Recals Reconsidered

[nemystic]

While this forum contains much information about the Insight’s recalibration function, the algorithms for determination and display of the SoC (battery pack State of Charge) have not been revealed, and consequently, some aspects of the “recal” remain a mystery to most Insight owners.

For those who may not be interested in the technical details, please skip to the last paragraph.

For those who may not be familiar; here’s a brief, simplified summary:

* Insight Recalibration is when the SoC indicator drops to a very low value at a relatively fast rate, which is not consistent with normal charge / discharge of the battery pack.

* After a recal, the regeneration (charge) priority changes until the iSoC (indicated State of Charge) reaches approximately 35%. (The SoC indicator isn’t necessarily linear, so reference to % may be misleading; but if you’re counting the bars in the display while in motion, you are not driving safely!)

* SoC for NiMH (and some other types of batteries) isn’t easily measured. It’s assumed that the Insight’s SoC determination is based primarily on integrated current input and output (to and from the battery pack).

* Since the Insight’s 144 volt battery pack consists of many individual cells configured in series, aberrant behavior of only one cell can adversely affect the performance of the entire pack. (A chain is only as strong as its weakest link.)

* The immediate cause of a recal is assumed to be a recognized difference between the calculated and actual SoC. It’s also assumed that this condition is detected by voltage and/or temperature inconsistencies within the battery pack.

* Why some battery packs fail before others is not public information. The factors assumed to affect battery life include total mileage, travel terrain, and large temperature variations. However there’s compelling evidence that the most significant factor is the initial inconsistency of individual cell characteristics. (‘In other words, the “luck of the draw”.) Differences in battery cell characteristics are unfortunately reinforced by charge / discharge cycles and inconsistent temperatures within the pack.

* The earlier model year Insight’s BCM’s (battery control modules) have a sub-optimal charging pattern that can adversely affect battery life.

* The frequency and iSoC threshold of recals is a predictor of ultimate battery pack failure (sometimes referred to as the “code of death”).

I’ve recently observed in my own Insight that with recovery from a (negative) recal, there is usually a positive recal that occurs as the iSoC approaches 80 - 85%. This positive recal is a relatively rapid increase in the iSoC.

As an unofficial survey, how many have seen positive recals?

[james]

I see the positive recal quite often - every time I see a negative recal, in fact.

I probably stress the battery more than most people, since I drive across the Sierra Nevada every week or two. (From sea level to ~8500 ft on a fairly winding mountain road.) Without you-know-what (which I just installed last week) it's difficult to keep assist from coming on in the climb. When the pack drains to a certain point, which for me has gotten to be only about 1/4 down, it gets the negative recal where indicated charge drops to zero.

This may also throw an IMA code or codes, which may disable the IMA. I just reset them.

Charging continues at a normal "battery empty" rate (which means you have the engine doing 4-5 bars of charge while trying to climb a 6% or so grade) until it gets about 1/4 full, then it quickly jumps back to full charge indicated.

It's been doing this for over a year - about as long as I've done the drive. It also went quite rapidly from a state where the recals would happen about 1/4 from empty. The codes seem to be related to pack temperature. They virtually never happen when I drive up the local mountains, though I sometimes get recals then.

[Armin]

Hi Nemystic!

I'm glad you are reviving the recal discussion on a more technical level. Let me chime in with my recent thoughts...

I've been experiencing pretty extreme recals lately: I installed the dfiore battery pack (~150kmiles and sold to me after it was removed for code of death). So it was already "bad" when I started using it. I drove another 30k miles with it, with >300 charge cycles using the grid charger. I would grid charge to full with the car off, then allow a positive recal to occur by just turning the ignition on for two minutes before starting the engine. I am aware that this was probably not ideal for the battery, but the goal was not to extend it's life, but rather to study its dying!

In the end, negative recals happened around the 3/4 level on the SoC gauge. If I allowed it to recharge using the IMA, it would positive recal once the SoC got up to 5 or 6 bars, followed by the IMA light and code of death.

I haven't measured the capacity after finally removing that battery, but I'd guess it's very low now.

Originally Posted by nemystic

The immediate cause of a recal is assumed to be a recognized difference between the calculated and actual SoC. It’s also assumed that this condition is detected by voltage and/or temperature inconsistencies within the battery pack.

