NHW11 Traction battery autopsy

Excellent!

From the User's Manual:

Suggesting 10 megohms is the normal impedance.

So think of the meter as being a 10 Mohm resistor being inserted into a megohm resistance network and batteries. FYI, nice instrument.

Given there are active voltages from the batteries in the circuit, I need to think about this. Trying to measure the resistance of a battery is . . . not trivial. <wink>

Bob Wilson

 

Since the OP is trying to determine which battery module has a ground leak, why not remove both busbars, then use the Fluke 87V to measure voltage from the terminals of each module to the traction battery case.

If you don't measure any voltage then there's no ground leak. Hopefully module 2, 3 and/or 4 will show some voltage from a battery terminal to the case which would prove electrolyte leakage. Or perhaps visual inspection would reveal the leak based upon busbar corrosion nearby.

 

The leakage path is often visible. If it is not, the method you describe is the simplest and fastest way to isolate the leak. It's not a complicated process.

 

You don't need to know battery resistance to find a leak. Leaks are found outside of the module.

 

That is the technique used with the Megger. You don't want the battery voltage impacting the Megger measurement. This sketch might help:


The open circuit voltage increases by the voltage of each module or module pair. However, the leak resistance to ground is likely to be high but lower than the 1-10 Mohm impedance of the DVM. The DVM 'resistance' and the leak resistance forms a voltage divider network and the measured voltage will be proportional to the resistance pairs. But the easiest answer is to look for where the sign changes.

In reality, there may be additional leaks to ground of variable resistance values. So it can become somewhat complex. However, every place where the voltage achieves a relative low, absolute value, is likely to be a leak to ground.

Bob Wilson

 

You're right. And once you get up into the megohm range, resistances - especially power to ground resistances - start to act differently than "conventional" resistances. A resistance in the millions of ohms can change by a significant amount from day to day. This makes early detection difficult. All that matters is that the resistance is well above the minimum resistance required to establish adequate insulation.

No, there's no downside.

You might build such a rig if you were tinkering with a single module, and didn't have access to a pack.

 

The DVM . . . I'll fix the sketch and repost shortly. (OWCH, I'll have to redraw it but I did include example calculations.)

Unless it uses AC, a megger would not be my first choice in a circuit with active DC voltages. But I only have a cheap, Chinese knock off and keep looking for an excuse to use it. The problem is the DC voltage of the batteries could impact the megger response. I would expect that reversing the leads would give a diffferent result showing the battery voltage bias problem.

The megger would work fine if looking at one terminal but all batteries come with two. If the leak is on the other side, the battery voltage would bias a DC based, megger and reversing the leads would show it.

Bob Wilson

 

It's not a matter of choice. Megohmeters are not intended for use on live circuits.

 

Correct. In fact the MRC 989 allows programming of several battery profiles and modes which it keeps. The User's Guide is accurate and with a little practice it becomes easy.

The time depends upon the what you program in as the Ahr capacity. New, our modules have 6.5 Ahr which means 1 hour to do a 6.5 A charge or discharge which is called 1 C. In practice, most folks are using 0.1 C for the conditioning charge/discharge cycle. So for a 6.5 Ahr battery that would be 650 ma. However, our used modules are going to be closer to 2.5 A.

Now an undocumented feature, you can limit the charge put on a battery. Normally it would go from the minimium voltage per cell, 1 V, to the maximum based upon the dV. But it won't charge over ~120% of the capacity entered in the profile (or at least mine won't.) What this means is we can pre-charge the modules to 80% of capacity but I don't see a problem with going to 100% and equalizing the voltage before the assemble pack goes back into the car.

Bob Wilson

 

This is were I'm at the limits of my 'hands on' knowledge. The literature indicates that a series battery needs to have all cells as close to identical capacity as possible. This means after the survey, the weakest and if exception, the strongest will be left over. The goal is to get them as close as possible but like many things in life, we have to deal with the resources we have.

This is one area where the depot level rebuilders like Re-InVolt can maintain an inventory of surveyed modules. The larger the pool, the easier to pick matched modules for the whole pack. Better still, they've had time for 'lessons learned' which us individuals have to 'cogitate.'

The plot of your bench battery voltages gives a clue:

Notice the self discharge shows a pattern, almost every other one, about one cell low compared to the upper ones. So what makes sense is to rank the modules strongest to weakest then reassemble alternating weak and strong modules from the set that will be used. The battery ECU reports the module voltages as pairs so a weak and strong together should have similar voltage vs charge profiles across all 19 pairs.

In a perfect world, we would charge all of the modules to 80% of capacity and then 'normalize' the voltage for each module to be identical . . . using the MRC 989 to tweak them to as close to identical voltage as possible. Now I'm going to suggest something I've never read about but may solve the problem of equalizing the SOC of 38 modules.

Assemble the pack with the chosen modules making sure all the mechanicals are right and module orientation but DO NOT PUT ON THE BUSS BARS. Get all of the modules as close as practical with the MRC 989 . . . then wire all the positives together and all of negatives . . . one massive, parallel set of batteries. In effect, one massive, six-cell, 38 module battery with 38*Ahr capacity. This will ensure they are all at the same starting voltage, the closest possible to an identical SOC. Let it sit for 12-24 hours to equalize.

Remove the jumpers and install the buss bars on one side. Very cautiously install the second set as you'll have two, ~135 V, lethal packs. Check the sense wires at the connectors to make sure none are broken. Finish reassembly, install in the car, put in the interlock, connect the 12 V and it is time for the 'smoke test.'

Bob Wilson

 

Have you personally carried out this process before? If so, what were the results?