Chargery: AH measurements

[Cal]

I've been working on a DIY Coulomb counter. https://diysolarforum.com/threads/coulomb-counter-for-diy-bms.11610/ I've noticed the Chargery SOC and AH counts were way off base. With the Coulomb counter complete, here's a comparison of my counter and Chargery's serial data. The entire test encompasses 45 hours. The graph shown is just for the first 25 hours.

The first 15 hours occur with no solar activity. There's a quiescent load current of about 0.9A. The DIY AH counter starts from zero and ramps down for about 16 hours. The Chargery counter doesn't detect the current and AHs remain at zero until there's adequate solar current. After 16 hours Chargery has a -15 AH error.

At dawn (about the 15 hour mark), the DIY counter reacts 2 hours earlier to solar current.

At the 20 hour mark the microwave was turned on for two periods of 3 minutes each. About 160A to 155A are drawn during that time.

About the 23 hour mark (dusk) Chargery's curve flattens. Any current below an amp is registered as zero.



Chargery's AH/WH information is a disaster. The AH/WH/SOC info will be too high. In my case, errors can accumulate at 15 AH per day.

 

[grizzzman]

Ya, I agree. I use the bogart TM2030 for AH "facts" Gotta love it, A cell level BMS and a battery level BMS. Go figure.

 

[Cal]

I've been in email contact with Chargery (Jason Wang). He chose to respond to this thread to my email address. In the past I asked him what's the BMS current measurement resolution. He didn't answer the question, only stated the shunt is rated 300A @ 75 mV and the ADC is 12-bit.

I hope he doesn't mind I post his private message here.

i checked your latest post ...
we use 12bit CPU. to detection current under 1A, the resolution is not enough, if quiescent load current is about 0.9A, our BMS cannot detect it, so 15AH error is believable, but i don't agree Chargery's AH/WH information is a disaster.
the important thing is which device consume 0.9A from battery? as you know, our mechanical relay has a around 0.9A holding current, cause battery capacity is wastes, many users don't accept it, so we develop new DC contactor to solve the question. if other devices still drain 0.9A from battery when battery is not in charge or in discharge, the quiescent load current is too large (21.6AH per 24 hours), so it is not suitable for less than 30AH battery, otherwise the battery is easy to be damaged whatever BMS detect the current accurate or not.

BMS should / must detect charge or discharge current accurately, from 0.1-300A or 600A, it is difficult, but it can be realized. the question is we have to consider the cost. when charge current is less than 1A, many chargers will stop charging, when discharge current is less than 1A, i don't know which load drain 1A or less. or which load drain 1A from battery with BMS.

to solve the question,

1. use as small current shunt as possible to improve current accuracy
2. use 16bit CPU.

anyway, the active balance BMS is in developing, current under 0.3A will be detected at the same time.

any suggestions please let me know. if you know or find any BMS can detect less than 1A in up to 300A scope, please email me. very interesting in test it.

thanks in advance.

 

[Maast]

Great data! His reply makes me ask, have you run the same test with a 1.5 or 2A nighttime load? As I have a Chargery 16T I'd be very interested in the results and the idle from the inverter and other "ghost" loads (TV, night lights, etc) is about 5A. If the error is still very high I'll have to look into some other means of tracking the State Of Charge.

 

[BiduleOhm]

we use 12bit CPU. to detection current under 1A, the resolution is not enough, if quiescent load current is about 0.9A, our BMS cannot detect it, so 15AH error is believable, but i don't agree Chargery's AH/WH information is a disaster.

Hmmm... 12 bits is 4096 counts and 300 A / 4096 = 0.073 A (or 0.146 if you want to be able to measure negative currents too) so even if the ENOB is 11 bits the resolution is still significantly better than 1 A, even if it's not stellar of course.

I bet the main problem was accuracy, probably too many false readings at low currents.

the important thing is which device consume 0.9A from battery? as you know, our mechanical relay has a around 0.9A holding current, cause battery capacity is wastes, many users don't accept it, so we develop new DC contactor to solve the question. if other devices still drain 0.9A from battery when battery is not in charge or in discharge, the quiescent load current is too large (21.6AH per 24 hours), so it is not suitable for less than 30AH battery, otherwise the battery is easy to be damaged whatever BMS detect the current accurate or not.

Well I think we're all free to have the quiescent current we want on our batteries... ? That's not the problem here.

