NHW11 Traction battery autopsy

Welcome to the leading edge.

Using .7 A is ~ 1/10 C charge rate is a good conservative rate. Although higher currents can be used, this is pretty much what many battery spec sheets recommend and some may use. Furthermore, getting ~13 modules done is great progress although you do have the advantage of two MRC 989s. You've taken the right approach, get them above their self-discharge voltage. Once they are all 'above the line,' then follow-up with "Auto Cycle" that uses the charge-discharge and survey-cycle the modules.

What I found is the MRC 989 uses the 2500 mAh to set an undocumented over charge limit. From memory, it will limit the total charge to 1.2 * 2500 mAh ~= 3000 mAh. When the charge cycle completes, it should show the amount it charged to. Certainly it does with the Auto Cycle which we'll go over later. BTW, my original source on the 2.5 Ah capacity came from a Dept. of Energy fleet survey of the traction batteries after 160,000 miles and confirmed with the used, battery I'd bought several years ago.

You'll use "Auto Cycle" which does the following:

1. Initial discharge to 1 V per cell - this just brings the module to a baseline value.
2. Charge to either dV or 1.2*Ah setting - using .7 to 2 A rates, you'll measure the module capacity. With a high enough Ah setting, the dV (or temperature sensor) will bring the module up to the most it will take. Then the charge cycle stops.
3. discharge cycle - using .7 to 1 A rates, it brings the module back down to 1 V per cell. Now you have a charge cycle and discharge cycle and the ratio shows how as these modules age, their charge-discharge efficiency falls off.
4. second cycle - usually pretty much the same as the first, it provides a quality check and leaves the module as fully charged as it can take thanks to the dV trigger.
5. After the cycles complete, read and record the values - it has to be manually read out and I just record the values in a spreadsheet. In fact I manually record all interesting data in spreadsheets to analyze the curves.

What happens with a failed module is they get hot. Worse, they never recover. There is a plastic, fiber mat that separates the electrodes. When heat melts the fiber mat, it makes a short and that cell is irreversibly dead and thus the module. Another observation is the failed cells discharge too quickly.

That is what the 9 V battery does. It allows you to keep it powered up and bring it into the house. The bad news is the ~80 ma. draw will drain a 9 V battery quickly but it lets you get the unit into the house and if you'll bare two, offset spots on the 9 V wire, you can 'clip on' a 9-12 V DC 'wall wart' (NOTE: MRC 989 in power supply mode is constant current and does not work well) in the house and take the 9 V battery off. This will give you unlimited time in the house with the ScanGauge in front of w2co and vincent's thread on the codes.

Programming a ScanGauge requires it go through the vehicle OBD bus initialization. This means the ignition ON. Once initialization is done, a few seconds with a message, plug in the 9 V battery and you can remove the ScanGauge from the car and take it inside to finish programming. Then using a 12 V 'wall wart', you can remove the 9 V battery and program as many XGAUGEs as needed.

Vincent's most recent post 'cracked that nut.' He has figured out how to read the three stored code as 'hex' values and manually decode them. The ECU holds these codes until either cleared by an OBD scanner or connecting Ts, pin 14, to ground or disconnecting the 12 V aux battery (but this also clears all trim value.) There is no known 'history file' but just stored codes and some sort of 'snapshot' memory. A history file or better still, a USB thumb drive would be a great enhancement but . . . .

I'm going to recommend three XGAUGEs for the traction battery:

1. minimum module voltage
2. maximum module voltage
3. minimum module ID

The difference between the minimum and maximum shows how the weakest module(s) are doing. The minimum module ID lets you see if just one or more than one modules are showing the lowest voltages. If it stays on just one, that would be the one most needing replacement.

