Do all server rack batteries not measure low amperage?

[mvonw]

My server rack manufacturer told me that their batteries do not measure power (SOC) if it is below .5A. What this means, is if you have large battery bank and extended periods of low energy usage, the SOC will never update off 100%. Is this a common thing with all BMSs? What good is closed loop comms if the SOC is so unreliable? I actually think my batteries are not registering below 2A.

I started seeing my batteries go flat even though the BMSs were reporting very high SOC.

 

[OzSolar]

Which server rack batteries are you referring to?

I've always suspected that the meters can't help but be inaccurate at very low currents. If you've tried to make sense of your DVM's specs they appear to indicate they've got a specific range that they are accurate in. Above or below that range and they can't be trusted.

Same thing will happen to any size of battery that you leave hooked up to your monitoring system but have turned off the charging sources for whatever reason. I've learned this lesson the hard way a few times. EG: I've put together a few systems in my shop for testing then unhooked all of the loads and charging sources except a the Smart Shunt or Cerbo GX in preparation for delivering and installing at the job site. For a variety of reasons the 1 or 2 weeks before the project start date got pushed back to 2 or 3 months and I found the batteries had engaged thier LVD when I got back to them.

I assume this is the reason that the SOK server rack batteries I've been around go to sleep after 24hrs of no activity. What I can't sure sure is if the mA draw of the smart shunt or the LED status indicator of generic state of charge meter would be enough to keep thier BMS awake.

 

[Lt.Dan]

It's pretty typical with most all BMS's. It's hard to calculate low current so accurately. Some BMS's let you change the "threshold" because some will have stray readings, where they think current is flowing, but its not, leading to even more inaccurate readings.

If you just use your very expensive batteries that you paid for, then it's a non-issue. It's also why you don't buy too big of a BMS than you will be using. For example, no sense in a 1000A BMS (just hyptothetical) when you will only use 50a at one time. The 1000a BMS is going to be the most accurate at its rated capacity, same with a 50a BMS.

 

[mvonw]

Mine are Trophy. With a big battery bank (mine is 720AH), I have to imagine the average amp draw for many installations has to be low for a good part of the day, which would mean the SOC is always way off, obviating the benefits of closed loop communications. If the BMS values are some unreliable, why do people make such a big deal about comms with their inverters?

 

[BentleyJ]

BMS communication with the inverter at this point in time is really only about being able to include the BMS info on the inverter menu but in some cases it over-rides the user menu settings. Many seem to believe that closed loop communication will somehow make the system work better when in fact it sometimes does just the opposite.

Regarding shunt accuracy: The problem is primarily due to limitations of the Analog to Digital converter and the size of the least significant value. A typical 8 or 10 bit microcontroller only has 256 or 1024 discrete steps and since it has to handle both positive and negative values only half of the steps are available for charge or discharge current meaning each step is going to represent a big jump. For example a 100A, 50mV shunt is only going to produce 0.5mV with 1 amp passing through. If a system has no loads other than the internal electronics and only using 200mA then the shunt is only producing 0.1mV.
The other issue is how the internal BMS electronics are powered. If directly from the battery connections or via the cell voltage monitoring wires then that current doesn't even pass through the shunt. Lastly, since current flows from high to low potential, no matter what side of the shunt one were to place the BMS power wire ( assuming it had a separate one) the current would only register on the shunt during charging or discharging depending on which side it was on.

 

[Lt.Dan]

Mine are Trophy. With a big battery bank (mine is 720AH), I have to imagine the average amp draw for many installations has to be low for a good part of the day, which would mean the SOC is always way off, obviating the benefits of closed loop communications. If the BMS values are some unreliable, why do people make such a big deal about comms with their inverters?

Even a 1a draw over 24hrs is only 24Ah of your 720Ah. That's only a 3% deviation across 24hrs. Not much to worry over. It's not like the battery is showing 50% overnight.

 

[HighTechLab]

SOK rack battery doesn't measure below .5a - we can turn this down, but we found that the noise we get is typically higher than the actual current drawn in low-current scenarios. It is a 13 bit ADC +- 8 counts. This means it has a current resolution of .0122a, and an accuracy of +- approximately .1a. At 0a, this means it could be reading .1a or -.1a.

But, for solar systems systems with battery banks sized properly, they don't really have a chance to see current below .5a much at all...Figure C/5 rate for charging and c/20 rate for discharging. Let's say it's a 100ah battery we're discharging and you drain your battery only to 50% at night (50ah) in 20 hours, thats still 2.5a current (average) to stay out of the deadband. I'm giving it the devils advocate, because I see most systems drain their batteries to 20% or so on the longest 20 hours of dark days...but the fact that these BMSs recalibrate their SOC at 100% means that even if there was a little bit of deviation, the next day (or two) when the sun comes out they are back in calibration again!

This however is one reason why I always recommend against purposely stopping charging on these rack batteries at anything other than 100%. Also for larger systems that don't drain as deeply, the simple solution is to put a main system shunt. Now you have a precise instrument instead of the tolerance of 10-30 shunts operating at the edge of their capability.

 

[mvonw]

the simple solution is to put a main system shunt.

Thanks for that explanation. It makes sense now. I just purchased a victron smartshunt to see if it will be more accurate than the Sol ark 12K built-in shunt I have. My battery bank is larger than necessary because my cabin is off grid, and when I’m not there, i run a bunch of low draw devices such as starlink, routers, cameras, and refrigerator, etc. On snowy winter days, I don’t always get 100% charge. If I’m not there for a couple of weeks, I began to realize that the BMS SOC were way off from the inverter, which was probably closer to reality. My main take away is that closed loop communication isn’t worth the trouble…