Honda Insight Forum banner

221 - 240 of 268 Posts

·
Registered
Joined
·
7,165 Posts
Discussion Starter #221
Interesting work. Your first batch of charts suggests to me that a 72 cell LTO pack (Fit LTO) would be just about perfect within the stock management window. Most of the values are within the 2.2 to 2.4V range, while the high and low values are sparse, plus low would be higher and high lower with an LTO pack that has at least half the resistance of the NiMH one...

Here's another view, of each stick pair voltage, measured about 14 hours after the car was last driven.... Anyone wish to guess which are the strong and weak stick pairs in this pack - the ones that are the most imbalanced under load - based on this measurement?
What do you mean by 'most imbalanced'? Cells within the stick pair are most imbalanced? Or taps that are most imbalanced from other taps? Imbalance in charge state/capacity or some other performance metric?

Also, are we seeing loaded voltages, resting voltages, what load, etc.?

Sounds like you're asking us to guess which one 'sags' or 'peaks' the most under discharge and charge loads. If that's the case it's really impossible to say, though in general, within the middle-charge state voltage range, the lowest voltage taps tend to be the weakest. So I'd guess tap 9, maybe 10...
 

·
Registered
Joined
·
889 Posts
Your first batch of charts suggests to me that a 72 cell LTO pack (Fit LTO) would be just about perfect within the stock management window. Most of the values are within the 2.2 to 2.4V range, while the high and low values are sparse
That's what I thought at first, but I thought about it some more and expanded the data at the edges. It's not really sparse in the real world. It amounted to only maybe a minute or two of my drive, but over a year this could be many events.

So yeah, this data is NOT a green light for safe LTO operation in my opinion. It needs more testing before any statement can be made about LTO suitability.

I don't recall anyone putting an LTO pack through the full range of driving conditions and posting results, like how much time is spent at a particular voltage, as I tried to do with this data. I'm building my own LTO solution, and do not want my investment to get anywhere near the knees of the LTO charge curve. This data says, "if your LTO pack has the characteristics of a NiMH installation, that is a possibility." We know that an LTO pack has different qualities than NiMH, but unless someone has done this testing, we can't conclude that this doesn't apply to LTO.

I'm just hoping that this data prods folks to include data logging in their LTO builds - perhaps, with the amount that microSD cards can hold, and the fact that you can add SD card logging so cheaply to a project - so that we can get a lot of data on these things and know in confidence what will happen in the corner cases.

an LTO pack that has at least half the resistance of the NiMH one...
which to me means that the car may just run the pack down more until it does reach that voltage.

The second post is a different question. A lot of time is spent looking at and talking about tap voltages. Here are a bunch. What do these tell us? (People are frequently told to measure the tap voltages. What should they conclude from them? From these in particular?) :)

I wanted to post this before I posted the results because I am no longer sure what value tap voltages measured in the driveway provide.
 

·
Registered
Joined
·
7,165 Posts
Discussion Starter #223 (Edited)
I don't see a problem. Definitely no problem at the top: Your chart looks like it doesn't even show a blip above 2.7V, while Fit LTOs are supposed to be able to float-charge at 2.7V, with minimal degradation... And actually, the same could be said about the bottom - 1.5V being the recommended minimum for Fit LTOs, where your chart doesn't even show a blip below that...

True we're looking at tap voltages here, so we assume cell balance - and if they weren't balanced you could have at least relatively more serious 'transgressions' (unless 'slope detection' plays a role, too). But it seems like picking nits, even to me, someone who's usually nit picky...

More data would be good regardless; but I don't think it's necessary to prove viability.

Sean's chart:




which to me means that the car may just run the pack down more until it does reach that voltage.
Not sure what you mean. If the LTO has half the resistance, won't it just lower the high voltages under load, and raise the low voltages under load? The pack will be considered 'empty' or 'full' at the normal empty and full thresholds/metrics/algorithmic junctures... The question is, simply, do those match well with the LTO or not?

