Honda Insight Forum banner
661 - 680 of 796 Posts

·
Registered
Joined
·
134 Posts
I am posting my first observations after some months spent with LiBCM.

As a whole, the system works pretty flawlessly. Consistent power delivery and better capacity, or so it seems. I started with a healthy NiMH pack, so it wasn't as clear a boost as those who went from flailing packs to this. The biggest difference is in the current hack. I upgraded because I was tired of the grid charging and super low ultra deep discharging that I did every few months, thought I almost never had an IMA light related to degradation. I guess there are three main small issues that I experience from time to time; two of them seem intermittent. Some of these may be due to the fact that I don't have OBDIIC&C (thereby IMAC&C) or a calpod switch.

1.) Limited current hack - Sometimes when flooring the car in second or third gear, the ~20 kW of power will drop to ~11kW within a second or a few seconds from initiating. Even with accelerator pedal floored, the car will cut down assist to normal levels. The time that it decides to cut it out varies, and sometimes I won't get it at all. I will have to lift off the accelerator then re-floor it to get the extra boost from the current hack. I am not sure if the system will limit the current for safety reason, but my guess is this is some intermittent connection issue with the current hack PCB. Maybe my solder joints...

2.) Regen cutoff at lower limit - It seems that sometimes the system will reapply max regen at the lower RPM cut-off in each gear right during clutch pedal depression. Seems a lack of hysteresis as the car will reapply full regen right after disabling it due to low RPM cut off. It's as if the RPM goes up just enough after regen is stopped to make the car think it can regen again. By then the clutch pedal is depressed, and things are off. The issue is that this creates quite a jerk at the cutoff point during deceleration. It doesn't happen all the time, but the thing I've learned to do is just depress the clutch sooner before it shuts regen off.

3.) SOC - I know this has been hashed out before, but even after tweaking the voltage offset, I can't get the car to consistently let me regen at high SOCs. For example, when the SOC is at 80% after charing overnight, I can't regen in the morning until I drain the pack to the mid 70%s, then I have the capability to regen past the 80% mark I charged to in the morning. Seem like a weird resolution issue, but the lack of regen in the higher SOC range makes it difficult to retain a SOC high enough to keep out of forced regen territory. I am pretty good at keep a full pack, but when it won't let me regen, it makes it difficult to keep it from always draining.

Once again, I apologize for the SOC related comments as I know this has been discussed on end before. As for the other points, if these have been already discussed and I have missed that, please forgive me.

Thanks for all the hard work on this, John. It's nice to not have to grid-charge or do ultra low and slow discharges to keep the car happy anymore.
 

·
Linsight Designer
Joined
·
3,891 Posts
Thanks for the detailed feedback!
1) This is expected behavior, due to the ECM throttling back power after a few seconds. You need MIMA or IMAC&C P&P to have continuous 20+ kW assist.

2) I'm not sure I understand the behavior you're describing. It doesn't help that I almost never downshift while coming to a stop. Can you elaborate more?

I have on occasion felt 'bucking' regen while on a gentle downhill slope. I suspect this is due to how LiBCM spoofs the back voltage. Specifically, as regen current increases, LiBCM spoofs a proportionally higher pack voltage. I believe that when the pack is full, the MCM sees this higher (spoofed) pack voltage as an "ok NiMH battery is full" and thus reduces regen. The lower regen then causes LiBCM to decrease the spoofed voltage again, which causes the loop to repeat again.

Peter has previously pointed out this issue, and has proposed alternate strategies. I haven't looked at it yet, but it's on my TODO list.

3) LiBCM won't let you regen above 85%... even getting from 80% to 85% is pretty tough, as the MCM doesn't really want to regen when the battery is that full. One solution is to (further) spoof the SoC value LiBCM sends the MCM, which would allow regen all the way up to 100% SoC. However, battery lifetime suffers dramatically once you get above 85% SoC (we're talking hundreds of cycles versus tens of thousands). I am being VERY conservative with the SoC range during the open beta period, since the goal is that these batteries should retain 80% of initial energy storage after 10-15 years.

Also, if you want to manually assist and regen all the way to the safe operating limits, you need to use MIMA or IMAC&C. Otherwise, you're stuck at the mercy of whatever SoC the ECM thinks is best.

...

I agree it is wonderful to finally just use the IMA system as I please, without worrying about all the unknowns in batteryland.
 

·
Premium Member
Joined
·
2,864 Posts
Even with a full pack(78%) I have never seen more than 17KW of assist. Even commanding it with Peter's slider pot mod.


Anyway, I'd like to note my LiBCM is at 26 hours, and I'm just posting this so when I come to complain Monday morning because I forget to do it tomorrow, everyone can laugh at me.
 

·
Administrator
Joined
·
13,900 Posts
Even with a full pack(78%) I have never seen more than 17KW of assist. Even commanding it with Peter's slider pot mod.
CVT or Manual?

Maybe also be due to MCM firmware variations.
 

·
Registered
Joined
·
5 Posts
Hi from Morecambe in the UK,

I re-flashed my 2001 CVT JDM G1 with the 7.04 firmware yesterday and all remains good for me.

Very little to report aside from the fact that I am very happy with the conversion and performance of LiBCM.

The battery can seem pretty quick to run down sometimes but it is also pretty quick to regen so not a big deal.