This is exactly what I've been pondering with this worn-out battery. I totally agree on the part of difference between the calculated and actual SoC. But how does it determine the actual SoC???

I figured, since it recals easily and often on my bad battery, I should be able to observe what triggers the recal. I built a panel with 10 voltmeters to monitor all subpack-voltages using Mike Dabrowskis pogo-pin connector to connect to the BCM. I also had meters for battery current and total voltage. I couldn't monitor temperatures, but there were many recals in situations where the battery was cold (shortly after starting after long rest).

I was unable to find any pattern in the voltages causing the recal. It is not triggered by differences between sub-packs! That battery had two packs with much higher internal resistance then the rest, causing them to be lower in voltage during assist and higher during charge. However, the behavior of the IMA was not as expected: rather than triggering recals, it limited charge / assist current to keep this difference under 1V. Most times, this meant no more than about 15A in either direction.

This current-limiting mechanism was somewhat slow, and if I floored the accelerator (or the joystick on the you-know-what) quickly, I could get >50A assist for a few seconds, with one of the bad subpacks dropping to <8V while the others where still above 12V. However, this never triggered a recal.

Another symptom I observed was that the DC-DC converter would turn off any time the total battery voltage dropped below 100V for more than a second or so. This could happen if I floored it quickly and the current limiting mechanism didn't catch me fast enough. The only way to restart the DC-DC was to turn off the ignition. This, too, didn't trigger recals.

On the flipside, I could park the car in auto-idle-stop and allow the battery to drain to a recal, while the sub-pack voltages are all within 0.2V of each other and the total pack voltage never drops below 140V.

The only hypothesis I have left is that Honda uses a complex set of parameters based on current, voltage and temperature over time to trigger recals. I.e. If we have 50A assist when the battery is at 50C and the voltage on one module drops more than 500mV/min... Unfortunately, I don't have the instrumentation required to gather enough data to find such an algorithm.

I did, however, develop a subjective "feel" for the true state of charge from looking at the meters. Usually, the recals occured when I expected them from watching the voltages and current. But it's all very fuzzy and I could never pinpoint a certain set of conitions that caused it.

[tuna1]

Thanks Armin for your results.Very interesting results!

Thanks
Ronny

0

[nemystic]

Thanks James and Armin for the comprehensive information.

Last year, in an experiment, the battery pack from a wrecked Insight was found to have widely variable low voltages on each respective sub-pack. This was likely because it had been inactive for a long time.
I installed this pack and recovered the SoC to a useful level. I don't recall the specific details, but the SoC indicator was showing very unusual patterns, like nothing I'd ever seen before. It was as though the SoC algorithms "learn" the characteristics of a battery pack over multiple charge / discharge cycles, and use that for the SoC determination and display. Since this "other" pack didn't behave like the "familiar" one, the algorithm may have been confused.

When the pack from the wrecked Insight was later removed, the voltages of each respective sub-pack were found to be consistent, well within 1%.

After re-installation of the original pack, the behavior and SoC indication was the same as before it had been removed.

I wonder if "overcharging" by an external source (grid), or otherwise, has a cell balancing effect by restoring full charge to the weaker cell(s), while potentially damaging the stronger cells through overheating and venting?

While I'd suggest discretion in content, to preclude this thread from becoming locked, I'll hope to see more feedback on positive recals.

-Brian

[james]

Well, here's some more. Did the trip back from the Bay Area last night, the first time with you-know-what installed, and also the first time after a rear-ender that smashed shut one of the rear vents.

I think I had a total of 4 recals on the trip. I would gradually work the battery down to its recal point, generally using less assist than the norm. After the indicator dropped to zero, I'd keep on trying assist until it wouldn't give any more, then let it do background charge. Each time it did a positive recal: slow charge to about 1/4, then a quick jump to full charge indicated.

I really don't understand what could be triggering the recals. I originally thought it was a low voltage, as they would happen with relatively high assist (which is all the normal system gives). With YKW I used less assist, and the recals happened at about the same indicated charge level.

On a related problem, has anyone come up with a good way to clear the IMA light while driving? I hate having to stop and fumble with the fuse.

[uhtrinity]

On my mini vacation this summer I also noticed and started to get an understanding of what might be going on.