At the top end, there's at least a few management thresholds that can/will mitigate propensity toward and/or prevent over charging. Ultimately, you got a 19.2V tap voltage max under load, and something like a 17.4V resting voltage, which will trigger regen throttling. hmm, just thought of something: If you're running 72 cells you're not using taps for management. You should just run 70 cells, 7 cells per tap and use the tap management...

That would give you a max voltage of 7 X 2.7V=18.9V and a minimum of 7 X 1.5V=10.5V (or 7 X 2V seems more realistic, better, at 14V).

Most NiMH tap operation is between about - call it 1.2 X12 and 1.5 X12, or 14.4V to 18V - which translates to 2.06V to 2.57V for your LTOs at 7 cells per tap... Sounds pretty good to me.

Let's see, 17.4V resting: 17.4V / 7 cells = 2.486V - your Fit LTOs in this 7 cell tap config. would charge no higher than about 2.486V per cell. That sounds great based on what I just looked at the other day, in the context of insightbuyer's 12V LTO schtick...

To get any more detailed we'd need to dig-in to the stock 'management thresholds' in a finer way, not something I want to get into at the moment...

edit: OK, I'm ready to add at least a couple more things here.

There's two major low-end voltage thresholds in stock form: 12V and 13.2V.

12V I think is the rockbottom low-end voltage the BCM (or MCM) will allow, analogous to the top-end 19.2V. If you go below 12V at tap the MCM will quickly throttle assist, bringing tap voltage initially back up to about 14.4V and then easing-in assist to uphold a voltage of at minimum about 13.2V, or 1.1V per NiMH cell on average. There's finer details here that I'm not sure of, but we don't need to get into that now.

With 7 Fit LTOs, a rockbottom 12V tap would be 1.714V X 7. 13.2V would be 1.886V. I think the 1.886V value is the more important one - it's the threshold that comes into play more often and over longer sustained durations.

So, theoretically, the stock BCM would allow these voltages. The question/s though is how often or likely would you even see those and/or at what load would you see those.

If you had a totally stock system, the current limits would come into play before you actually dropped tap voltages that low, at least until a tap were near depletion... I can't remember exactly what kind of voltage drop LTO users have been seeing in their builds, but I recall it's been stellar, as in very little drop, even at hacked high currents... hmm, I scrawled a little figure on my desk at some point: "106V at 121A." I believe that was one value Peter P. measured with his small LTO pack. I think that was 48 cells, so a nominal of 110V.

So, you're not gonna see much drop under load, at stock limited currents, but even at higher hacked currents. i.e. these low voltage thresholds, 12V and 13.2V, aren't even critically bad for the LTOs, not being lower than 1.5V at cell-level, yet chances are you won't even get near them at least most of the time...

The next step would be to talk about the near empty threshold: if during most usage when cells are charged they won't get close to these low voltage thresholds, what about near empty? That's gotta be another post at some later date. In general, the nominal state of charge determines a lot of things, so it partially depends on what if anything you're doing with nominal state of charge. And, as far as I can tell, the low-end value that's analogous to the high-end 17.4V resting voltage, is around 14.4V: If a tap drop to 14.4V at rest or low current, the BCM calls it empty. Before that happens, though, it can be called empty if a cell drops out - steep slope happens.

14.4V / 7 cells = 2.06V. Not sure what the rebound resting voltage is for the Fit LTOs when at or near depletion... Off the top of my head that sounds a little low, like we'd prefer it to be a little higher, but I don't know. end edit

A lot of time is spent looking at and talking about tap voltages. Here are a bunch. What do these tell us? (People are frequently told to measure the tap voltages. What should they conclude from them? From these in particular?) I wanted to post this before I posted the results because I am no longer sure what value tap voltages measured in the driveway provide. [emphasis added]
They often don't tell you anything. Your ~200mV range probably reflects some fairly minor charge-level difference, and a propensity for your taps to drift over time. Maybe. Not much else. Even a ~100mV difference can be nothing or it can be a single cell almost empty - so no problem or a serious, major problem. Which just shows how potentially worthless voltage tap measurements can be.

