a number of batteries in parallel you have a common voltage from them all and when you look at the cell voltages. They are also all very close Yet each battery can be displaying a totally different soc. I have seen in my 10 battery stack one battery down at 10% soc. And others at 60% yet all the cell voltages were within 10mv of each other we have communication between battery’s why can this not be used to sanity check the value agains the other battery’s ?

[@babaluto]

CC has been my only way of calibration. A shunt inherently drifts with temperature either ambiently or as a response to current flow. The more charge/discharge cycles outside of the top off voltages, there will be drift. I have not been able to establish any reliable references to correct the SOC outside of top off. I schedule a rop off mid season as a means of recording individual cell performance and calibration. Maintenance is not always automatic.

[@sunshineossi]

Oh, another video about the JK's😅 No matter what and how I calibrate the JK's, after a few days the SOC drift starts again and then it gets worse from day to day. Why don't the JK's get it? After all, it works with other BMS too🙄 Anyway, thanks again for the many videos from you Andy😉 By the way, my new video about the JK inverter BMS is coming out today. I'll go into our experiences and problems in the first 3 months of operation. There was also one complete failure, which wasn't so nice. Kind regards from the currently also sunny hot Germany😂

I just ignore the BMS and go with the readings from my Outback Radian inverter system shunts with the Mate3s controller and Flexnet DC current flow analyzer module which is actually designed for lead battery use.. It is most likely not super accurate either but I get a kind of point of reference and also watch for the quick rise and fall at the ends of the battery capacity cycle.. I still get red lights flashing that my capacity has bottomed out when I have 30% capacity left so certainly not accurate.. Thanks again Andy, you are a top go to guy when it comes to LiFePo4 batteries and JK BMS..

[@OffGridGarageAustralia]

Yeah, I do the same with the Victron Smart Shunt and use that as the point of truth!

[@raybiondolino9778]

Andy, love your channel !!! I’ve learned more from you in the past few years than from any other resources I’ve researched… Have you heard about the new OverKill Solar BMS that’s coming out soon? 2 models, the Pathfinder & the Nomad. Can preorder only, but the specs are impressive! Just FYI If I’m ever in Australia I’ll look you up!

[@OffGridGarageAustralia]

Thank you very much. I'll have a look at the new Overkill BMS.

[@9111logic]

Very informative experiment Andy, it would be interesting to do the same even for us Seplos users and confirm the accuracy. Thank you for sharing and for your time 🙏

[@OffGridGarageAustralia]

Thank you for your kind feedback. I can do the same with the Seplos and Pace BMS and present the outcome. The 100A JBD/Overkill BMS in the top shelf is very accurate and SOC is always close to the Victron Smart Shunt. So, they must have a different technique to measure and display SOC.

[@9111logic]

@@OffGridGarageAustralia it might be the different algorithm if that is what they're using, unless they use something more basics 🤔

[@AshleyL-.-L]

Dear Andy, any information or update on the SOC 100% stop charging issue for Growatt, Luxpower inverter etc? I updated my BMS to 15.30, the coulomb counter plus SOC 100%Volt (single cell reaching the balance start voltage) still cause the SOC to reset 100%. This results in early charging termination as usual.

[@brendanmcorrie8699]

G’day Andy, I noticed this with JBD BMS’s too, I just reload my settings file into each BMS and its OK for about 2 months. I know it’s a temporary fix but I’ll do until I find a better solution. Cheers

[@scotth9681]

Thanks for this video, Andy. I also have found JBD SOC values to be much more reliable than JK. Now I know it's not just me 😅 Several times i have had a nasty surprise thinking SOC was 40%, then seeing it jump to zero as a cell went undervoltage. It would be good to see JK fix this issue with what is an an otherwise very good BMS, even if it is just a software workaround.

[@OffGridGarageAustralia]

Thanks a lot for your support 🙏👍 Yes, this is one thing which needs to be fixed on a long list of further issues and requests. I don't know how JBD does it, but their SOC calculation is on par with the Victron smart shunt.

[@dominikw.8978]

For now we don’t know if the error is non-linear or linear with the distance to the calibrated amperage. 1. Linear: It would be sufficient to have an additional correction factor i (float!) for the inclination where i = 1.000 means no error. I = I_shunt + i*(I_shunt - I_calib) 2. Linear: Alternatively the BMS could be teached a zero amp point. 3. Non-linear: Multiple calibration points are necessary including interpolation and extrapolation.

[@OffGridGarageAustralia]

I would say, it is not linear from what I observed.

[@geekmystique]

@@OffGridGarageAustralia Assuming they programmed the BMS for a linear curve: As most BMS will have very similar shunts a software update with a more complex calculation taking some non-linearity into account would make the SoC measurement way more useable through a firmware update. Maybe you can share this hint with the JK guys to have a go at this. The only thing they need to do is do a wide range amp test against their shunt and figure out the curve- this is not at all a new challenge for programmers, similar implementations exist to compensate for poor ADC inputs on many micro controllers.

[@OpenSolarEnergy]

I use my hybrid inverter to charge the battery with 5A fixed in order to calibrate the current. Calibrating for 5A allows me to have a good measurement for powers between 0 and 500W which is most often the case. I compared the curve of a shunt with my JK-BMS for a month and the curve was very close even without recharging at 100% every day.

[@OffGridGarageAustralia]

OK, if you stay in this power range for most of the time, this may work.

[@rcinfla9017]

The ten 1 milliohm chip resistor in parallel form a 0.1 milliohm current shunt. The copper bars are just to beef up the PCB copper foil for high current. Because the ten chip resistors are distributed along the PCB there is PCB copper foil thickness tolerance which has some effect on the actual value of 0.1 milliohm net shunt resistance. The voltage drop across the 0.1 milliohm net resistance current shunt is amplified before injected into uC's analog to digital converter. Lot of places for errors to creep in, including DC offset drift vs temperature on op amp. Victron smart shunt and 712 monitor has precision temp compensated 0.1 milliohm shunt resistor, self-calibrating DC offset correction op amp, and higher ADC resolution for reading the shunt voltage drop. That is what you are paying for. If you want to attempt JK BMS current calibration your best reference is to use the number from the Victron smart shunt. If you load inverter with just a resistive heater the battery current should not jump around much.

