The ADS1115 his a higher resolution in PGA1 mode (4.096volt).
So aim peak voltages at that, not at 5volt.
Leo..
Sorry for the delay. I appreciate your input.
I have spent quite a bit of time evaluating potential solutions to my unique needs. Briefly put., I need three things.
- I need a voltage representing individual bank currents to run the peak detector.
- I need to take that stored voltage, digitize it and send it to the MCU.
(that voltage does not need absolute accuracy because it will used only as a bank to bank comparison) - I need a digitized total bank charge current somewhat accurate but reasonably reproducible. This will be used to determine end amps and float currents in the 3-10amp range.
My issue is the dramatic imbalance on the shunts.
Total discharge current ≈400amps Total charge current ≈60amps.
A bipolar adc with a 400amp discharge range cannot accurately discriminate in the 4amp float charge current range
I am willing to go in depth as to how I am trying to resolve this but only if there is an interest.
In the meantime, are there any unipolar ADC modules you may be aware of?
? That's only 1:100.
A 15-bit (uni-directional) INA226 has a 1:~32000 ratio.
Leo..
You make a good point. Do these numbers make sense?
ADS1115 full scale =500amps (could happen under surge conditions)
500amps ÷ 32000 = .0156 amps =16ma LSB
worst case of trying to discriminate 2 amps
.016amps ÷ 2amps = .008
± LSB would be ±1%
In which case the ADS1115 would serve my purposes?
Also note that an LM324 could have an input offset error of 4mV.
Which get worse if you amplify that.
Leo..
Hi, @barryfields
How will you know which battery current you are measuring if you put all the voltage drops in parallel?
You will need one 228 ADC per current shunt.
Tom.... 
I do not need to know each individual banks current.
Any inconsistencies between banks will be evident in the results of the separate peak detector mentioned in the first post.
I would still use one INA and one shunt per battery.
With those currents there also will be voltage differences between battery and measuring points due to battery cablesamd connections.
You draw the positive as a rail, but that's only theory.
Leo.
A quick calculation shows that a 0.5 inch lenght of 1/4" x 1 1/2" aluminium bar has the same resistance as a 100 Amp shunt. Are you sure you're measuring current at the right places.
Aluminium seems a poor choice at those low voltages and high currents.
And I'm not even talking about contact resistance.
Leo..
What value of current are you expecting to get from all the shunt outputs paralleled together?
It won't be the total of all the currents from each bank.
Your circuit in post #1 is measuring what, battery voltage or battery current?
Don't forget ALL your batteries will show the same voltage.
Why not have ONE shunt in the output bus-bar to your loads and measure there?
Log the current, as you will be measuring the TOTAL output current, you can detect and measure the peak discharge.
ONE SHUNT...
Tom....
FROM POST 1
Current handling might not a problem, but volt drop could be.
Every inch of busbar is a shunt that could throw off your current measurements.
Leo..
I would differ.
You are correct that any buss or connection resistance would affect the battery current. However, because the shunt is in series with the battery it will accurately measure the actual current through that battery. I see no parasitic resistances parallel to the shunt.
The ADC will see this input as if it were connected to a 400a 75mv shunt.
Not my point.
Look at VBus of the INA228 in post#26, connected to the four positive battery terminals.
Practically, there will be three shunts between the four positive battery terminals.
That will throw off the measurements of the INA.
Every drawn wire in that diagram that carries significant current is a parasitic resistor.
Again, use four INA chips and shunts with Kelvin connections to the INA.
Leo..
Not the current measurement. I am only monitoring voltage and current NOT power.
The bias currents for Vbus ,Vin+ and Vin- are less than 3na.
The worst distance between Banks is about 5 feet or .1mΩ . One voltage connection will be accurate enough for my specific purpose.
Remember, my peak detector will indicate any problematic parasitic problems.
How different do you believe the voltage levels are if measured on the buss next to the bank attachment points?
Sorry I didn't look closely at your diagram and thought you had the shunts in the positive lines, which is the norm for INA chips.
Having the shunts in the negative battery lines creates the same problems though.
Volt drop between battery negatives adds or subtracts from the true shunt voltages with your unconventional resistor adder.
I'm done here. I don't see why you're fighting the use of four INA chips.
Leo..