You only get anywhere when you start looking at change over time, under load like you're doing, etc. If voltages are near perfectly balanced - your pack is probably fine, great. If voltages are seriously varied, like maybe more than 300mV, it likely reflects a serious, outlying performer... And these kinds of judgments need context, background info, to actually draw conclusions... Like if a person says, 'I got an IMA light', 'my charge gauge dropped to the bottom', etc., and person measures taps and finds one is like half a volt lower than the others, there you go, prime culprit.
 

·
Registered
Joined
·
36 Posts
I thought I'd communicate on this thread as it seems timely for my process now. Hooked up the OBDIIC&C and the Prolong harness and verified things. Now I want to measure tap voltages. Where is the process for measuring this? Sorry for the simple question here, but can I just reverse probe the plug for each pair to ground to measure? I'll post the data here and then start the UDD process for the first 5 pairs
 

·
Registered
Joined
·
7,165 Posts
Discussion Starter #225
^ I think everything's answered or linked in the original thread you started, like in the following linked post or the one below it: Getting up to speed (literally!)

You should continue your inquiry in that thread...
 

·
Registered
Joined
·
7,165 Posts
Discussion Starter #226 (Edited)
Sean, if you want to do more logging and graphing, I think it'd be really useful to do some discrete logs and graphs. For instance, instead of logging and graphing over varied usage over a long period of time, break your datasets into 'max discharge load', 'max charge load' and the like. One dataset per event.

Perhaps you hit a hill and start logging (or note the time) and try to hold assist steady - that'd be one dataset. A second one might be several bursts of full throttle full assist. Here I think you'd need to have some method to mark your datatsets, like mark when the full assist starts and ends, same for the next full assist event, etc., so that you can then collate the data into 'full assist events' and take an average or something.

I think the most telling datasets would be charge (regen) near full and discharge (assist) near empty... Both of these I guess at the highest powers you can obtain.

Do you have an OBDIIC&C? If so here's a couple other datasets that'd be really useful:

-positive recal under a consistent, background charge current. The only way I know of to do this easily enough is to start with a near car-full pack, such as one that has recently pos recal-ed, use some assist to bring charge down a bit, then reset SoC to 60%. That will invoke a background charge of about 6-7 amps at normal cruising and the BCM should pos recal somewhere between 60% and 70%. This will show you what tap is triggering your pos recal - which is charged most, or at least has the highest voltage under charge load.

-Do the same - only this time it's neg recal. Drain the pack until neg recal, allow pack to charge up some, then drain again under as controlled a current as you can manage, until neg recal again, This will show you which tap is causing the neg recal, as well as suggest the threshold/s in play.
 

·
Registered
Joined
·
889 Posts
Not sure what you mean. If the LTO has half the resistance, won't it just lower the high voltages under load, and raise the low voltages under load?
I don't think so. Doesn't the BCM look at voltage? It doesn't care about current and can't measure resistance.

At the top end, there's at least a few management thresholds that can/will mitigate propensity toward and/or prevent over charging. Ultimately, you got a 19.2V tap voltage max under load
Yes. This is exactly what I'm concerned about and why I'm bothering with collecting data.

19.2V tap voltage * 10 = 192V pack voltage / 72 = 2.67 V. That's at the end of the charge curve without load; under load you need to define how much of a load. Here's my data, charging from 0V (yes, that's how the cell arrived) to 2.95V:

87989


By my calculation, if someone is not regularly balancing their LTO cells and one cell is allowed to drift 5% higher in SOC than the others (ie, it reaches 20AH while the others are still at 19.5AH) then by the time the rest reach 20 AH the voltage of the leader will be past 3.5V - damage zone.

We don't have any real data, I think, to know the rate at which an LTO pack that is not being balanced sees voltages drift apart, if at all. But lack of data doesn't mean it can't happen, thus my concern (and a call to implement always-on data logging in LTO installations.)