[@egbertegberts5384]

I've added a Victron's smart shunt for this. I have more confidence in this platform for measurement purposes. The shunt is set to Battery Monitor

[@OffGridGarageAustralia]

Yes, I have the same. But what are people doing which don't have Victron gear?

[@gibo1971]

@@OffGridGarageAustraliaI have selectronic and the only time the battery matches the inverter is full. I was thinking of running battery in pylontech mode as inverter can read this but to be honest no trust bms is up to being correct . Happy to get Victron shunt and raspberry pi to be able to monitor effectively it that works.

[@egbertegberts5384]

​@@OffGridGarageAustraliaI don't think there is an effective way to calibrate the current over the full range to guarantee an accurate SOC reading. Just one argument validates my position: by resetting the SOC to 100% when just one cell reaches a certain voltage. I have it at hand myself: one of my new 305Ah EVE cells reaches that set voltage much sooner than the others (voltage difference 180mV). The BMS then reports 100% while the smart shunt is at 50%. No calibration can cope with this😅

[@Shellyfi]

Just started to get worried that is Andy OK - 6 days and no video, but phef.. video came out

[@OffGridGarageAustralia]

Yeah, we had family over for a few days. I had this video already in the pipeline but needed to finish it. 🥈

[@Carsten-PV-DIY]

​@@OffGridGarageAustraliahope you had a good time with your family over in the sunny hot 200A Australia

[@samsuffit6229]

Dear Andy thank You for this Video, I would like to come back to the process of charging the pack with the Wanptek charger that I also own since one year (thanks to You ;-)). I would like to slowly charge my GobelPower pack (14Kw) that I also bought one year ago (thanks to You too ;-)) but I am not 100% sure which setting to use, so I prefer to ask... I saw in You first attempt that You had to deselect the OCP because it triggered as soon as You started the charge. Then in the second attempt 6the CC turn red and the the CV turned off i.e. the Wanptek directly switched to constant current and adapted the voltage (You set 61 Volt ...). And what's about the cables, may I use the cables delivered with the Wanptek, are they thick enough ? Should I use an extra fuse ... even if the Gobel pack has already a 200A switch (mine is from end of 2023 and has not anymore the extra fuse internally) ? Thank You for Your help !

[@captainobvious9188]

I've learned this from a bunch of the older JK BMSs I've had running for years - Their current sensing isn't high enough resolution, so in and out currents that aren't "smooth" cause it to diverge from its actual SoC *a lot*. I've had a PV string -> MPPT charger -> JK BMS DIY battery -> Blue grid-tie inverter with limiter, running for five years in this configuration, and because the MPPT charger is doing PWM in and the GTIL is doing PWM out, it sits around 48V (14s LCO batteries) but thinks it has 95% SoC. If I fully charge them it only takes about a week until its fully off again.

[@OffGridGarageAustralia]

Best solution is maybe to use an external shunt as battery monitor and let the BMS show and calculate whatever it wants. That's how I run my battery shelf with the Smart Shunt and it woks well.

[@raystone4673]

Thanks for video.Yes I believe any readings with clamp meter should be on a scale, where the meter range is close to what your'e reading. They're most accurate at full scale. Don't measure 10 amps on a 400A meter range. It's like when you take 5.8 kg of scrap copper to a buyer, and he says "put it all on this 4000 kg scale," and pays you for 5.0 kg. It's not right. Maybe put multi turns of the sample cable through jaws of clamp meter, (clean it where jaws touch) to bring up current on say a common 80A scale little clamp meter. External available 75mV shunts used to be 1%. Now I think 0.5% are common. Just for the calibration, you could wire one in, and use a 3.5 digit or better DC multimeter to measure volts drop across it. (User can get advice on how to then calc Amps from that I if they don't know how to.) They are much more accurate on the DC volts scale, than a clamp meter is on the 100A range. Even a $30 Jaycar one is +/- 0.5% on DC volts, and I've seen 0.25% ones for a bit more or same price. For something so important, we could spend a few $'s, and sleep easy.

[@keyem4504]

I'd set the calibration current to my most common use-case to get most accurate results. However, most BMS have a minimum threshold below which they don't count the current anymore. This will lead to problems over time as well.

[@OffGridGarageAustralia]

Yes, correct. The JK has a 0.3A threshold. Any current below that will not be counted for.

[@aaabbb1579]

Hi Andi, you need 2 calibration points. Zero current and max current calibration for proper linear curve. But there's more for a proper SOC. you need a very high resolution ADC to measure very small currents - long time standby drains the battery with small current by the bms. You need to take into account a bleeding self discharge parasitic current in the cells not counted by the shunt depending on temperature, also charge / discharge efficiency (heat in the cells and bms), Balancing also destroys charge, and there is the Peukert exponent too. Complex math by current, time, temperature current direction, ADC reslution errors... Victron s,art shunt does a great job, except for self discharge leak over temperature and time and balancing losses. Also when the cell capacity depends on cell tempeature - so does the SOC too. That means it is important at what temperature you do a capacity test. Have fun!

[@Arjen46]

There also seems to be a resolution issue as it keeps jumping with 0.4 amps. Mostly on the lower side after calibrating. These inaccuracies will add up over time, especially in a ‘low’ season where you don’t hit 100% SOC. So it will certainly thrift over time again, and you might even give a hypothetical prediction based on this in inaccuracy.

