HV Monitoring / Reconditioning Questions and Insights

Hi all! New to reconditioning. Gen3 owner since early 2012.

Background
* I'm an EE and pretty good around DIY stuff. Turns out my woodworking tools and skills have also been handy
* I didn't know about the battery fan (and need for maintenance) until obvious signs of trouble (strange SOC drops, etc)
* I've now got:
- A good OBD (LX), and both the HA and Dr Prius apps (and torque for good measure but not using it much)
- Prolong Deluxe
- Some replacement modules: I was desperate to have a functioning vehicle for a recent 5200+ mile three week trip...
- A good (Fluke) voltmeter, remote thermometer, and lots of computer equipment including PSU's. No RC chg/dsc boxes.
- Home-built module tester: two headlight bulbs in series (110W total)

Progress to date
- Initial use of Prolong was probably hurried, but it was clear (Dr Prius) that my pack was in trouble. At least one bad block... maybe a few.
- Got five modules - sadly two were worse than what I have, but three were good enough to get us on the road
- During our travels, mid-day the HV shut down with a code (IIRC, we had gone down a mountain, then going up - so massive charge then discharge, and suddenly deltaSOC went bad enough that it disconnected the HV. I power cycled the car, cleared codes, waited a few minutes... and amazingly all was well. No further issues.
- Ordered three more replacement modules, thinking I needed at least some stability and no more dead ones.
- I pulled the battery (much easier now that I'm used to it! I even found the nut I lost the first time LOL)
- Did a two minute 110W load test, and attempted to match high/low pairs for the reinstall
*** I now think this was a mistake: not enough of a test to discover the weakest modules? ***
- Reassembled, and just finished a complete three cycle (134v, 104v, 84v) Prolong sequence. Final charge of ~36 hours ended at 246 volts, higher than I've ever seen.
- As I write, I'm doing a simple grid load test suggested by someone on this forum: use the discharger to get the HV down to 200 volts, and see how the blocks are doing.

An Unusual Twist
I'm probably just blind, but have not seen anybody mention doing the following for process monitoring:
* Attach a battery keeper to 12v so you don't drain the AGM (I am amazed how easy it is to drain the AGM if not in ready mode!)
* Turn on the Prius in Aux mode (two presses without the brake
* Run Hybrid Assistant "Battery Check" to monitor block voltages during various process steps (particularly the deep-dive discharge!)

IMPORTANT NOTES:
* The ECU does not like what it sees! You'll get various codes... and occasionally the car will turn itself off to "save the battery" (after about an hour?)
* DO NOT allow it to go into Ready mode. Very bad...

Questions
* Obviously, discharging to 84 volts means each of 28 modules is down to 3 volts on average. In my case, the disparity was huge: one block at 13 volts, a few a bit lower, then some dropped even down to 2 volts!

Q: Clearly, I didn't do a sufficient load test (or something) on my modules. How much of a load test is enough?

Q: What's the actual best/important parameter for module-matching? I'm beginning to think it may be IR rather than delta-V-under-load... but I've not yet found a good way to test per-module IR.

Q: Is there a good way to test per-module internal resistance, without spending a bunch more money?

Q: Is IR consistent for a given module, at any SOC level (or at least any typical range of SOC?)

That's enough for now. More fun, charts, photos and more to come...

 

I've removed the question about weak modules. Obviously it's a distraction. LOL

 

Any thoughts on my questions?

 

Answering some of my own questions:

1) Found a very good pair of Gen3-specific videos on HV removal/replacement. There are nice simplifications available for Gen3, compared to the Gen1/2 process.
* Removal:

* Disassembly:

2) Directly measuring IR (using electronics meters) with any accuracy is very difficult, unless one has some unusual (and costly) meters.
* I have a nice Fluke meter. But it only measures ohms to 0.1 +/- two digits. We're looking for numbers in the 0.0nn range
* The result can be had by measuring voltage and current accurately, using a known high precision small resistor (I own an 11 ohm 100W... could get a 1 ohm 100W if needed...) Our modules are over 6 volts... my meter measures 6.nn +/- ... (if it were below six, I could get 5.nnn which might be enough. Current/watts is going to be similar (6 volts / 1 ohm = 6 Amps; 6 x 6 V x I = 36 watts)
* Rather than purchase an RC charger/discharger with IR calculation, I decided to continue, using what I already have.