Here's a different view of the NiMH samples from the earlier drive test (rescaled by multiplying each stick pair sample by 10 and then dividing by 72) in bins that better expresses the concern that's driving me to suggest better testing is needed:

87993


I chose the bins based on the following:
  • Below 2.0V - from the start of the knee to zero. Risk of cell reversal to a cell having the least stored charge in a pack whose balance is not maintained
  • 2 - 2.1V - a buffer from the start of the knee. Provides a comfortable buffer which may increase equivalent number of cells.
  • 2.1 - 2.275V - lower half of usable capacity
  • 2.275-2.45V- upper half of usable capacity (using the voltage Peter has set his passive balancer to as the upper limit)
  • 2.45 - 2.6V - a buffer from the passive balancer voltage to the start of the charging curve knee
  • Above 2.6V - all samples above 2.6 volts; risk of damage to a cell that has the most stored charge in a pack whose balance is not maintained
I used a logarithmic scale for number of samples because the point is to avoid the red zones. When I build my LTO pack I want it to stay well between the knees of the charge curve at ALL times irregardless of current.

So the reason I'm saying a better test is needed is because this is against NiMH cells, not LTO cells. Also, this is against only one version of BCM, A01. Not sure how an 030 or 070 would behave or different versions of ECU and MCM. These have changed over the years by Honda to address battery warranty issues.

(I'll talk about the Android phone screenshot - tap voltages - in another post.)

Enjoying the tech dialog! :)
 

·
Registered
Joined
·
889 Posts
I think the most telling datasets would be charge (regen) near full and discharge (assist) near empty... Both of these I guess at the highest powers you can obtain.
The last chart does that in a way. I'm emphasizing those zones where we typically go with a low pack under assist and a high pack under regen.

I have been noodling on what I need to log with a high rate IMA data logger for a long time. (Been noodling on the equivalent for the main ECU too.) The data one can pull through the data link connector (the OBD2 port) is too slow to be correlate to certain events like when you first demand heavy assist. Any thoughts about which MCM and BCM signals I should log are welcome. This high rate battery data is really nice, so I'm targeting maybe 50 ms per data set. Will need to come up with something other than a spreadsheet to massage it.
 

·
Registered
Joined
·
7,165 Posts
Discussion Starter #229 (Edited)
^ I'll have to think about that later.

Re. your previous post - how to keep this short and sweet?

I think we're missing each other on the resistance thing. The lower resistance of the LTO pack would mean this: all the driving you did and logging of loaded voltages with the NiMH pack would be different - the highs wouldn't be so high, the lows wouldn't be so low. By a large amount. The example I posted about Peter's 48 cell LTO pack is a case in point: he had a nominal pack of 110 volts - and saw a voltage of 106 volts at a massive 121 amp current! Only a 4 volt drop.

121 amps is higher than the stock system will allow, so if one were to stick to stock, there's even less chance of overshooting your cells' min and max voltage thresholds.

Now, this covers like 95% of driving, so the remaining issue is the 5% where you might be near empty or full. That involves questions about how the BCM deals with empty and full and perhaps what you choose to do about it. It also has to do with cell balance - like you've pointed out. I agree with your general thresholds, like 2V and 2.6V, and I agree with your assessment of the impact of cell imbalance, in the sense that, yeah, if cells are imbalanced you can't rely on tap voltage thresholds to save you...

I think you're probably overestimating the risk of damage from short duration incursions above and below 2V and 2.6V. And I think you're overestimating the potential frequency of those incursions...

In a broad brush view, I definitely agree with your assessment that balance matters and that, if there were no slope detection, there'd be a risk of overcharge/overdischarge if cells within the 'management unit' are imbalanced, even by a small amount. I'm pretty sure there's slope detection at the bottom, and I think there might be slope detection at the top, though I'm less convinced of that. The key thing at the top would probably be resting voltage - 17.4V at tap, 2.486V at 7 cells per tap, 2.417V if 72 cells total. So if that's all there is to 'protect' the top, the question is, What kind of headroom do these resting voltages mean for your cells - what state of charge do these resting voltages represent? 100% minus that equals headroom... In your graph it actually doesn't look like there's a ton of headroom above these voltages - 10%, 20%?* - so it wouldn't take much charge imbalance for a cell to breach 2.6V early - all else being equal, in a hands-off approach...

I'll have to leave it at that for now.

* I had put 5-10, but it looks more like 10-20, which is actually a decent amount of headroom...
 

