Honda Insight Forum banner
761 - 780 of 799 Posts

·
Premium Member
Joined
·
2,865 Posts
You have some really well balanced cells... my pack gets 6 mV out of balance after a couple weeks of hard driving.

...

Here's my initial testing on @thibble's PCB:
-Summary: there's nothing wrong with the hardware... at least it's working the way I expect it to.
-My hunch is still that 12V_IGNITION or 12V_IMA is stuck on in @thibble's car.
Couldn't you just set an emergency limit in LiBCM that no matter what the cars current state is, that at some low arbitrary voltage LiBCM force shuts down to prevent catastrophic damage to the cells? This is probably way too simple of a solution because I don't understand anything but. Meh.
 

·
Registered
Joined
·
238 Posts
I noticed the default setting in the config file was for a 40% current hack. In the installation video it says not to change the config file if you are not using current hack. Wouldn't someone need to change the setting for a 0% hack?
 

·
Premium Member
Joined
·
1,342 Posts
I noticed the default setting in the config file was for a 40% current hack. In the installation video it says not to change the config file if you are not using current hack. Wouldn't someone need to change the setting for a 0% hack?
When Mudder made the original videos, the firmware was shipping with the current hack disabled. However, since then, he found out that something like 99% of LiBCM users are using the current hack and so he changed the firmware defaults to enable the 40% hack, but he hasn't redone the video. So, yes... if you are not using the current hack, you would need to change the setting to disable it.
 
  • Like
Reactions: mudder

·
Linsight Designer
Joined
·
3,904 Posts
Discussion Starter · #765 ·
Couldn't you just set an emergency limit in LiBCM that no matter what the cars current state is, that at some low arbitrary voltage LiBCM force shuts down to prevent catastrophic damage to the cells? This is probably way too simple of a solution because I don't understand anything but. Meh.
This code already exists, but doesn't work when the key is ON, as LiBCM hardware cannot turn itself off when the 12V_IMA rail is powered (i.e. the key is ON). The 12V_IMA rail is specifically designed to turn LiBCM back on the next time you turn the key on.

I noticed the default setting in the config file was for a 40% current hack. In the installation video it says not to change the config file if you are not using current hack. Wouldn't someone need to change the setting for a 0% hack?
See the errata below that video:
Update: Since nearly every Open Beta customer has installed the +40% current hack, the default current hack setting in config.h is now set to 40%. Therefore, if you don't have the +40% current hack PCB installed inside the MCM, then you will need to manually select the "0% current hack" option inside config.h
As @joeaax1j mentioned, I will eventually roll all the beta changes back into the final videos... for now, I've logged all changes alongside the videos.

It's balancing.
I'm not sure that's actually the case.
Unless I'm looking at the pictures out of order, last night the delta was 126 mV, whereas today it's 263 mV. So it looks like it's unbalancing.

Plug a laptop into the USB port, open an Arduino Serial Monitor, type $DISP=CELL, then turn the key on. Post the individual cell data so I can see what's going on.


Nevermind... I was confusing your picture with @FlyingHood395 's (from last night).
 

·
Registered
Joined
·
238 Posts
I'm not sure that's actually the case.
Unless I'm looking at the pictures out of order, last night the delta was 126 mV, whereas today it's 263 mV. So it looks like it's unbalancing.

You must be thinking of @FlyingHood395 as I only posted the one picture.

When I started it up last night, the Delta was 369 mV, now the timer says 47480 (13.1h) and the Delta is 231 mV. There was a point where I did see the Delta increase for a few minutes before coming back down. Earlier this morning it went from 260 to 265 and then back down.

I will still check out the individual cell data for the fun of it. Thanks for sharing the procedure.

Also, I hope that it's okay that I'm grid charging outside of the car. The outlets that I can reach outside turn off during the day. I have the 90 watt charger.
 

·
Registered
2003 Honda Insight
Joined
·
43 Posts
Quick update before I drive home to see my family:
I grid charged until the delta was .009, then I installed into the car. My first driving impression is HOLY MONSTER. Very quick! Keep in mind, I've never had a working battery, so this feels like a game changer to me. One thing I've noticed on the readout screen, the delta jumped up very quickly, now at .120 or so, but moves around quickly. SoC went from 70 to 40 in about 4 miles of city driving (edit: I was not remembering to regen brake, so it did eventually go back to ~SoC60 and a ~d .007). I'm thinking when I get home I'll get the data that mudder asked about in post #752 and then grid charge it for more time.
 