[@OffGridGarageAustralia]

That is probably the threshold of the BMS measurement accuracy. Below 0.4A/0.3A it just shows 0.

[@zoe..d]

That doesn't explain the 0.4 ish A jumping at your midpoint testing of 15A. It's almost like the BMS is internally on some thresholds for switching internal circuits for measurement and you get the fluctuations despite rock solid power from the bench top supply. Hard to know without going to component level and deeper knowledge from their "engineers" asking you for the solution.

[@kappa7c]

Is it possible to use in a GX system a Victron smart shunt for soc and current measurements and the JK for others controls? For example the GX can be configured to use the soc from the smartshunt to stop the discharging under a predefined value and still use the JK information to set the charge current, float voltage, react to BMS alarms, ecc?

[@Raphael_Hofmann]

Having a calibration curve for their shunt could be the best solution. But it is probably not that easy to implement. They could meaure a couple of shunts and extrapolate a curve with a mathimatical function. But it might be easier to have some kind of linear ramp with a offset.

[@Timuragashka]

There is clearly a hardware issue: when we try to calibrate the shunt, we can't achieve accurate results. If we calibrate at low power, the high power readings are incorrect, and vice versa. To solve this problem, we could adjust the feedback loop of the operational amplifier responsible for the shunt readings. If we prefer not to modify the hardware, we can address this in software by adding multiple calibration points-both for low and high currents. Then, the algorithm can calculate a curve based on these points, leading to more accurate current readings.

[@VolodymyrPavlyuk]

The shunt is sensitive to temperature changes, so for accurate readings you also need a temperature compensation curve.

[@OffGridGarageAustralia]

I wonder how JBD does it. This BMS seems to be fairly close to the Victron Smart Shunt and stays that way throughout winter when not fully charging for weeks/months.

[@InspectorGadget2014]

As we have noticed the current sensor does not (always?) seem to be a linear response, we have opted to calibrate at 50% of max current allowed. Whilst it does show some inaccuracies in the lower/low currents, on average is it fair enough for us/me. Until BMS software allows you to have multiple data-points to input a reference-currents (let's say at every 10% increase, so 10%, 20%, 30% up to 100% and tune the real values compared to the displayed values), we believe that calibrating at 50% is good enough for now. I do believe the BMS often does partially measure but also partially does some calculations (algorithm?), as measuring requires (very) accurate components, temperatures and sufficient voltages and currents, that BMS manufacturers are not bothered as much for accuracies.

[@devonvankraft]

Andy Lets GOOOO .. if JK shows real solutions i will stay their client ;) Man i love your excellent work! Greets from the Netherlands.. Home of The Victrons! OCP = Over Corrent Protection.. s.p.a.t. =..?

[@TheRonskiman]

Solar Panel Alignment Tool

[@ronhellenbrand]

Hi Andy, Did calibrated with Victron shunt while discharging with 100A, and mine SOC reading during high current discharge are +/- correct. Problem are the small loads. If you apply AC at your multiplus II the current the battery sees is a sine wave, even @ 0 Amp. The reading of JK @ shunt 0A than does jump from 3,5A to - 3,5A. Turning off AC in, you will see lower amps than the 3,5A. JK should implement average measurements say running average of 1-2 sec, than current measurement and SOC will be correct!

[@OffGridGarageAustralia]

But discharging means negative numbers which cannot be set in the BMS.

[@ronhellenbrand]

@@OffGridGarageAustralia Just verified, put in the positive amps number during discharge and it will calibrate. during discharge @ -100A just calibrate with 100.

[@VladBYT]

I have the same problem with my JK BMS on a 12v system. Over time, my solar panel stopped charging because the MPPT is controlled by the BMS. Even if I have calibrated the current, it is still off on small values. Andy, can I update my firmware so i can have the SOC option? I have a new JK BMS with the latest version and he has this option but on the older version is not there. I have spoken with Hankzor from where I bought the first one and they say is no firmware update . Thank you !

[@OffGridGarageAustralia]

If you have the older BD series of the JK-BMS, there is no option to update the firmware. Only the newer inverter BMS can be upgraded.

[@dean5263]

I asked AI to write this : I just finished watching this incredible KZbin video that truly blew me away! From the moment it began, I was captivated by the excellent production quality and the engaging narrative style of the host. They meticulously broke down the complex topic, making it not only informative but also entertaining with a perfect blend of humor and visuals. The use of real-life examples really brought the concepts to life, and I found myself reflecting on my own experiences as I watched. The pacing was on point; it kept me hooked from start to finish without any dull moments. Furthermore, the call to action at the end was both motivating and thought-provoking, encouraging viewers to think critically about the topic discussed. I can't recommend this video enough to anyone interested in learning something new while being thoroughly entertained! It's definitely one I'll be revisiting and sharing with friends.

[@OffGridGarageAustralia]

What a lot of sh!t

[@DavidPrue]

The 100% voltage should be based on either (1) AVERAGE cell voltage or (2) Pack Voltage. If you have even moderate (~0.100v) deviation then establishing 100% SOC based on the highest cell is ludicrous. Pack voltage makes the most sense, but average cell voltage is a usable alternative.

[@SailingThelKen]

Hi Andy, thanks for all the vids and info, 100,000 soon 😁. I was wondering if you can ask the JK software eng about the parallel limit on the inverter bms, why ? any plans to go above 4 etc etc

[@OffGridGarageAustralia]

You can parallel as many as you want. There is no limit. Just make sure each battery has its own safety device, breaker or fuse.

[@SailingThelKen]

Thanks for the reply, I know you must be a busy boy replying to comments. A year or two ago I asked JK about parallel use with the B2A8S20P and was told it was not recommended. In reality I have had 4 x 24v banks running well to date. Wouldn’t the dip switch’s limit the master / slave numbers on the newer BMS ?