·
Administrator
Joined
·
12,434 Posts
I have been noodling on what I need to log with a high rate IMA data logger for a long time. (Been noodling on the equivalent for the main ECU too.) The data one can pull through the data link connector (the OBD2 port) is too slow to be correlate to certain events like when you first demand heavy assist. Any thoughts about which MCM and BCM signals I should log are welcome. This high rate battery data is really nice, so I'm targeting maybe 50 ms per data set. Will need to come up with something other than a spreadsheet to massage it.
There are various ways to do this but it depends what data you want of course.
Getting the individual tap voltages requires some sort of isolated system.
You might already have something suitable?

If basic IMA data like pack voltage, pack temperature, pack current , pack soc and BCM flags are ok then the BCM outputs a constant serial stream of such data at 60ms intervals on the BATTSCI line that can easily be logged directly into excel using a BCM Gauge a serial/usb adapter and PLX-DAQ software. We have been doing that for at least ten years.

Or you could install your own device inside the BCM that connects to the isolated ADC inputs from the taps. Again we did this ten+ years ago or more, mine was called the BCM Tap Voltage Monitor or something.. It did what the BCM Gauge above did but also collected tap voltages.

Massive amounts of details, schematics, firmware for all these devices are on here.

If you want instant data even faster from multiple system then you have to pick it up yourself from the actual sensors or inputs which requires more sophisticated and complicated electronics. Again that's all doable but you would need to design yourself a bespoke system or use off the shelf expensive stuff with suitable ratings and comm etc etc.
 

·
Registered
Joined
·
889 Posts
@retepsnikrep Thanks for the tips. BATTSCI and METSCI should be good enough. The voltage data is coming from the BCM's ADC inputs.
 

·
Registered
Joined
·
337 Posts
Looking back over my charts of tap voltages, I see that I've driven with some pretty extreme tap voltage imbalances. I can now conclude that tap voltage imbalance alone is not an issue for the BCM. You can drive with major tap voltage differences and the BCM will not bat an eye. OK, maybe it bats an eye, but it doesn't trigger any codes or the like.
When you say "...it doesn't trigger any codes or the like.", may not be exactly correct? The reason I'm wondering about this is because Topt installed resistors on the HV battery taps to send a perfectly balanced battery signal to the BCM and successfully prevented the IMA light from coming on. See: Using a resistor matrix to simulate a balanced pack with...

Based on his results, it would seem that the tap voltages do come into play with regard to triggering codes. I installed the resistors and had similar results. My 2001 CVT would eventually throw a code, but my 2000 MT hasn't so far.

The CVT is the car I'm monitoring the real tap voltages on a separate display as I drive, while the now added resistor matrix sends a perfectly balanced pack signal to the BCM. For me, without the added matrix on the CVT car I would get an IMA code much sooner than with the matrix, so tap imbalances can surely come into play with regard to triggering the IMA light.
 

·
Registered
Joined
·
7,165 Posts
Discussion Starter #234 (Edited)
^ I think your's and Topt's success probably has to do with avoiding pre-mature empty or full signals, not avoiding imbalance per se. Or perhaps avoiding one of the two more extreme imbalances I mentioned in that post (pasted below) - which have more to do with real, underlying problems.

I'd have to know what IMA codes you got before you installed the 'resistor matrix', to tell what's really going on...

Oh, also, I would really recommend against such a matrix - unless you're willing to let your IMA usage devour individual cells over time. I.e. if the pack's expendable - go for it.

But if you really want a high quality, high functioning pack, this is definitely not the solution. The codes/problems you're likely avoiding are extreme problems. You're basically letting usage drive individual cells over a cliff... It might be viable if your usage were monitored closely and you took manual management steps. Otherwise, it's only a matter of time, or perhaps sheer luck, until one cell after another gets destroyed.

...more extreme voltage differences can matter. For example, we know that one of the trouble codes will trigger if a tap's voltage deviates from the others by more than 1.2V for something like 24 seconds or longer within a narrow usage current range (like between -10 amps and +20 amps). But that would signal a major major cell failure. It looks like anything less than that, though, doesn't really matter, the BCM won't make a fuss...