·
Premium Member
Joined
·
2,865 Posts
Quick update before I drive home to see my family:
I grid charged until the delta was .009, then I installed into the car. My first driving impression is HOLY MONSTER. Very quick! Keep in mind, I've never had a working battery, so this feels like a game changer to me. One thing I've noticed on the readout screen, the delta jumped up very quickly, now at .120 or so, but moves around quickly. SoC went from 70 to 40 in about 4 miles of city driving (edit: I was not remembering to regen brake, so it did eventually go back to ~SoC60 and a ~d .007). I'm thinking when I get home I'll get the data that mudder asked about in post #752 and then grid charge it for more time.
Keep in mind if you park the car with more than 40% SoC, it will auto balance the rest out. My Citrus was at 0.015 when I took it off the charger to take it to work, and it was 0.001 when I came out to go home.

I think it's above 40% SoC...
 

·
Linsight Designer
Joined
·
3,904 Posts
Discussion Starter · #769 ·
There was a point where I did see the Delta increase for a few minutes before coming back down. Earlier this morning it went from 260 to 265 and then back down.
FYI: The cell voltage delta can increase during the bulk charge phase... and will then drop back down as the most full cells continue to discharge. As long as the trend is downward, local maxima along the way are expected.

Also, I hope that it's okay that I'm grid charging outside of the car. The outlets that I can reach outside turn off during the day. I have the 90 watt charger.
Absolutely ok to grid charge/balance outside the car.
 

·
Linsight Designer
Joined
·
3,904 Posts
Discussion Starter · #770 ·
Quick update before I drive home to see my family:
I grid charged until the delta was .009, then I installed into the car. My first driving impression is HOLY MONSTER. Very quick! Keep in mind, I've never had a working battery, so this feels like a game changer to me.
Great to hear. Ha, going from no IMA to LiBCM with +40% current hack is about as far a swing as you can go.

One thing I've noticed on the readout screen, the delta jumped up very quickly, now at .120 or so, but moves around quickly.
There's a lot of 'noise' in the measurements during heavy regen/assist. When the IMA power is close to zero, the cell voltages will stop jumping around.

SoC went from 70 to 40 in about 4 miles of city driving (edit: I was not remembering to regen brake, so it did eventually go back to ~SoC60 and a ~d .007). I'm thinking when I get home I'll get the data that mudder asked about in post #752 and then grid charge it for more time.
If the pack is initially balanced (e.g. 0.001 mV delta @ 75% SoC), and then you assist until the pack is mostly empty (e.g. 30%), then seeing a large 'resting' cell delta indicates that the different cell types do in fact have different capacities. For example, let's say one cell type is 5 Ah, whereas the other cell type is 4 Ah... then if we start from 'full' (let's just assume 100% SoC in this example), and then remove the same energy from each cell (e.g. 3 Ah), then we'll find that the 5 Ah cells will be at 40% SoC (3 Ah / 5 Ah), whereas the 4 Ah cells will be at 25% SoC (3 Ah / 4 Ah). If we then lookup the Voc SoC value (i.e. we calculate SoC based on resting cell voltage), then we'll see that the 5 Ah cells are resting at 3.665 volts (40% SoC), whereas the 4 Ah cells are resting at 3.570 volts.

Therefore, if you see a large resting delta voltage develop as the pack is discharged, then that means the cell types do in fact have different capacities. Note that if this is the case, then you will see the resting delta voltage decrease as the cells recharge back towards full.

So if you have different cell capacities, then you'll want to change the parameter CELL_BALANCE_MIN_SoC to '70', so that cell balancing only occurs when the pack is full... otherwise LiBCM will constantly try to 'balance' your cells, whereas they're already balanced (based on the cell full condition).

...

FYI: I will state any action items more succinctly once I review your data.
 

·
Linsight Designer
Joined
·
3,904 Posts
Discussion Starter · #771 · (Edited)
Keep in mind if you park the car with more than 40% SoC, it will auto balance the rest out. My Citrus was at 0.015 when I took it off the charger to take it to work, and it was 0.001 when I came out to go home.

I think it's above 40% SoC...
Yes, the default CELL_BALANCE_MIN_SoC value right now is 40% SoC... but I will be increasing that soon, as it tends to cause LiBCM to hyperactively balance the cells. I've explained the reasoning before (and you can deduce it from what I just wrote in post #770)... but basically until I've written a per-cell battery model, balancing should only occur only when the cells are close to full.

FYI: There's no harm in keeping the value at 40%... it just means LiBCM will spend more time than it needs to balancing cells out.

Related: I will also be increasing the cell balance setpoint in a future version... it turns out each LTC6804's reference accuracy isn't good enough across temperature... so for example if one LTC6804's reference varies from 0 to +1 mV off, whereas another varies from 0 to -1 mV off, then LiBCM will also spend too much time balancing cells. There's really no need to balance cells to within 1 mV... I'll ultimately change this value to twice the LTC6804's specified reference accuracy (probably a few mV... haven't done the math yet).