Going back to when I bought the car in the late spring, if I drained the IMA battery on a hill it would slowly move down to roughly 10 bars, then dive to two and do a forced recharge. I have done this maybe three times prior to the trip two weeks ago.

On the trip I crossed quite a few passes and drained the battery to zero twice, and fairly low another 1/2 dozen times.

During one of the full drains I had what i think is a positive recal. The battery went all the way to 2 bars, but I held the accerlerator which forced assist unti there were no bars, when it had charged back up to about 16 bars it then jumped all the way to 20. During this time I was still able to force a charge as I was on a grade and could just tap the brake lightly. After stopping to take pics however it wouldn't charge any more.

The next time I did a full drain it went to about 6 - 7 bars before diving to zero. But this time it charged all the way to 20 without a positive recal, and I again charged using the brake pedal every chance I got even though the meter was pegged at 20 bars. I did this from the last pass for roughly 15 miles of 5 - 8% grade down into Vernal, UT. I didn't have a chance to drain the battery fully again, but I was able to take it down to 5 - 6 bars without a negative recal, and seemed to be able to use assist slightly longer.

What appears to be happening:

The computer estimates where empty and full are, and by completely draining the battery you can force it to lower the empty point, and by forcing more charge you can reset what it sees as full. This seems to fit what I observed and what was mentioned in previous posts. Of course bad batteries throw a bunch of other variables into the equation.

[Mike Dabrowski 2000]

uhtrinitys results are similar to what happened with Ed Zandees silver insight.
My theory is that the SOC and limits are actually two different pieces of software. the SOC is strictly based on the colulmb counting, and the limits are based on possibly a voltage over time determination.
Set the ABC on you know what to about 12-20A, and let it continue to charge after it reaches full on the SOC. Based on our experiment on the return from Madison, I was able to continue to charge the pack for the approximate equivalent of 5-6 bars after I passed the "full" mark. Once that started to be limited (MAX volt detect?) assist was able to be drawn well past the recal point, even off the bottom of the scale. It seemed that before this exercize, the SOC guage bottom end had been shifted up by 6 bars, but the normal SOC based upper limit was not allowing regen once the skewed SOC scale reached the top. The fact that we were able to continue charging past the top bar would seem to indicate that the pack was never getting fully charged when under the SOC based upper limit.
My pack will limit regen at the top of the SOC guage, and will allow assist to about 2 bars from the bottom where it gradually fades. Both ends basically follow the stock IMA response even with you know what asking for more. The fact that the regen was not limited when we did the pack whack, until well past the top bar, and the subsequent complete fix of the recal on the next power up, seem to indicate hat the SOC and the limits now aggree as they do on my pack. Ed has reported that he still has not experienced a recal since the trip.
I have been saying that my pack never has recals, and It has not. I will many times leave the ABC running with 5-12A, and forget about it. I will look up and see the SOC right to the top, and the abc is in limit. So I may have been whacking pack all this time without realizing it, and therefore the lack of recals?
James , you may want to try the ABC background charge past the top, and see if it changes anything. Not sure how a positive recal would fit into this theory, but it may indicate a different pack condition.
If we keep poking at this we may finally get a better understanding.

[james]

"The fact that the regen was not limited when we did the pack whack, until well past the top bar..."

I think I've seen something like this on mine for as long as I can remember. Say I start with a partly discharged pack at the top of a long steep downgrade. It will recharge going down, with 4-6 bars of charge showing without throttle or brake applied. Braking will progressively increase the charging up to all bars lit. This continues until the battery gauge shows full. At that point charging will stop for some seconds, then it goes into what I think of as "topping up" mode, with 4-5 bars of charging that isn't increased by braking. That continues for a while, then all charging stops.

I do need a long downgrade to see this: 10 miles or so at 5-6%.

[nemystic]

Thanks for the additional interesting information James.
It's clear that the battery capacity of the Insight doesn't match well with mountain pass terrain, and the stress on the battery pack is evident.

What I've found is that while the Insight's stock assist indicator accurately displays the outgoing current, the regen indicator does not.
I monitor the current (as a voltage from the hall effect sensor) and have observed that the regen indicator can show a high level of input (such as when braking) sometimes when there's actually none (except the hidden charge that's not measured by the hall effect sensor). The magnitude indicated is often misleading as well.

It will be interesting to find out if the "topping up" charge you observed on the display is for real or not.

Keep us all posted!