There's also another really extreme voltage tap deviation that will trigger a code, a P1568-66. I forgot that I probably got that once due to an extreme voltage deviation. The DTC sheet says a difference of 4V for 2 seconds or more. One time, I started a drive with a couple taps that were deeply discharged. I had grid charged for about 5% worth of charge, but I think that was probably not enough to bring the cells up to the normal operational 'voltage plateau'. Upon assist, it's possible most taps were around say 15V while these deeply discharged ones momentarily fell to about 11V, or just below 1V per cell, which can be typical under the right circumstances. So, the difference in tap voltage would have been more than 4V.
 

·
Registered
Joined
·
7,165 Posts
Discussion Starter #235 (Edited)
Just wanted to at least scrawl a quick working summary of the BCM/MCM management thresholds mentioned in previous posts, plus others. It can be hard to keep them all straight...

Starting from top-end working our way down the charge state range. Voltages are tap voltages, some are estimates or approximations:

19.2V - max voltage allowed, absolute regen limiting

~17.4V resting - regen limiting invoked, time limits and usage behavior come into play. For example, the first time 17.4V is measured, regen limiting will be enabled and last for some timed duration, or until you use assist and drop voltage.

~17.4V under load or ~17V resting for timed duration, or perhaps steep slope detection or perhaps amp-hour counting - postive recal, nominal state of charge jumps to 75% or 81%, pack is considered mostly full or full.

This is a tough one, for me. Usually I see pos recals when total voltage is around 174-176V, and when my cells have been known to be well balanced I can use that total as a proxy for tap voltage - so 17.4V to 17.6V, and this is at about a 6-7 amp charge current. The threshold seems to be adjusted for resistance, so it's higher at higher currents. Either that or there's always a measurement at a resting state - and that resting voltage is around 17V. So, the BCM considers the pack 'full' or at 75% when a tap's resting voltage hits 17V and stays there for at least some seconds, it seems. It can't just hit 17V and then say 5 seconds later fall to 16.9, 16.8, it has to stick for like, maybe 15 seconds...

72% nominal - charge state value will stay at 72% until the BCM sees one of the pos recal thresholds, that is, if you don't discharge with assist first. Similar behavior happens at the bottom, at 28%.

70.2% nominal state of charge - when background charge is in play, charge state is taken up to a nominal 70.2% and background charge is disabled.

~65% - if you have headlights ON, background charge always seems to kick in at about 65% nominal charge state. Headlights OFF, it doesn't.

~52% nominal - green CRHG bars tend to disappear if in forced charge mode, even though the charge is still taking place. That forced charge becomes a 'background' charge.

~38% nominal - assist throttling automatically kicks-in, much harder to invoke assist, takes more throttle and you get less of it.

Note that 'nominal' percentage movement is simply 65mAh per percentage point in the 28% to 72% nominal charge state range. There's a slight difference with BCMs that pos recal straight to ~81% - because those BCMs also make a weird jump and weird drop at the top; I can't remember exactly what it is, I think they drop from 72% straight to 66%, or maybe 75% to 69% - something like that.

28% nominal - IF you can actually drag your pack this low, the nominal will hang there until one of the 'empty' thresholds is breached, while you continue to discharge/use assist.

Steep slope detection - although I haven't measured this exactly, directly, rough measurements, proxies, context, and logic lead me to believe there has to be a tap voltage slope calculation/detection in the BCM. When a cell is nearing empty, its voltage starts to plummet fast. Even in a string of 12 cells, you'd be able to discern a cell nearing empty by the steepness of the tap voltage slope. If this is detected by the BCM, the pack is considered empty and forced charging is enabled. Nominal charge state drops to 25%.

~14.4V resting - If by some miracle your cells are so matched and balanced that a tap voltage can reach 14.4V resting, the pack is considered empty.

~14V - pack will be disabled via main contactor, I think, if a tap's voltage drops to 14V at very low current. DCDC is disabled.

~13.2V - generally the minimum sustained voltage allowed under assist load.