BMS should / must detect charge or discharge current accurately, from 0.1-300A or 600A, it is difficult, but it can be realized. the question is we have to consider the cost.

That sums it up: cost cutting from a 14 or 16 bits ADC to a 12 bits one (probably internal to the MCU too; explaining the accuracy problems...)

anyway, the active balance BMS is in developing, current under 0.3A will be detected at the same time.

Meh, come on, spend the 3 or 4 $ on a proper 16 bits ADC and get easily down in the 0.0x A range.

 

[Cal]

Hmmm... 12 bits is 4096 counts and 300 A / 4096 = 0.073 A (or 0.146 if you want to be able to measure negative currents too) so even if the ENOB is 11 bits the resolution is still significantly better than 1 A, even if it's not stellar of course.

I bet the main problem was accuracy, probably too many false readings at low currents.

Meh, come on, spend the 3 or 4 $ on a proper 16 bits ADC and get easily down in the 0.0x A range.

Here's a plot of Chargery BMS early morning solar current readings. Around the 17.1 hour mark Chargery detects enough current to register 1A. It then starts flipping back and forth between 0 and 1A until there's a solid 1A present. It looks to me resolution is 0.25A. Chargery decided not to measure below +/- 1A. That's not good.

Edit: comparison to the DIY current readings.

 

[Cal]

Great data! His reply makes me ask, have you run the same test with a 1.5 or 2A nighttime load? As I have a Chargery 16T I'd be very interested in the results and the idle from the inverter and other "ghost" loads (TV, night lights, etc) is about 5A. If the error is still very high I'll have to look into some other means of tracking the State Of Charge.

Don't need to run the data with another load. Solar clearly shows when the BMS decides to give up current measurements. Don't know which shunt you have, but any currents between +/- 1.0A are reduced to 0A for a 300A shunt.

Jason says resolution is better with a 100A shunt. However I believe neglecting current lower than 1A still apply to 100A BMS. I installed a 100A Chargery BMS in another RV that I own. I suspect it has the same problems.

You will get into trouble if you rely of Chargery SOC, even with a 1.5 to 2A quiescent load.

 

[BiduleOhm]

Here's a plot of Chargery BMS early morning solar current readings. Around the 17.1 hour mark Chargery detects enough current to register 1A. It then starts flipping back and forth between 0 and 1A until there's a solid 1A present. It looks to me resolution is 0.25A. Chargery decided not to measure below +/- 1A. That's not good.View attachment 23584

Yep, steps are clearly 0.25 A and what's weird is that there's some dissymmetry: sometimes they're a bit under and sometimes a bit over. The 0.5 A steps are always right on the line however.

My guess is that those steps are software related (and the fact they're 1/4 A and not a random weird value like the ADC steps tend to confirm that) as they don't match with ADC steps even if you discard the LSB (11 bits would give 0.293 A steps). They must be doing some rounding of the raw value to the nearest 1/4 A value.

Why they don't use the full resolution (I mean 12 bits isn't super great to being with so better not waste anything IMHO...) and do some averaging to avoid most of the false readings (and have a higher ENOB at the same time) is a mystery to me as it so easy and effective. Their weird 1/4 A rounding must take more code than averaging would need ^^

 

[Craig]

This is unfortunate that it is not 100% accurate but I guess if you want complete accuracy you need to pay much more for a batrium or other high end expensive device.

I'm not sure if bad information is better or worse than having a Daly BMS with no information.

maybe someday soon @BiduleOhm will have his BMS available for a reasonable price.

But until then we are stuck with whats available.

 

[BiduleOhm]

maybe someday soon @BiduleOhm will have his BMS available for a reasonable price.

Well, I work on it as fast as I can but I just got a new job a few weeks ago (that's why I didn't post anything new on it since then) so not a lot of free time available at the moment; it should get better with time tho, I hope ?

For the price I don't really know yet what it'll be but obviously more expensive than chinese BMS, and less expensive than the Batrium. If we use the rule of thumb of twice the BoM cost for the finished product then we would be around 500-600 USD for the 5 boards (and less than that if you don't want the optional boards of course) but that's really just a quick guess. Of course I can also sell kits (or even just the PCBs) which would cost less. I didn't think about all that as selling it wasn't the initial goal and as I need to have a fully functional unit before going further anyway.