[@OffGridGarageAustralia]

@@SailingThelKen The switches are only of interest if you have them communicating to each other and the inverter. But without, there is no limit.

[@MR-ub6sq]

Dear Andy. You are a wonderful, humble and funny media personality. The information you provide is important. Thanks! I have a question about cooling LiFePO4 cells. If in some situation it turns out that the cells require cooling, and the JK BMS provides information about the temperature with the help of two sensors. It can be used to automate the heating of the cells for charging, as the JK BMS includes a port for the cell heating element, but apparently no connection for the cell fan. I don't need a heating element myself, because the cells are always charged at room temperature. But the limited housing of the cells can cause the need for cooling. The cells are C40 type and there are "channels" between them due to the round profile of the cells. So I'm asking, do you have a video of how, for example, I could utilize the JK BMS's temperature measurement and cell heater connection to achieve cell cooling with a case mounted fan?

[@ptrwsk]

I have 5 identical JK BMSs and observed that also Voltage needs calibration because connected to the same busbar and equalized for long time Jk shows different voltage but when checked by probe they are all the same.

[@OffGridGarageAustralia]

Yes, voltage can also be calibrated but is not important for the SOC drift or inaccurate measurement of the current. The JK has several issues here.

[@neemnc1185]

On my Sunsynk, what I did was to disconnet solar and then turned on Grid Charging the battery option, and set the Grid change rate to 20A. This match the clamp meter reading as well and then I set that in BMS caliberation. Also while not a fix, can we get JK to provide a option to set SoC level manually, atleast that will allow us to set the SoC to a more resonal number when we know it has drifted.

[@photovoltasunnyboy740]

Dear Andy , as the BMS Amp's mesurment is almost inaccurate (in my case Seplos V3.0 with actual latest firmware 1.3 as V11/200Amps revision) , if we connect a SmartShunt via VE Direct to the Cerbo GX, would it possible to base the DVCC on the SOC mesured by the SmartShunt insteadof the BMS?

[@OffGridGarageAustralia]

Yes, absolutely! Set the BMS as the battery monitor and leave the Seplos BMS connected for alarms etc.

[@sorinlepadatu873]

this is due to voltage drops on the conductors, differences in the length of the conductors, section of the conductors. there are many variables.But mostly with my system I am satisfied with the results

[@mathiasschuldt4805]

Hello Andy I have a Pace BMS Vers. 1.0 and the recipe has not worked for a few months the numbers are full but the SOC value remains at 34/ 35% with a battery the other three have around 80% can you tell me how I can fix this

[@Mike-01234]

What about the voltage calibration? I have 2 meters a Klein CL800, and a Fluke 27 both read differently by 10-12MV then the JK BMS, and the inverter. The inverter a EG4 6000xp reads differently then my Fluke 27 when put the probes right on the circuit panel where the ring terminals mount to. Maybe it doesn't really matter I have to increases my charge voltage and float voltage little to get it to balance.

[@hanskroeger6239]

JK B2A25SRP is using an external Shunt to measure the current. More accurate! Propose to use also an external shunt for other Jikong BMSs.

[@typxxilps]

I had done the JK 15.30 firmware upgrade successfully and now I am wondering what the heck this float charging is doing cause it is now pushing 290 Watt into the battery for hours even though this is already at 100% SOC. How long does floatcharging usually take for the JK BMS cause I am wondering if that is really usefull or if I should turn the battery simply off with a HomeAssistant Switch provideded by BAT Mon ? Any ideas what to do or a link to a video ? I have searched for float and absorption but did not find any or must have overlooked one .

[@sergiejsergieh]

Hej mam problem. 48V BMS JK kiedy magazyn jest naładowany cele są na poziomie 3.50 i rozbieżnośc 0.009v. Kiedy zaczynam używać i mam -10A cele znowu się rozjezdzaja o 0.020V czemu?

[@Slay3rOne]

I did use a benchtop PSU to calibrate mine some time ago, but I don't have one with enough current output, mine tops out at 3.2A. But I noticed the same issue when calibrating at a certain current, it just reports wrong values going higher or lower afterwards. I think JK should really provide multiple calibration points, maybe a 0 current as well like the SmartShunt has. Without multiple calibration points it's never gonna work right I guess. Also, we really need to be able to calibrate with negative values (and why no decimals...), because then a simple dummy load with resistors could do the trick instead of having to charge the battery at a high current. I hope they can improve the calibration procedure!

[@OffGridGarageAustralia]

I guess, calibrating the BMS with smaller currents will be very inaccurate when higher loads are connected.

[@wazoo7016]

Hi Andy and community. This could be a common issue due to the use of poor adc readings without some kind of auto-scaling, maybe they use a 12bit adc to read whole current range on a single range for 100A charge/discharge capable bms. Probably the lower current will have the higher error. And maybe the use of linear calibration methode. I think it needs some kind of curve calibration with 2 or 3 points. 🧐 Low cost engineering for a low cost bms. I do not know 🥺

[@JohnMckeogh]

thank you Andy

[@robm.4512]

I’ve called my clamp meter a GTA meter for years: Guess The Amps.

[@capebee4839]

I would start by re torquing all connections. Especially those inside the battery to see if you get a more consistent flow of current with less variability

[@Jan-g3i]

Wie bei jedem Analogeingang müssen 2 Werte abgeglichen werden. 1.) Der Offset von Verstärker und AD Wandler bei einem möglichst kleinen Wert. 2.) Die Verstärkung von Verstärker und Shunt bei einem möglichst großen Wert. Dann mehrfach wiederholen um sich anzunähern weil 2 und 1 nicht unabhängig sind. Fehler dazwischen kommen dann von missing AD Codes und Unlinearitäten welche

[@tiborchren]

Would it be possible to add a software option in the BMS that when the Victron SmartShunt is in the system to take values from it back to the BMS? Victron SmartShunt can also be connected to Solar Assistant, so somehow the data from it can be used in a non-Victron system as well.

