I have a battery which supplies a true-sinewave inverter, and I'd like to measure the battery conductance/resistance/impedance/admittance in situ.
The inverter current varies as I=Io.sin^2(ωt) almost reaching zero around the zero crossings of the inverter output voltage.
I have a current shunt and a voltage divider connected to A/D inputs, to measure the battery current and voltage.
Does anyone have a neat algorithm that could derive the battery conductance?

 

It will be difficult to do with the battery connected in circuit. The internal resistance is usually calculated by measuring the battery open circuit terminal voltage and again with a resistive load. The internal resistance is the difference between the two voltages divided by the load current. The internal resistance could be calculated in circuit if the terminal voltage was measured at two different load currents.

 

You've thrown around a bunch of terms that all mean different things, so I'm guessing you're not really sure what you are asking about.
Are you talking about the internal series resistance of the battery?
In other words, the deviation of an actual battery from an ideal voltage source.
If so that is fairly easy to estimate, at least when the battery is fully charged, and it may not remain constant as the battery discharges.
What kind of battery chemistry are we talking about?

For all kinds of information on batteries, I recommend Battery University.
https://batteryuniversity.com/

 

Are you talking about the internal series resistance of the battery?

Yes.

Good call on Battery University, I've used it before for other stuff, and it is excellent. I'll see what they say about conductance meters.

 

Conductance = 1/resistance

 

The internal resistance is usually calculated by measuring the battery open circuit terminal voltage and again with a resistive load.

Easier said than done - have you ever tried it? I should have mentioned it's a lead acid battery.
On connecting the load the terminal voltage does what is known as "coup de fouet" - due to the surface charge, a big dip, back up again to the original voltage then down again, and then drifts slowly down as the battery starts to discharge.
I've seen conductance meters which claim to inject an AC current, and I have an AC current as part of the system . . . .

 

Like everyone has said it's tricky to measure online with loads.

I use a set of switchable load resistors to get a rough calculation of battery internal resistance. First to reduce surface charge with a voltage stabilization check and then to switch load currents for a resistance calculation.
https://forum.allaboutcircuits.com/...c-controlled-battery-array.32879/post-1457747

 

Yes.

Good call on Battery University, I've used it before for other stuff, and it is excellent. I'll see what they say about conductance meters.

I apologize if my original sentence sentence seemed harsh, but I was seriously trying to understand the potpourri of related but dissimilar terms. No offense was intended. I'm glad I was able to interpret your meaning.

 

What I'm really looking for here is a smart algorithm to calculate the answer (why I put it in the microcontroller section, not the analogue section)
Assuming I have a series of samples of V and another series of samples of I, (sampling rate to be determined) what I'm trying to design is a smart bit of software which will take my samples of I and V and give me G or R (whichever way up you do the division).

 

I apologize if my original sentence sentence seemed harsh, but I was seriously trying to understand the potpourri of related but dissimilar terms. No offense was intended. I'm glad I was able to interpret your meaning.

No offence taken.

 

No offence taken.

Hope you find a good solution. The answer may be a "pas de deux". Namely the battery and the microcontroller.

 

What I'm really looking for here is a smart algorithm to calculate the answer (why I put it in the microcontroller section, not the analogue section)
Assuming I have a series of samples of V and another series of samples of I, (sampling rate to be determined) what I'm trying to design is a smart bit of software which will take my samples of I and V and give me G or R (whichever way up you do the division).

There are no really smart algorithms for this beyond basic circuit theory. Most of the fancy electrical conductance testing measures are varying forms of voodoo trying to remote-view the internal condition of an electro-chemical battery using the bones of only electrical measurements. Don't expect too much actual accuracy as most battery health measurement system have huge slop factors.

 

There are no really smart algorithms for this beyond basic circuit theory. Most of the fancy electrical conductance testing measures are varying forms of voodoo.

It's not easy to avoid stepping in the doo doo when you do voodoo.
Sorry, I couldn't resist.

 

It's not easy to avoid stepping in the doo doo when you do voodoo.
Sorry, I couldn't resist.

Stepping in it is fine. The burning detail is not.

 

Watch the guys trying to consume stumps with various methods of fire. Some are environmentally conscious, others are not.

 

If you can measure the instantaneous voltage and current of the battery while under load and the load varies, then you can just subtract the difference in voltage and the difference in current at the high and low points and divide the two differences to get the battery internal resistance.

 

I think I have spotted the shortcut:
statistically, the mean-square value of the ripple current is its variance (σ^2), and variance is easy to calculate
σ^2 = Σ(x^2)/n - (Σx/n)^2.
I'm already calculating Σx/n because that's the mean value
So it's just a matter of calculating σ^2 for I and for V, dividing one by the other, and taking the square root. (Only one division, and the square-root routine also only runs once). Then I have the admittance or the impedance, which is close enough to what I need as there isn't much phase shift at 100Hz.

 

look at battery gas gauges.

 

Ohmic measurements (internal resistance and conductance) can indicate battery changes but it can't indicate what specific (SoC, sulfation, ...) internal electrochemical changes are happening.

https://en.wikipedia.org/wiki/Randles_circuit

It does give valuable information that can be used to infer battery conditions but it's much like trying to discover that's wrong with the gas powered engine of a electrical generator by analyzing the electrical output characteristics under load.