Time to get back into some technical work
I have been working on a way to detect if a cell goes low in a LIFEPO4 pack without monitoring every cell. The method works only if the cells are top balanced. The pack I did the development work with is a 42 cell 40AH GBS pack from Elite power solutions using there basic balancer board that sits on top of the cell. This is a stand alone board with no commutation to any readout board or computer. So I have no wires coming off the cells for monitoring.

The method I proposed is based on the principle that if you top balance then when the pack starts to be drained one cell, the cell with the lowest capacity will drop in voltage first. This being said if one broke the pack into two half for measurement purposes  

Positive














Center tap














Negative


Now if we  measure the balance between the two half's when the pack is near drained one cell is going to drop in voltage first causing a measured imbalance between the two half's secondly the voltage difference between the two half's will become noisy meaning the voltage will bounce around as the cell hit there knee and drop in voltage. This setup can also detect which cell is fully charged first as it's voltage rises above the other cells again causing a pack imbalance.

This is how I set up to measure the balance between the two half's

All the resistors are .1% 100ppm drift very high precision wire wound resistors
Observations:

1. The balance between the two half of the pack has never deviated more than 20mv
2. A one volt drop in a single cell is 200 mv
3. When the pack is full charged and in balance mode I see no more than 70mv
4. I I intentionally unbalance the two half's .1watt drain on one half of the pack for 48 hours, when the pack started top balancing I observed 1 volt of unbalance, The balances did there job and the pack settled out at 70mv

For a first test to manually see if this is a viable method to detect a first cell or a series of cells at there knee voltage I concluded it is a good method to detect the point in which one should stop using a battery pack and call it drained. This method is very poor at determining state of charge and can only detect when to stop using a pack or risk cell damage. In combination with counting Watt-hr per mile to determine range I feel a almost fool proof method for a battery gauge and  protection can be developed without running wires to monitor every cell or communication links daisy chaining from cell to cell.