[@OffGridGarageAustralia]

That could be possible from a technical perspective. But is this really a solution for the SOC drift of the BMS? The JBD/Overkill BMS stays very close to the Victron shunt during winter, so they must have figured it out...

[@Juergen_Miessmer]

How should a smartshunt know the current of each battery? It only meassures the total current of the whole system.

[@stephan.scharf]

Is it not possible to use solar direct current charging input instead output for calibrating? Excuse my dumb question, I'm not a BWL student ...

[@jaytheblader6701]

A series string of 4 x 12v halogen flood lamps would be a suitable resistive load for calibration. Or a couple of pieces of copper pipe in a bucket of saltwater

[@typxxilps]

DIFFERENT APPROACH: I have 2 PV systems with 8 GROWATT inverters used as a FULL FEEDER which is feeding 100% into the grid and passes the official energy meter and a usual surplus feeder for the home (attention 2 growatt inverters are part of the full feeder, not just 1) . I have compared the GROWATT metering of the past month with the official number. The GROWATT CLOUD / INVERTER figure was roughly 2012 kWh (for both), the official one showed 1999 kWh. That is just +0,6% more. And that gives me the "hope and believe" that the growatt measuring seems to be quite accurate while the JK BMS was way way off which I had assumed from the beginning cause each cell had been ZKE TECH measured multiple times each of them above 282 Ah, when the JK BMS told me that we did not get more than 241 Ah out of the battery. So we were pretty disappointed about the cells for months . Growatt shows me a draw of 1040 W from the batterty and JK BMS pretendes to deliver 892 W. while Voltage values were close to each other 52,1 to 52,2 V over the whole SOC. Hence I used this 1040 / 892 Factor to "CALIBRATE" or should I say "SPAT-ibrate" the JK BMS from 16,9A to 19,6A. Why ? Cause in that case the 282 Ah weakes cell would determine roughly the battery capacity cause if you divide 282 Ah by the "JK off factor" of 1040/892 you get the 241 Ah figure the JK BMS has always pretended to have delivered. First night is over with a constant discharging of 1040 W x 12 h = 12480 Wh while 36 Ah still remained in the battery x 51,2 V= 1840 Wh in total an estimation of 14.320 Wh BUT THERE IS A TINY CATCH: losses My 1040 W drawn are shown as the draw figure of the battery and then split into 900 W for the house and 140 W feeding in. DC 1040 W usually does not equal AC 1040 W without any losses. I assume we have to expect the usual 3,5% losses shown on the inverter data sheet or roughly 0,5 kWh losses (14,2 kWh). But again: As long as I am pretty confident that the growatt figures are accurate I can ignore that and simply have to wait another night with then maybe 1190W x 12h . Of cause a "you can not do it that way method" but why not if that is better than the JK guess work ? THe official meter is calibrated , Growatt is delivering far more closer figures than I had ever thought and if the Growatt Watts are measured right i will trust their Wattmetering too cause that is AC based. If I get 1190 W x 12 h = 14.280 Wh I can live with that or even lower figures. Of cause not really a solution for everybody cause it might require a lot but at least worth sharing the thoughts how to calibrate the JK BMS without a clamp guess o meter. I think JK needs to improve their calibration and I also have to admit this big coil on the V15 which should have been in the center of the round whole is not in the middle and the built quality and craftmanship were rather terrible, I mean all those manual labour steps you can check. Loose screws is not what you want to see at delivery on such devices. Yes, easy to fix but you have to become aware of that. After their firmware upgrade which had worked flawlessly (following your instructions from a video 6 months ago) I have ordered another batch of 2 JK Inverter BMS.

[@DanBurgaud]

1:50 I use a bench power supply, programmed to deliver 10.0A @ 1.0V and measured again by a clamp meter (same as the one you are using 2:51). Then connect power supply directly at JK's B- and P- terminals then calibrate with a clean DC! DONE! This method does not need a 60v bench supply. Even using an inexpensive 0-10V, 0-30A BUCK converter can do this trick. But at higher current flow, JK's reading would mismatch the clamp meter. If I do the reverse (calibrate at high current) it does not match at low current.

[@OffGridGarageAustralia]

Ah, yeah, good trick! The current measurement is always off at the other extreme, no matter how we calibrate.

[@john_in_phoenix]

So, you think the shunt in a $50 power supply is accurate? Ohms law is your friend. I do have a calibrated electronic load that I use, but most people don't. A good Fluke meter can get you an accurate 10 amp measurement, but really you want whatever is "normal" load. Fluke meter will measure voltage and resistance VERY accurately. Go to home depot and get a water heater element. Measure resistance, measure battery voltage. Solve for current.

[@DanBurgaud]

​@@john_in_phoenix Here are the flaws in your suggestion: #1. You cannot calibrate fractional current. Only whole numbers like 10.0A or 30.0A or 50.0A. Using your method, you need to exactly match the battery voltage to come up with whole number. (Watch the video, Andy mentioned it here) #2. JK does not allow you to calibrate in negative values; you cannot calibrate when in discharge. The only way to calibrate is during Charging and the Chargers seldom delivers constant current, nor deliver whole number. In this video, Andy used a 60V, 20A bench supply; most people do not have this and is not economically practical (at least $500) to invest for a once/twice use. A 20A buck converter costs less than $10.0 in case a bench supply is too expensive.

[@john_in_phoenix]

@@DanBurgaud Sounds like a JK problem.

[@enyasalida3736]

That was my idea also, I can even include the Victron shunt in the circuit and a 20A-Multimeter. And of cause turn off the Multiplus...