Edit:
The changes mentioned above are in v0.7.8.
 

·
Linsight Designer
Joined
·
3,904 Posts
Discussion Starter · #774 · (Edited)
I've just pulled the 'prerelease' firmware branch back into 'main'.
No worries if this doesn't mean anything to you... it just means beta testers who weren't specifically using the 'prerelease' branch will now get all the changes I've made since 2022APR08. We're now all running the same code again, until I make more changes to 'prerelease'...
...starting tomorrow! I'm finally back to full time LiBCM firmware development. Boy I've been busy these past several months.

FYI: I've increased the target minimum cell delta voltage to 2.2 mV, so don't expect LiBCM to continue to get the delta down to 1 mV... it'll be more like 3 mV now.
 

·
Registered
2003 Honda Insight
Joined
·
43 Posts
Once the pack gets all balanced out... maybe late tomorrow:
-Go for a long assist-heavy drive (e.g. enable calpod when you would normally regen). Get the pack all the way down to maybe 30% SoC.
-Turn the key off and wait ten minutes.
-Plug a laptop into LiBCM's USB cable.
-Open an Arduino Serial Window.
-Type $DISP=CELL.
-Turn the key ON (but not started).
-LiBCM will start spitting out all the cell voltages. Log these in a text file and post it here.

What we're looking for is that the cells in all three modules are at the same resting voltage, which indicates the cells have the same ampacity. It's not the end of the world if they aren't the same, but it'll be nice to know.
I finally got around to getting the pack to a lower SoC (which felt somewhat abusive to not regen). When I parked, the SoC read at ~28. After 10-15 I read the data with the Arduino $DISP=CELL. WOW! Not great at all! And SoC dropped to concerning levels but still above the 3.100 danger level printed on the readout screen.

Also, I've noticed from driving the past day that there seems to be a dependance with the delta reading and SoC levels. Somewhere around SoC50 to SoC70 the delta gets small, like less than .010. But lower SoC the delta increases to high values.

Pack (FrankenPack):
Bay 1: 12S+ from Bumblebee
Bay 2: 18S- from Bumblebee
Bay 3: 18S+ Blue Energy Module



temp:37
Old SoC: 5%, New SoC:5%
IC0,3.3227,3.3204,3.3235,3.3204,3.3199,3.3193,3.3210,3.3208,3.3216,3.3225,3.3230,3.3191
IC1,3.3222,3.3236,3.3238,3.3244,3.3236,3.3239,3.6006,3.5981,3.5995,3.5993,3.5994,3.5981
IC2,3.5976,3.5994,3.5991,3.5988,3.5974,3.5931,3.5971,3.5987,3.5973,3.5993,3.5975,3.5995
IC3,3.5933,3.5962,3.5930,3.5955,3.5948,3.5965,3.5962,3.5956,3.5958,3.5931,3.5943,3.5940
Discharging cells above 4.2000 V (0x): 0,0,0,0,

f the pack is initially balanced (e.g. 0.001 mV delta @ 75% SoC), and then you assist until the pack is mostly empty (e.g. 30%), then seeing a large 'resting' cell delta indicates that the different cell types do in fact have different capacities. For example, let's say one cell type is 5 Ah, whereas the other cell type is 4 Ah... then if we start from 'full' (let's just assume 100% SoC in this example), and then remove the same energy from each cell (e.g. 3 Ah), then we'll find that the 5 Ah cells will be at 40% SoC (3 Ah / 5 Ah), whereas the 4 Ah cells will be at 25% SoC (3 Ah / 4 Ah). If we then lookup the Voc SoC value (i.e. we calculate SoC based on resting cell voltage), then we'll see that the 5 Ah cells are resting at 3.665 volts (40% SoC), whereas the 4 Ah cells are resting at 3.570 volts.

Therefore, if you see a large resting delta voltage develop as the pack is discharged, then that means the cell types do in fact have different capacities. Note that if this is the case, then you will see the resting delta voltage decrease as the cells recharge back towards full.

So if you have different cell capacities, then you'll want to change the parameter CELL_BALANCE_MIN_SoC to '70', so that cell balancing only occurs when the pack is full... otherwise LiBCM will constantly try to 'balance' your cells, whereas they're already balanced (based on the cell full condition).
I will change this in the morning, clearly the capcities are very different.
 

·
Linsight Designer
Joined
·
3,904 Posts
Discussion Starter · #776 ·
Looks like the Blue Energy 18S module has quite a bit less capacity than the 5AhG3 module.

I haven't characterized the Blue Energy cell yet, but if it's similar to the 5AhG3 cell, then:
-The Blue Energy 18S module's cells are at ~3.32 Voc, which is ~6% SoC.
-The 5AhG3 18S and 12S modules' cells are at ~3.59 Voc, which is ~28% SoC.