Bob Wilson

 

After testing a 12 V wall wart and two bare areas on the 9 V line, my B+ is on the connector side of the diode, I did the following:

1. connected ScanGauge to NHW11 - aux battery voltage brought up the display
2. turned car to ON - this enables the ECUs
3. ScanGauge shows "Connected"
4. Plugged in 9 V, sustainer battery
5. Turned off car
6. Removed ScanGauge from OBD connector leaving the pig-tail with sustainer battery connected
7. Went into house and clamped on 12 V wall wart
8. Entered programming mode: MORE>MORE>MORE>XGAUGE
9. select the gauge number to EDIT
10. Edit using the table below

	Column 1	Column 2	Column 3	Column 4	Column 5	Column 6	Column 7
0	Lowest Battery Block Voltage	82D5F101AE	044145AE	2810	000100010000	Vlo	XX.X volts
1	Highest Battery Block Voltage	82D5F101AF	044145AF	2810	000100010000	Vhi	XX.X volts
2	Lowest Battery Block Number	82D5F101AE	044105AE	3808	000100010000	blo	XX
3	Highest Battery Block Number	82D5F101AF	044105AF	3808	000100010000	bhi	XX

Vincent's table.

Vincent reports that if two XGAUGEs are programmed from the same TXD, only one will display at one time. I'm planning to find out on my morning drive.

The drive in I monitored the Vhi and Vlo which are two separate XGAUGE values. Shortly after pulling out of the driveway, their values 'froze.' A quick unplug and re-plug brought the display back and both values showed nominal, expected values and updates. However, the display was blacking out as if the FUEL>HYBRID had been forgotten. At lunch time, I check to see if this cleared. Also, I changed the polling interval to SLOW after I parked. This is on the hypothesis that if no response comes in fast enough, the ScanGauge decides to blank the display.

Bob Wilson

 

Ah yes, the setup. This should not have to be done in the car but if you have a wall wart, you may be able to do it in the house. Vincent and w2co may have some better ideas but this is what I'm using

• MORE>SETUP>ENGINE>1.5L :: matches car
• MORE>SETUP>FUEL>TYPE>HYBRID :: keeps display running with ICE off
• MORE>SETUP>FUEL>TANK SIZE>12 GALLONS :: matches car
• MORE>MORE>RATE>SLOW :: polling rate, may require car initialization

There may be other values that Vincent and w2co use. I'm learning the tool too. Regardless, you can see the potential.

This is good because it means the lowest voltage pair is not fixed. I have a similar pattern although more low modules are included. Seeing the low block distribute across a larger number of modules is better. What we didn't want to see is a single module pair reported over and over again. That would be the signature of a failed module.

Now this begins to suggest several strategies:

1. Fault-isolate the leak to ground in the running car - clear the cause, probably a leak to ground, and return to service.
2. Rebalance the modules in running car - using Vincent's data as well as per module voltages when the pack is removed, restack the modules to balance the modules so #7 is less frequently the low block and the low block spans a wider range.
3. Measure capacity of all modules and rebuild the running traction battery with the best matched set of 38 modules.

Now you might consider that you are likely to have 36-37 left over modules and this begs a couple of experimental techniques:

• traction battery 'topper' - the remaining modules are configured to apply a topping charge to the traction battery after being house current charged. This is a home-made, PHEV.
• 12 V system topper - a pair of modules in series have a nominal voltage of 14.4 V and the 12 V system normally draws current from the DC-to-DC inverter. A 'gang' of modules configured to supply voltage to the 12 V system would offload the DC-to-DC load, a low-voltage, PHEV system. The NiMH batteries would be charged 'in the house.'
• Sell the good modules on Ebay for those who want to 'patch' a traction battery with a failing module or experimental purposes. You'll be able to report their Ah capacity and we can even go over how to measure their internal resistance.

Bob Wilson

 

Why not? A healthy battery pack can exhibit that behavior.

In addition, a failing battery pack does not necessarily exhibit that behavior.

However, the OP reports that the vehicle has not run for two years. What does that tell you?

 

.

 

Good conservative numbers and approach.

The RC community is more aggressive with their batteries. The lower end chargers often use battery temperature to cut-off which is pretty abusive. Upon overcharge, an NiMH battery generates hydrogen and oxygen gas and these combine in the battery to generate heat. Needless to say, this is rough so the long delay makes sense. RC hobbists are not always patient.

I set my delay to 1 minute, the minimum, and have not had a problem. With a conservative charge-discharge cycle they should never have excessive heat and need a longer 'cooling off' period.