[@holgersteinhaus8530]

Hi Andy, in my camper I observed another SoC issue with the older (non-inverter) JKBMS: if the current falls below a certain threshold (0.5A?), the JKBMS just assumes 0.0A. This causes a huge amout of SoC drift in just a couple of weeks, if the battery is just sitting in idle state, powering only the cerbo at 5..7W (0.2A @ 24V). I guess the inverter BMS is doing the same. Maybe it would be interesting to discuss that topic with the JK engineer?

[@OffGridGarageAustralia]

Yes, it's a know issue with all BMS. The JBD is still very close to the Victron Smart Shunt during Winter, so I wonder what their secret is to keep the BMS so accurate.

[@loucinci3922]

This week, I find my Dell laptop battery not lasting as long. I believe it is because I am using a USB scanner and it charges the handheld scanner. I want to add a battery pack to the USB-C charging port. I wonder if I can just connect +/- from a YELLOW or RED tool battery (18/20v@3Ah) to a USB-C cable +/- to provide extra run-time? These battery packs have their own BMS. The laptop has its own BMS. I am not sure what is in the charging brick. The specks on the brick read 20v/4.5A/90W and smaller. I think it would work. What do you think?

[@DIY500AMP]

When you try to calibrate at 20A you have more voltage drop over the cables distance. When you use 10A the voltage drop is less and the number will be a bit off from the calibration. Will be hard to do. How long are your cables and how thick are the wires? Even the temperature will change the numbers.

[@OffGridGarageAustralia]

That does not matter. We are dealing with a series connection of battery, cables connections, power supply, ... the current is the same, no matter how good or bad the connections are or how long the cables might be.

[@DIY500AMP]

@@OffGridGarageAustralia thanks for the clarification!!

[@johnauxillos7804]

Calculating SOC depends on how you use the battery. For me, when its rainy season and we dont get to fully charge the battery, i estimate soc by looking at charge/discharge patterns over the past few days and the lowest cell voltage. When its sunny day where we can do full cycle, i rely more on previous day charge minus a little for the round trip. Given that estimating SOC depends on how you use the battery, JK can just put the most basic soc estimation and give the stream of data to home assistant for analysis or to create better soc estimation based on all data.

[@wayne8113]

Thanks Andy

[@OffGridGarageAustralia]

Thanks Wayne!

[@rdflo6739]

I'm no electronics guy, but i would think that jk can incorporate a simple shunt measurement device in their hardware design. Basically a resistor bridge with a current sensing amplifier that accurately tells the real shunt resistance to the bms. This will automatically compensate for shunt resistance changes as the shunt heats out cools. I'm sure there is plenty of literature about this from the automotive world. The rest is a few lines of code.

[@OffGridGarageAustralia]

I'll pass this suggestion on to JK. But I doubt they will do such a hardware upgrade. It would mean they will then have three different hardware which they have to provide the software for.

[@zoe..d]

​@OffGridGarageAustralia and yet they provide temperature sensors already in the BMS. Seems they could use its own temp sensor and some code. If X°C temp then some offset factor to the calibrating current figure.... May as well use what they have already built in to it....

[@hairband80s69]

Wondering if you set the Multiplus to "charge only", would it still jump around?

[@OffGridGarageAustralia]

This has nothing to do with the Multiplus. The battery is connected to the Phoenix inverter, not the Multi.

[@EitanTsur]

I think the only viable option is allow for shunt resistance measurements. IE: take a 4 wire bench multimeter and measure the shunt resistance to get the actual resistance value down to several decimal places past the zero. I have a 7-1/2 digit multimeter that can do kelvin measurements, having the option for shunt resistance values could be useful. That said, the shunt will vary in resistance based on temperature as well, so without some sort of profiling of the shunts across a practical temperature range, and a temp sensor on or near the shunt, I don't know that we can really rely on the measurements it provides.

[@OffGridGarageAustralia]

That's why I use the smart shunt as point of truth and ignore all the other SOC measurements. But this is not a real solution for the SOC drift of the JK BMS.

[@EitanTsur]

@@OffGridGarageAustralia I think the only way to solve the problem is for JK to profile and program the actual shunt value as part of their QC process and use a much higher accuracy than they currently are factoring for their current measurements and math internally. The current inaccuracies could easily be explained by a rounding error if they aren't going deep enough in decimal places. Part of the problem may require a hardware modification in order to account for temperature drift and the other part will require software changes in order to increase the resolution

[@maikrolf9356]

is it possible to have an option that we have multiple inputs on calibration? for example 5A 20A or 100A? so the curve is more precies?

[@typxxilps]

our JK V15 with Firmware 15.30 has no float charge issue. I guess it runs quite usuall which means that it powers the EVE LF280K cells after 100% SOC for about 1 hour with roughly 350 Watts till it drops to 0 and is on idle. Will add the comment on a more recent video as well that it might be recogniced. I have done 5 battery cycles tests during the last late summer days which meant running down to 2,6 V and then charge the cells again till noon or so. And during these charging sessions I observed the floating behaviour and itt looked and felt quite well, of cause duing cloudy hours in the noon things are running a bit different so I have not investigtated the power current and voltages for every minute cause on 2 days clouds were coming and going within 10 seconds or so.

[@davidpenfold]

Looking at my REC BMS manual, it states (in the SOC section, there's other detail on the shunt measurement): "Temperature and power correction coefficient are taken into consideration at the SOC calculation. Li-Po chemistry algorithms have an additional voltage-to-SOC regulation loop inside the algorithm. Actual cell capacity is recalculated by the number of the charging cycles as pointed out in the manufacturer’s datasheet."

[@CaptainProton1]

I'm a bit worried about the size of the main cables on the 200A JK BMS, mine get so hot at 160A and more to the point that I think the rubber coating on the cable is going to melt. Are they really good for 200A continuous

[@Juergen_Miessmer]

It is no rubber it is silicone. Silicone can handle higher temperatures.

