Chargery: AH measurements

[Capt Bill]

Got a chance to check Chargery new current measurement capabilities. The threshold for current measurement changed from 1A to 0.5A on a 300A shunt. Current resolution remains the same at 0.25 A / step. Graph shows solar charge for the day (130W panel). x-axis is time in minutes.

View attachment 29326

Thanks to Chargery for improving current measurement.

Nice to see Chargery making Improvements. Re: "new current measurement capabilities. The threshold for current measurement changed from 1A to 0.5A on a 300A shunt." IS that mostly from Chargery's recent firmware update to v4.2 ??? ,,, or is that newer shunt ... part of that accuracy improvement picture too? ... just wondering if there is any why to know ... :+)

 

[Bob B]

Nice to see Chargery making Improvements. Re: "new current measurement capabilities. The threshold for current measurement changed from 1A to 0.5A on a 300A shunt." IS that mostly from Chargery's recent firmware update to v4.2 ??? ,,, or is that newer shunt ... part of that accuracy improvement picture too? ... just wondering if there is any why to know ... :+)

Yes .... the 4.02 software update.

According to Chargery, the 300 is supposed to go from storage to charge at .3A, but with Cal's logging and my visual testing, it takes .5A to go from storage to charge.
My results from watching the display deviate from Cal's logging after that for some reason. As I increase the charge from .5 up to 1A, the display updates every .1A.
Not sure why the logging and the display visual updates are different.

 

[Chargery]

Yes .... the 4.02 software update.

According to Chargery, the 300 is supposed to go from storage to charge at .3A, but with Cal's logging and my visual testing, it takes .5A to go from storage to charge.
My results from watching the display deviate from Cal's logging after that for some reason. As I increase the charge from .5 up to 1A, the display updates every .1A.
Not sure why the logging and the display visual updates are different.

The text is from manual V4.2, with 300A shunt the minimum current of 0.5A can be measured, with 100A shunt the minimum current of 0.3A can be measured, with 100A shunt, on new Active BMS, the improvements is realized by optimizing software.
use 16 bit ADC, 0.1A must be measured with 100A, 300A even 600A shunt. it is still in developing.

 

[Chargery]

Got a chance to check Chargery new current measurement capabilities. The threshold for current measurement changed from 1A to 0.5A on a 300A shunt. Current resolution remains the same at 0.25 A / step. Graph shows solar charge for the day (130W panel). x-axis is time in minutes.

View attachment 29326

Thanks to Chargery for improving current measurement.

thanks Cal, your testing result is correct, and as same as my tesing. if use 16bit ADC, 0.1A shoud be detected on 300A shunt.

 

[joeblack5]

Not sure that i follow the above calculation.

16bit adc has 65536 steps full range that is 32768 for positive ( charging )and 32768 for discharging.
With a 300 Amp range the resolution is then 300/32768 =0.0091 Amp or 9 mAmp.
so the chargery does not have a 16 bit adc or has severe problems keeping it stable to have to limit it to a resolution of 100mAmp.

 

[Factory400]

Not sure that i follow the above calculation.

16bit adc has 65536 steps full range that is 32768 for positive ( charging )and 32768 for discharging.
With a 300 Amp range the resolution is then 300/32768 =0.0091 Amp or 9 mAmp.
so the chargery does not have a 16 bit adc or has severe problems keeping it stable to have to limit it to a resolution of 100mAmp.

A 16bit ADC does not effectively deliver 16bit performance. Although with oversampling and other techniques that slow it down, you can get fairly close. As I have said before - the analog side of the circuit is delicate and must be carefully considered to avoid noise that will start eliminating bits.

Let say for argument that the effective number of bits is 14 which is easy to get.
That ends up being around 37mV resolution.

Only managing 100mV resolution seems like all sorts of design problems.

 

[grizzzman]

.

 

[joeblack5]

factory400, I agree on all that. This is a slow process so over sampling and most other tricks would be fine.
Just puzzled that chargery mentioned 16 bit and then confirms a measurement of 0.1 Amp, wonder if he just made a decimal mistake.

 

[Factory400]

factory400, I agree on all that. This is a slow process so over sampling and most other tricks would be fine.
Just puzzled that chargery mentioned 16 bit and then confirms a measurement of 0.1 Amp, wonder if he just made a decimal mistake.

Well.....considering the existing unit is 12bit but delivers something like an unstable 8bit ADC - his estimate may not be too far off.

 

[Cal]

Not sure that i follow the above calculation.

16bit adc has 65536 steps full range that is 32768 for positive ( charging )and 32768 for discharging.
With a 300 Amp range the resolution is then 300/32768 =0.0091 Amp or 9 mAmp.
so the chargery does not have a 16 bit adc or has severe problems keeping it stable to have to limit it to a resolution of 100mAmp.

I don't believe Chargery has been using a 16 bit adc. His new units might be, but not what I have. My graph shows 250 mA granularity for a 300A shunt. The green print above says 0.5A for 300A shunt.

 

[Factory400]

I don't believe Chargery has been using a 16 bit adc. His new units might be, but not what I have. My graph shows 250 mA granularity for a 300A shunt. The green print above says 0.5A for 300A shunt.

It sounded like a hypothetical expectation for a future product. The current one that uses a 12bit ADC is dismally implemented IMHO.

 

[BiduleOhm]

My guess is that it uses the ADC integrated to the MCU with no external voltage reference. The theorical number of bits is probably 12, as most integrated ADC are, but the ENOB is far lower.

 

[Factory400]

My guess is that it uses the ADC integrated to the MCU with no external voltage reference. The theorical number of bits is probably 12, as most integrated ADC are, but the ENOB is far lower.

STM32 internal ADC is what I believe is being used. There is a reference on the board, but I don't believe that STM32 has an external reference input - could be wrong on that.

 

[Cal]

STM32 has an external ref input. With 12 bits, the absolute best you can get is 300A/2^11 = 150 mA

I've been playing around with a INA226. It gets 6 mA granularity with the 300A shunt.

 

[Factory400]

STM32 has an external ref input. With 12 bits, the absolute best you can get is 300A/2^11 = 150 mA

I've been playing around with a INA226. It gets 6 mA granularity with the 300A shunt.

That sounds reasonble for that part. I have been using those for many years and really like them. Once they are programmed - they do all of the hard work and just spit out the results.

Now - they just started shipping the 20bit version called the INA229 which I think is a drop-in replacement. That part coupled close to the shunt with a simple uC could be a very good energy monitor.