 

[Craig]

Well, I work on it as fast as I can but I just got a new job a few weeks ago (that's why I didn't post anything new on it since then) so not a lot of free time available at the moment; it should get better with time tho, I hope ?

For the price I don't really know yet what it'll be but obviously more expensive than chinese BMS, and less expensive than the Batrium. If we use the rule of thumb of twice the BoM cost for the finished product then we would be around 500-600 USD for the 5 boards (and less than that if you don't want the optional boards of course) but that's really just a quick guess. Of course I can also sell kits (or even just the PCBs) which would cost less. I didn't think about all that as selling it wasn't the initial goal and as I need to have a fully functional unit before going further anyway.

No worri I was more trying to explain these things are not easy nor cheap.

 

[BiduleOhm]

Yep, exactly; technical problems are pretty much always solvable, but that needs time (and often increasing BoM costs too... ^^). Costs however you can't really do much more once you already optimized everything you could, besides cutting corners on the design and/or using lower quality components (both of which are really not a good idea on something like a BMS IMHO).

 

[Cal]

This is unfortunate that it is not 100% accurate but I guess if you want complete accuracy you need to pay much more for a batrium or other high end expensive device.

It's far from 100% accurate for me. My solar harvest is in the range 30 - 40 AH day. That's an error of 38 to 50% per day! For me the display is a disaster. At the beginning I relied on the SOC reading. The battery then discharged to 0V.

 

[Bob B]

I think the reality is that those of us that have the current Chargery models are going to be forced to have a separate battery monitor. I have been using this one on a temporary basis for testing and it seems much more accurate than the Chargery. https://www.amazon.com/gp/product/B07FGFFHC6/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&th=1 I may eventually add a Victron monitor to my system.

I think the problem will eventually be solved with a new hardware platform .... kinda like what is happening to displace the old relays with the new DCC.

 

[Craig]

It's far from 100% accurate for me. My solar harvest is in the range 30 - 40 AH day. That's an error of 38 to 50% per day! For me the display is a disaster. At the beginning I relied on the SOC reading. The battery then discharged to 0V.

Not saying it is not a disaster I dont use the shunt at all and use voltage I know roughly where I am at at all times. Sorry I'm not a purist at all but volts can still give you a rough idea.

 

[Craig]

It's far from 100% accurate for me. My solar harvest is in the range 30 - 40 AH day. That's an error of 38 to 50% per day! For me the display is a disaster. At the beginning I relied on the SOC reading. The battery then discharged to 0V.

did not the LVD settings work?

 

[Cal]

There was no LVD. I would have installed a disconnect had I know Chargery SOC measurement was crap . I've been RV'ing for over 30 years. Never had a lead acid battery discharge below 50%. I know how to take care of batteries. Just need reliable instrumentation.

 

[BiduleOhm]

Small question: is there a datasheet of some sort for the BMS?

 

[Craig]

There was no LVD. I would have installed a disconnect had I know Chargery SOC measurement was crap . I've been RV'ing for over 30 years. Never had a lead acid battery discharge below 50%. I know how to take care of batteries. Just need reliable instrumentation.

This is not at all meant to be argumentative. But I feel that solely using columb counting as battery level is a bit short sighted even with the best BMS in the world. Sooner or later your battery is going to wear out and sooner rather than later with lead acid. once that battery starts to go your columb counter is going to think it stores x Ah when it actually only stores Y y AH so it will therefore be off.

I am not trying to say that the Chargery issue should be ignored I'm just saying everybody needs to realize that just using Columb counting will not cut it either.

Now as to your use case I really find it hard to believe that someone of your expertise wouldn't also have some sort of idea what his battery volts are. You absolutely should be irritated that the counter is wrong, I do not have a problem with that at all. But I dont think its fair for you to stand on your high horse and say that the Chargery ruined your perfect record of maintaining batteries. Any monitor should be "trusted but verified" to steal a phrase from a past President.

 

[Cal]

But I dont think its fair for you to stand on your high horse and say that the Chargery ruined your perfect record of maintaining batteries. Any monitor should be "trusted but verified" to steal a phrase from a past President.

Say what?

Not standing on any "high horse"

Just presented the facts on Chargery current measurement and AH calculations.

I get the feeling YOU, as a seller of Chargery hardware, sound a bit disgrumpled.

Don't shoot the messenger!