12V - absolute minimum voltage under assist load. As soon as 12V is measured, assist is throttled rapidly to bring voltage back up to about 14.4V. After that, if assist continues, voltage will be throttled every time it dips to 13.2V. Initially, perhaps at higher charge states, the 'throttle' voltage is a bit higher than 13.2V. Some of this is a bit variable, I think, as it depends on how much assist you're requesting. For example, the above scenario would be more 'text-book' for a full throttle full assist, which can last up to about 4 seconds. But if you invoke full throttle full assist in say 4th or 5th gear on the MT transmission, the BCM/MCM seems to allow minimum voltages below 13.2V but higher than 12V...

20% nominal - charge state you see when you do a reset, such as pulling fuse.

17.6% nominal - absolute minimum, such as if you drain your pack and don't hit any of the other thresholds first. I've never seen a value below about 17.6%.


I'll probably try to fill this out more as time goes by, perhaps add some key NiMH characteristics/voltages that likely inform some of these management thresholds.
 

·
Registered
Joined
·
889 Posts
^^^^^ sticky candidate

- just realized that this may need to be the first post of new thread to be a sticky. I think it's DEFINITELY worth it.
 

·
Registered
Joined
·
7,165 Posts
Discussion Starter #237
^ I don't think anyone 'does' stickies anymore. I've suggested much better things (of others) in the past and no one seems to pay attention.

I came back here to post this link to another one of my threads that has quite a lot of detailed reasoning and observations about stuff in the previous post. I re-read most of it and still find most of it true, sound. I had thought I needed to write some stuff - but most of it's already written: Contemporary musing on Insight pack and management

I'm really trying to hone-in on my "moving window theory of charge state" idea. The more I experiment, the more I believe it has to be true. But I also want to understand how it's happening, at a deeper level... My brain is all mucked-up when it comes to really, deeply conceptualizing charge state and related things. It's like I have some core, fundamental flaws in the way I frame the problem, so I can't find the right answers.

For example, a few concepts, issues: charging in terms of 'amps'/current, and 'capacity' vs. charge state.

I'm so stuck on the idea of charging in terms of 'putting in amp-hours' - that I think I miss the real picture like 65% of the time. Charging and discharging seem like complicated things when I really think about them...

Here's a concrete example from an experiment I did a while back. I wanted to gauge the impact of temperature on charge/discharge efficiency. Before the test I imagined the cell charged cold would 'put out' less amp-hours than the one charged warm. But when all was said and done, there was no difference - in total amp-hours. There was however a difference in the shape of the voltage curves: the cold cell had a higher voltage charge curve and a lower voltage discharge curve.

The cold cell was charged to a lower 'charge state', despite 'taking in' and 'putting out' the same number of amp-hours as the warm cell... When it comes to Insight NiMH cells, with massive voltage hysteresis, and possibly this 'moving window' charge state, it becomes really difficult to measure the charge state. Or to put it another way, even if I track amp-hours in and out, I really don't know what 'charge state' my cells are at - this is especially true as time passes from a known charge state point, and because, like I said, voltages can be all over the place at different true charge states.
 

·
Registered
Joined
·
889 Posts
I'm really trying to hone-in on my "moving window theory of charge state" idea. The more I experiment, the more I believe it has to be true. But I also want to understand how it's happening, at a deeper level...
The answers are in what is going on electromechanically inside the cell. It would be valuable to learn this stuff; we should start a new thread on it
 

·
Registered
Joined
·
7,165 Posts
Discussion Starter #239
"electromechanically"?, not chemically? Why do you think it'd be mechanical and not chemical? I hardly know what electromechanical is, or would be...

As far as starting a new thread - I don't think it's worth it. Too many of those types of threads already, no one really cares. If you have things you want to say, explore, though, just do it here... I'd have to think about it more just to be able to better describe the idea. Part of my problem is that I've never had a new pack, so I don't really know how new cells behave, hard to tell what's likely to be an 'old, used cell' phenomenon and what's likely to be common to the cells and/or chemistry in general.