Other than shortening the delay, these are good conservative numbers. But once they have completed charge-discharge cycles, read out the values from the memory and you'll have the Ah capacity. Then you have options:

• increase charge rate based upon Ah capacity - higher current, faster charge and the same with discharge. However, you are likely to see more heat.
• Adjust charge/discharge rates to moderate temperature - once you have some metrics, you'll be able to adjust the currents and monitoring the change in temperature, keep it modest. I typically see a 5C (~10F change from ambient) temperature change in my Prius so that would be a good target.

Literally, you are at the leading edge. Other than Re-InVolt and possibly a few others, no one else is doing this quality of work in Prius open forums. You are at the same place I am. WELCOME!

Bob Wilson

 

My apologies for the error; I had been reading the thread initiated by lapp, and had confused the back stories in two threads. However, the corrections regarding minimum and maximum voltages still stand.

 

This is using the 0.7A charge and 1.0A discharge with the 20 minute pause still?

Ok, do the math:

2.6 Ah / 0.7 A ~= 3.7 hours :: assumes 100% efficiency​

I prefer to use low values and often the cycles take a day. Now that you have an initial Ah value, it is perfectly OK to change the 'C' charge rate:

• 1/10 C = 0.250 A
• 1/5 C = 0.500 A
• 1/2 C = 1.25 A
• 1 C = 2.5 A

That seems about right. I found it only saves the value if it completes the cycle, not a running score.

With mine I see two values for each cycle:

• discharge value
• charge value

What I found is a failing module will often show a charge value but the discharge is nearly instant. I've seen this mostly in the self-discharged modules and it is a puzzle.

My hypothesis is that after a sufficiently long self-discharge, the metal hydride loses some aspect of its structure or chemistry. I found one reference that suggested vanadium can pass into solution in the KOH electrolyte. So far, I have not found any way to recover a battery that has passed into this state.

It is a real puzzle: (1) charge appears to follow NiMH profile but with a high voltage, and (2) discharge appears to be nearly instant as if there is no charge there. This is why the recent thread about recovering a traction battery after two years is so puzzling.

There has been one report that a capacitive discharge into the module can recover. Certainly that was true with NiCD cells that formed dendrites. But the risk is to the plastic separator mat. I have this vision of the 'arc' both vaporizing a dendrite and blowing a hole in the separator.

I wasn't kidding when I said you are at the leading edge of our collective knowledge.

For background, you might want to start with the Wiki article:

Nickel-metal hydride battery - Wikipedia, the free encyclopedia

One of the more interesting news reports this past year has been development of a high temperature, separator material . . . one that can go over 100 C. It won't help modules built 10 years ago with weak terminals. But it is interesting to see if we can find a solution to this problem.

Bob Wilson

 

No argument here!

What version of software? I'm seeing each charge and discharge value.

This is what I'd found, another undocumented 'feature:'

2810 / 2500 = 112.4% :: I'd remembered 120%, sorry
​

Now the other thing is what is the actual battery capacity? IMHO, it is the discharge mAh, not the charge side. What you're seeing the loss of charge-discharge efficiency.

Based upon these numbers, 2-3 A is certainly OK as well as increasing the Ah setting. I see no problem with going to 7000 mAh and letting the dV stop the charge cycle. For grins, record the starting and ending temperature.

GREAT WORK!
Bob Wilson

 

Hummm, I thought there was a way of checking. No luck.

I've kept it at 5 mV as I don't have enough information to make an informed guess.

Bob Wilson

 

Good deal. I used 100F as my temperature limit.

This is reasonable. Roughly how far along are you out of the 38?

You may want to program the traction battery temperature probes. Charging is exothermic and they can get warm.

The 0.2 V difference is pretty close to what I'm seeing with my traction battery. As long as it is below 0.3 V, you are in good shape . . . other than we need to see about the P3009 eventually.

I'm not sure what that is. I know some values are read out that are effectively fixed numbers.

I'm not sure about DTC but it is curious one pair of modules always shows high. We're usually interested in a consistently low pair.

Bob Wilson