[@martinbohm8239]

We need a new Hardware Version with a real shunt like Seplos v3

[@OffGridGarageAustralia]

Seplos V3 is terrible in SOC measurement as well!

[@CliveTrezona]

Is this clibration problem only for the JK inverter BMS? Do we need to calibrate all the other versions too?

[@Juergen_Miessmer]

Yes, all JKs need to be callibrated.

[@marcvandaele1878]

Hi Andy I have a question about your set up and JK bms,not directly connected to this topic( i don’t know how to contact you another way)coz i do something similar If i have 3 banks batteries with each there own breaker(2p16s608Ah) connected to busbar and afterwards main breaker My bms is JK -PB2A16S20P and will limit charge current to 120A Someone told me that it’s not so good idea this way coz there will be nuisance/imbalance between the 3 banks and breakers will always trip,one after the other…I don’t really believe this if i see your experience with it,but it’s running in my head😢 Bitte konnen sie mir die richtige losung mit teilen Grusse aus die Philippinen Marc auswanderer von Belgien😊

[@Dutch_off_grid_homesteading]

Heya, don't know if there is a solution as you work with AC and DC coupling as you said AC is a sinus wave and DC is a CC and you will have losses in the cables and inverter and in contact's. maybe JK can programm the BMS for a average wen discharging the battery but then still you gone have a differance.

[@OffGridGarageAustralia]

The cable loses and connections will not matter in a series connection.

[@peterle987]

the min resolution of the bms is 0.4A - so lower currents may differ easier by one quantisation level/step like 5A are displayed as 4.6. or the 15A is once 14.9 - 15,3 - 14.5. Always 1 step up or down. And yes - highest possible currents needed because of this bad resolution of current meassurement.

[@SunInLithium]

The real issue!! is with the fact that when the BMS reads 100%, it causes all in one inverters to stop changing, no matter the voltage. There needs to be an option to Stop counting at 99% until the RCV Times is up. Also, they should consider mabie adding a Peukert exponent

[@OffGridGarageAustralia]

JK is aware but has not developed a solution yet.

[@SunInLithium]

Good to know. I think the solution should be to add an option to limit the SOC counter to 99% until the RCV time is over. And then rest to 100%.

[@ChrisEldridge]

Looks like you are only allowed to do an offset calibration and not a scalar calibration. If those parameters were available I would do a 5 point calibration at 10%, 25%, 50%, 75%, and 90% to try and get the best resolution.

[@cdrbmw]

Can I calibrate the zero point also? I think mine drifts around zero.

[@anthonybianue9206]

I think we need to take into account the voltage drop in the wire and resistance inside the bms. Feels like the higher the current the less the resistance matters

[@davidabineri908]

To update bms software, do we still need to goto your friend web site for the proper code as before?

[@OffGridGarageAustralia]

No, update works without code.

[@zoheb4you2002]

I need to buy 48volt bms for my deye inverter, which bms do u recommend?

[@OffGridGarageAustralia]

Totally depends on your system design.

[@TheDefpom]

It looks like the firmware needs more calibration points, it is common for the ADC to not be linear in response (called INL), especially if they are just relying on a microcontrollers built in ADC. They probably need to add points for ZERO, 5A, 10A, 20A, 50A, 100A to ensure that the accuracy is as good as possible, in a pinch they could do a ZERO, 20A, 50A setting, to give some kind of response curve. The SOK batteries built in BMS uses a zero set point, and then a current calibration for charge, and another for dishcharge.

[@VolodymyrPavlyuk]

JK‘s current measurement issue is caused by hardware because there’s no proper shunt resistance on the board, and the frontend chip has ADC of low resolution. It’s simply impossible to have accurate readings.

[@OffGridGarageAustralia]

It works very well in the JBD-BMS. This is always very close to my smart shunt reading and SOC does not drift much during winter.

[@daisydonald14]

Hi. I've been dealing with this problem for 6 months. The problem is that there are big differences between charging and discharging. If I calibrate the BMS to 20A, it shows 20A when charging, but only 17A or 18A when discharging. This means that the SOC value is no longer correct later. I have now calibrated the charging current to be around 5A lower. This means that the BMS shows less when charging than it actually is. This also means that the SOC value increases more slowly. I can now charge my battery to 100%. But please note one thing, calibrate the current with the maximum value that you use. I charge with a maximum of 60A and then calibrate to 55A, for example. If someone charges the battery with 100A, then you also have to calibrate it at 100A to 95A, for example. And that is the whole problem with the JK, too big a difference between charging and discharging current. It would be best if JK engineers could build in a correction value. You calibrate the BMS when charging and when discharging you correct the measured value either up or down.

[@grua-vfe]

as @user-sq5ml9io4d has already mentioned: The float mode that no longer works in v15.30 is a much more serious error. Andy, please communicate this with JK, in my opinion this would be even more important than the drifting SOC.

[@OffGridGarageAustralia]

Float mode is not working at all any more? What is the exact issue? I haven't noticed anything in V15.30... EDIT: OK, got it. I'll test and get back to JK

[@enib3179]

will there be an update for the older JK aswell?

[@OffGridGarageAustralia]

No, cannot be upgraded.

[@GregOnSummit]

Did you calibrate the voltage too?

[@martinschoonover2812]

Are they using Manganin alloy in the shunt, which compensates for temperature changes?

[@OffGridGarageAustralia]

Hi🙋‍♂️

[@martinschoonover2812]

@@OffGridGarageAustralia Since the shunt is a calibrated resistor, the more current you pass through it, the hotter it gets. Manganin was developed as an alloy that would keep the resistance constant up to fairly high temperatures. My JK BMSs suck as a reliable measurement of capacity, and I calibrated them too.