I think my core question is something like this: I can get like-new performance (or at least performance that easily meets the car's max demands) with my 18 year-old cells. BUT, I can only get that within maybe a 45% charge state window. It doesn't matter where that window is - it can be low or high along an absolute amp-hour capacity range of 0-6500mAh. I just can't get it along the whole range within one discharge or charge. So, in essence, my cells seem to suffer 'capacity' degradation, just that it's nothing like the degradation most people have written about around here, or that I've imagined, or actually graphed when working with sticks and cells.

'We all' focus almost exclusively on 'amp-hour capacity'. It wasn't until not terribly long ago (3 years? 4 years?) that I started concentrating on voltage, the 'loft' and 'sag' in charge and discharge curves, etc., and now it's really starting to gel how 'capacity' degradation can look very different in real-world usage than I had imagined...

Anyhow, the 'moving window' theory probably applies to old, used cells - a description in part of how our cells actually degrade over time and usage. I can charge and discharge my sticks and see fairly normal, full-length charge and discharge curves - at currents under 20 amps. But these cells see a lot more than that in the car. When subjected to 'a lot more' in the car, they simply can't handle it across the whole range.

I recall new 'BetterBattery' cell graphs a long time ago that showed the cells capable of discharging at maybe 80 amps across the whole capacity range, with voltage remaining above about... well, I don't recall the voltage, but it was consistent across the whole range, more or less. It was probably above 1V, at least. I'm sure if I subjected my cells to an 80 amp discharge they would not uphold that current across the whole (absolute) capacity range...

It's still somewhat phenomenal to my mind that my cells can achieve full output in these ~45% range windows. That's probably not something most used cells do. Mine do it probably because of the various reconditioning and usage schemes I've tried. Most used packs probably end up with imbalanced cells and get driven with imbalanced cells until the packs work no longer - until P1449-78 codes. And maybe then 'we' subject them to band-aid fixes, like full grid charges. But the cells will never be the same again, they'll never perform 'like-new'. For my pack I've basically nipped the imbalance in the bud and tracked performance, adjusted when necessary, and exercised the cells across the whole range, including very deeply, the 'low-end structure'. It can't or hasn't restored all 'capacity'; but it has reconditioned and maintained that ability for max performance with the 'capacity' it has, those 45% windows...
 

·
Registered
Joined
·
337 Posts
I think your's and Topt's success probably has to do with avoiding pre-mature empty or full signals, not avoiding imbalance per se. Or perhaps avoiding one of the two more extreme imbalances I mentioned in that post (pasted below) - which have more to do with real, underlying problems
I'm not sure what you mean about "pre-mature empty or full signals". Wouldn't this imply an imbalance, if it's at the tap level? The total pack voltage the car sees is the same, with or without the matrix.

The code was always a P1449 on the MT, but I didn't get the blink code. My CVT doesn't have an OBDIIC&C, so I don't have the IMA code for it.

Also, I agree that this is not a healthy solution for any pack. I've started the install on my Escape parallel packs, and just did the matrix as an experiment, and for the convenience of not throwing an IMA code and stopping all assist while driving. Both packs, the MT and the CVT, are expendable at this point. Also, as Peter pointed out, this shouldn't be attempted unless the OEM PCT strips are in place on each cell and hooked up and working. A cell could get really hot and potentially cause a fire!

Honda says an imbalance of 0.61v should throw a code, but while observing the cell taps as I drive, I never see that much of an imbalance. The CVT eventually would trigger the IMA after driving several days with the matrix in place, so I suspect total pack capacity degeneration is the issue for that. Curiously, Topt says his MT hasn't thrown a code since installing the matrix, and I haven't either on the MT, and my MT pack is very low capacity. Right now I'm attempting a pack rejuvenation with my simple cycler on the CVT.

Please understand I'm not trying to be argumentative, but rather just contributing to the knowledge base. I'm also trying to sort out what levels of tap voltage monitoring I need to set on my Escape's second pack (the one in parallel), which is one of the reasons I installed the matrix. I currently have it set for an imbalance of 2.5%, which is a little more conservative than Honda's 0.61v tap criteria.

I for one certainly appreciate all the work you've done and taken the time to share it with the community. Kudos to you my friend!!
 
221 - 240 of 268 Posts
Top