3) So, I made the following assumptions and measurements. Harder to write down than to do it --

a) The Prolong system fully charged the HV... AND running the discharger down to ~200 volts most likely revealed the weaker modules.
b) I used Hybrid Assistant to record the draw-down for every module. (Attaching a voltage-keeper, and turning the car on in non-Ready mode)
c) This revealed the strong and weak module pairs (ie blocks)
- Any block still high has two very good modules
- Any block that went low or way low -- not so good
d) I clustered blocks in several groups, based on their performance during the draw down (b above)
e) I measured the terminal voltage for every module. Obvious which modules in the "bad" blocks were OK vs possibly weak
f) I did a standard 2 x 55W headlight draw-down test to get data on both my available replacement modules, and on the weaker modules in the pack. I used this to sort the replacements into where they fit within the weaker modules.
g) The weakest/dead modules are gone.
h) Did the usual sequence, clustering strongest+weakest modules into blocks, hopefully to minimize delta V
i) Reassemble, and do a new Prolong charge + (3x discharge/charge)

 

Something about this setup doesn't make sense: if I attach the resistor across the battery, won't Vload and ΔVr be identical?

More significant: this is exactly how I measured... and discovered there is not nearly enough resolution to learn anything.
Given a 10.6 ohm (200 watt!) resistor, a sample set of numbers I get are:
rOhm 10.6
Vopen 7.92
Vload 7.89
dVbat 0.03
dVresist 7.89
Iload 0.744
rBat 0.040

40 mOhm... but here's the hard part: This is VERY sensitive to the voltages, and even a great meter is +/- 1 in the last digit.
If I go up by 0.01 volt (splitting the value between Vopen and Vload), rBat is 54 mOhm.
If I go down by .01 volt, rBat is 27 mOhm.

So, my honest measure is at best 40 +/- 13 mOhm.

And these numbers don't match what the Prius says... not even close.

I have some ideas for a more dynamic way to measure... Will play with it.

 

I use No-ox too. Yep, easily could have some mOhms in wires and connections. After all, my rather valuable Fluke multimeter shows 0 to 0.1 ohms just from quiescent leads, and a heck of a lot more if they are wiggled. Way too sensitive.

I thought I had a better dynamic method idea... but nope.

Best bet right now without more equipmentis using the Dr Prius data recorder feature, which includes block IR. Slow, but can learn a lot by swapping modules.

At this point, I'm ready to buy an RC charger with IR measurement... IF that actually works.

 

I'm pretty sure I have found a reliable way to measure IR producing similar results to what Toyota does... I'll start a new thread. @mjoo THANK YOU! Your hints were crucial

 

@deis It's a LOT simpler than that in some ways
* Pull the entire HV battery system out after disconnecting. See the two videos I linked above.
* No need to clean plastic surfaces. They don't carry any power. (Reminds me of the home building inspector who insisted I needed to add a ground wire across a union that linked a copper pipe and a plastic pipe...eventually they were fired LOL)
* The copper bus bars are easily cleaned with a vinegar-salt mixture (milder than HCl which cleans even quicker)... (then dip/wipe with water+baking soda to stop the reaction.)
* I don't know where you live, but my battery modules emerge clean from the car. "years of grease picked up from the air"??? Yuck!
* DO apply good conducting grease to the contact bolts, and YES use a small torque wrench to apply the correct 48 inch pounds when tightening. Not a problem.

The hard part is proper testing of each module, and deciding how to make up module pairs, which make up the 14 "blocks" seen by the HV system. The 14 pair-blocks need to have a close match over time, particularly during acceleration and charging (which can be VERY high power briefly. When one block is quite different, you get a code and nor more EV power. See elsewhere for those discussions. I'm working on a new way to do that, making use of Internal Resistance measures...