[@svenmueller]

The problem isn't that the measurement is imprecise. The problem is that it deviates to different amounts. In some example that is somewhat close to your measurements: If you charge for 1h at 20A and the BMS measures 20A and then you discharge for 2h at 10A and the BMS measures 9A, it will drift (towards reporting a too high charge percentage, by 2Ah, whatever percentage that would be, in this example). Would it measure 18A and 9A respectively, it would be wrong only temporarily (report a too low SoC after the 20A charge) but be right again after the 2h, 10A discharge. Because it measured 18Ah in and 18Ah out. (Our really built BYD LVM has the same issue though)

[@OffGridGarageAustralia]

Exactly, some errors will compensate over time and the BMS is fairly accurate again. But now at the end of winter, it's all over the place and the JK have drifted a lot! The JBD however stays very close to the Victron shunt during that time. They must have a better technique in measuring and computing SOC.

[@adon8672]

​@@OffGridGarageAustraliaHi Andy. I guess you should report the better performance of the JBD to JK. Maybe they can look into it and reverse engineer.

[@sjdtmv]

The display looks like it is 4 or 8 bits, so there will be only set number of jumps in amps displayed, what would be ideal is a plus or minus adjustment instead of clearing the whole number with a new number, but I think that is still controlled by how many bits each jump as you increase or decrease the value

[@teapot_]

Perhaps you should be working for JK. I think you would be very valuable.

[@OffGridGarageAustralia]

I think, I kind of do already😜

[@tullgutten]

Might help with the calibration to have large capacitors and inductors on the BMS input to decrease any voltage ripple (every PSU have voltage ripple) Just be sure to slowly turn down the charging voltage before disconnecting to prevent overvoltage from the inductors. Might be safer to just go with capacitors and no inductors

[@OffGridGarageAustralia]

The inverter already has huge capacitor banks at its DC input. Adding more will make very little difference in this case.

[@dumitruionescu7545]

Hi please could you tell me what is the polarity of wire B0?

[@erwa8729]

I don't care about SoC of the bms, only of the smart shunt. In winter time there's also a problem with victron shunt. I ignore it.

[@theshuff]

I'm in trouble with a new jk inverter bms. Cant get to absorb phase with my sofar inverter. When bms hit soc-100-volt, inverter just stop voltage charge and i cant do any absorb or balance. Do you know if jk have any plan to untie soc-100-volt from other voltages ? or trigger 100% soc only after the absorb timer directive ? or do i have to revert formware to version where soc 100% volt was not a feature ? thanks.

[@AshleyL-.-L]

There is no such firmware. That SOC100%Volt setting is there since the initial release of the v14 hardware. Growatt,Sofar, Luxpower, EG4 etc or any Voltronic OEM design are affected. I also would really like to know the answer to your questions as well.

[@theshuff]

@@AshleyL-.-L i heard in another videos that older jk firmware reset soc only when ovp was hit ?!? maybe i'm wrong or it is pnly on older hardware jkbms.

[@AshleyL-.-L]

@@theshuff Nope, that is totally different BMS. The older jkBMS or non-inverter series cannot be flashed/updated at all. You are using new JKBMS which is capable of inverter-bms communication and this series has serious flaw with coulomb counting and SOC100% reset by single cell voltage issue which results in early charging termination problem.

[@loucinci3922]

Enjoyed. thanks

[@rainonsunday6510]

Maybe you need to solve the problem not by software but by hardware? The developer may release a separate mini board that would fix this problem. Customers would buy it, connect it and forget about the problem. And the new version of BMS would already have this board in it.

[@matthewwakeham2206]

My victron shunt can be 12% out after a few days. Not sure why. It'll go from 88% to 100% based on the voltage. That capacity doesn't seem to be missing so I think some electrons are not being counted.

[@OffGridGarageAustralia]

That sounds like a wrong setting in the shunt.

[@Jeff-ow4ud]

This is a question about the dry contacts on the BMS. What can they be programed for? Low battery voltage? High Battery Voltage? High Cell Voltage? Low Cell Voltage? Can all the above be selected together to turn this contact on/off? In canada you need a BMS wiht contactor and I wish to use the Dry contact ot do so with the above conditions set to trip the dry contact that will be controling the High Current contacotor I have any thoughts?

[@OffGridGarageAustralia]

See this video here where I go through all the settings of this BMS: kzbin.info/www/bejne/m4u7aHaheqqklZo

[@Jeff-ow4ud]

Hello Andy Yes I seen that, but what i would like is to see the ability to select the Battery Over Voltage, Battery Under Voltage, Cell Over Voltage and Cell Under Voltage all at the same time. If any of the 4 conditions are met then have the dry contact flip state. Can this be done with a firmware update?

[@OffGridGarageAustralia]

​@@Jeff-ow4ud Technically this can be included but imagine the mess having to program the ON and OFF values for each of your 4 trigger points. I guess that's why they have included option 10 (System Alarms) and the contact closes at any given alarm. I doubt JK will address all individual requests for specific use cases. The fw functionality would explode.

[@Jeff-ow4ud]

I can see the point. A thought would be to have them pull the same numbers from the other settings? OV ,UV, Low Cell and High Cell? I will look into the Option 10 and see how that works. May have to invest in the LCD screen. WIll it tell you what system alarm on the screen or just that there is a problem?

[@OffGridGarageAustralia]

@@Jeff-ow4ud The screen and the app still tell you what the problem is, the relay will trigger at any of the protections regardless.

[@cgutowski471]

Ugh. So no answer to the 100% SOC issue (single cell vs pack). Thought you mentioned it would be discussed in this next video.

[@AshleyL-.-L]

Indeed. I am kinda disappointed after having a high hope of it getting resolved.

[@SandeepSingh-ht3ug]

I'm guessing there could be ~50w loss (1amp @ 50v) between charger and the battery due to various factors such as wiring, connectors, fuses / breakers etc..

[@OffGridGarageAustralia]

The amps are not getting lost though. All components are in series connection and carry the same current. Connections, cables don't matter for that.