The SoC delta (between the two cell types) is 22% SoC, which is 1.1 Ah (5 Ah * 0.22).
This means the Blue Energy cells are more like 3.9 Ah (whereas the 5AhG3 cells are more like 5.0 Ah).
That's a pretty big difference.

You can still use LiBCM with this Frankenstein pack, but long term you should probably switch to the same cell type in all three modules. For the time being, I recommend changing STACK_mAh_NOM from '5000' to '3900' (i.e. 3.9 Ah); this is a line of code in the file 'config.h'. This will prevent LiBCM from over-discharging your pack.

Eventually LiBCM will do this for you automatically, but I haven't written the code for that yet.

...

I thought someone was sending me a Blue Energy cell/module... they even sent me a text saying they were at the post office right then... but it never showed up. So if someone wants to send me a Blue Energy cell/module, I will characterize it. That could, for example, let us see if the charge/discharge curve operates at a slightly lower voltage. If that were the case, then it's likely we could extract more energy from the cell without damaging it (e.g. maybe get up to 4.5 Ah, instead of 3.9 Ah).
 

·
Administrator
Joined
·
13,900 Posts
It's always usually a bad idea mixing packs. I would swap to all one or the other asap.
Ideally you want the BMS operating (balancing) as little as possible.

Even packs of the same type with the same spec/manufacturer can be problematic if they come from different vehicles with different usage profiles and ages etc etc :(
 

·
Registered
2003 Honda Insight
Joined
·
43 Posts
Looks like the Blue Energy 18S module has quite a bit less capacity than the 5AhG3 module.



I haven't characterized the Blue Energy cell yet, but if it's similar to the 5AhG3 cell, then:

-The Blue Energy 18S module's cells are at ~3.32 Voc, which is ~6% SoC.

-The 5AhG3 18S and 12S modules' cells are at ~3.59 Voc, which is ~28% SoC.



The SoC delta (between the two cell types) is 22% SoC, which is 1.1 Ah (5 Ah * 0.22).

This means the Blue Energy cells are more like 3.9 Ah (whereas the 5AhG3 cells are more like 5.0 Ah).

That's a pretty big difference.



You can still use LiBCM with this Frankenstein pack, but long term you should probably switch to the same cell type in all three modules. For the time being, I recommend changing STACK_mAh_NOM from '5000' to '3900' (i.e. 3.9 Ah); this is a line of code in the file 'config.h'. This will prevent LiBCM from over-discharging your pack.



Eventually LiBCM will do this for you automatically, but I haven't written the code for that yet.



...



I thought someone was sending me a Blue Energy cell/module... they even sent me a text saying they were at the post office right then... but it never showed up. So if someone wants to send me a Blue Energy cell/module, I will characterize it. That could, for example, let us see if the charge/discharge curve operates at a slightly lower voltage. If that were the case, then it's likely we could extract more energy from the cell without damaging it (e.g. maybe get up to 4.5 Ah, instead of 3.9 Ah).
One thing that @Battery Buyer mentioned that stood out to me:
"Panasonic Insight cells are 3.7 volts nominal.
Blue Energy EHW5 and EHW5B cells are 3.6 volts nominal."

So based on that quote and my experience, it does seem necessary to fully understand these cells (voltage and capacity) and to use them homogeneously (i.e. all 48 cells made from Blue Energy).

Unfortunately, there's only been that one sale with the 18S modules, and no 12S. @Recovering_Gasaholic has a thread where he is exploring converting the 36S module to 12S, which seems like it'll take awhile of problem solving and learning until a homogeneous pack is available.
 

·
Registered
Joined
·
238 Posts
I just purchased some ehw5 modules. I was going to have the seller one ship to you but then said that one just arrived from another member. Anyway, I guess I could ship you one in a couple days. Let me know if one shows up before then.


I thought someone was sending me a Blue Energy cell/module... they even sent me a text saying they were at the post office right then... but it never showed up. So if someone wants to send me a Blue Energy cell/module, I will characterize it. That could, for example, let us see if the charge/discharge curve operates at a slightly lower voltage. If that were the case, then it's likely we could extract more energy from the cell without damaging it (e.g. maybe get up to 4.5 Ah, instead of 3.9 Ah).
 

·
Linsight Designer
Joined
·
3,904 Posts
Discussion Starter · #780 ·
I was going to have the seller one ship to you but then said that one just arrived from another member.
I agree I said that... the box I thought was an EHW5* module turned out to be something else.

Anyway, I guess I could ship you one in a couple days. Let me know if one shows up before then.
Is this an "extra" module (that I can test destructively)? Or is it one you plan to use (that I can characterize resting SoC from and then send back)?
 
761 - 780 of 799 Posts
Top