Cheers, Steve V
 

·
Premium Member
Joined
·
2,864 Posts
CVT or Manual?

Maybe also be due to MCM firmware variations.
The turbo CVT car.

Actually this makes me think. How many amps of assist should I be seeing with a full pack and giving it the beans? The most I have ever seen is 103.
 

·
Administrator
Joined
·
13,900 Posts
The G1 IMA motor drive gives peak torque/power/current at about 2000-2200 rpm.

CVT owners likely won't generally see as high peak current numbers on the display as manual cars because the rpm rises so quickly that back emf etc from the motor reduces the current before it has time to register.

A manual car in a high gear takes a long time to gain rpm so it can sit in the sweet spot long enough for the high currents to be detected and displayed on the screen.

Also the CVT MCM does have slightly different IMA programming to the MT version.
Try an MT MCM and see if that makes any difference, you won't hurt the car.
 

·
Premium Member
00 MT
Joined
·
48 Posts
2.) Regen cutoff at lower limit - It seems that sometimes the system will reapply max regen at the lower RPM cut-off in each gear right during clutch pedal depression. Seems a lack of hysteresis as the car will reapply full regen right after disabling it due to low RPM cut off. It's as if the RPM goes up just enough after regen is stopped to make the car think it can regen again. By then the clutch pedal is depressed, and things are off. The issue is that this creates quite a jerk at the cutoff point during deceleration. It doesn't happen all the time, but the thing I've learned to do is just depress the clutch sooner before it shuts regen off.
40% current hack definitely makes things a little odd if you tend to regen all the way down to the cutoff even without LiBCM. There's enough time between beginning to press the clutch and regen turning all the way off that it'll take the engine from 1000rpm to 0 if you don't time things perfectly (particularly in second gear), or will let the rpm bounce back up. Before the current hack I used to downshift a gear or two and regen, then downshift again and regen to a stop. With the current hack, I'll only drop down to 3rd gear which still gives a ton of braking if you need it, and it will very cleanly transition to autostop at 20mph without stalling or causing any issues.
 

·
Registered
Joined
·
134 Posts
2) I'm not sure I understand the behavior you're describing. It doesn't help that I almost never downshift while coming to a stop. Can you elaborate more?
Sorry for the lack of clarity. It is a bit strange to describe. I think weldoak describes it well in post #671. However, in my case, I don't necessarily have to be in second, nor do I have to downshift during regen to get this to happen. Matter of fact, I don't typically downshift. I just lightly depress the brake in the given gear I am in to activate full regen which I will cary to the lower RPM cut-off for that gear. It is at that point right when I am depressing the clutch and the car shuts off regen that the car seems to reapply full regen right after it cut it off causing a massive jerk, or a near engine stall if the clutch is depressed enough to remove the rest of the drivetrain from the equation. I hope that provides a little more clarity.
 

·
Linsight Designer
Joined
·
3,891 Posts
Discussion Starter · #673 ·
I just lightly depress the brake in the given gear I am in to activate full regen which I will cary to the lower RPM cut-off for that gear. It is at that point right when I am depressing the clutch and the car shuts off regen that the car seems to reapply full regen right after it cut it off causing a massive jerk, or a near engine stall if the clutch is depressed enough to remove the rest of the drivetrain from the equation.
I wasn't able to reproduce this in my car, but let's see if changing the following config.h parameters makes it go away:
1) Open config.h
2) Uncomment the line //#define VOLTAGE_SPOOFING_DISABLE (i.e. remove the '//').
3) Comment the line #define VOLTAGE_SPOOFING_ASSIST_AND_REGEN (i.e. add '//' to the beginning.
4) Upload the code
5) See if the behavior persists

FYI: Some people get an immediate CEL when the above steps are performed. If that happens to you, then bail back to the initial settings. Either way, report your results.

...

A second test (to perform, separately):
1) Set the current hack PCB to 0% (all switches off).
2) Uncomment the line //#define SET_CURRENT_HACK_00 in config.h
3) Comment the line #define SET_CURRENT_HACK_40 in config.h
4) Upload the code
5) See if the behavior persists
 

·
Linsight Designer
Joined
·
3,891 Posts
Discussion Starter · #674 ·
Update on @BLS ' LiBCM CEL:
Tonight we spent a couple hours troubleshooting the system. We tested many low hanging fruit issues and everything looks good. Note that I sent @BLS both a known-good MCM and a known-good LiBCM... and the issue persists. I also sent him a bunch of test equipment, which we're putting to good use. I think I've got @BLS hooked on Anderson connectors ;).

We then started to get into some 'spicy' high voltage debugging, at which point I accidentally shorted an alligator clip across the HVDC contactor... the QTY3 1750 uF capacitors inside the PDU charged really quick across the bridge! That's why there's a pre-charge resistor.

We decided to call it a night, so that I could come up with a safer test procedure (which I have now done). Turns out it's going to be much safer to just hook up that programmable power supply I sent him.

Remember to always wear your PPE!

We're going to reconvene tomorrow night.
 

·
Linsight Designer
Joined
·
3,891 Posts
Discussion Starter · #675 ·
I'm fairly certain we figured out @BLS ' issue. I believe the issue is a poor mechanical connection inside the OEM BCM temperature sensor wire harness. At this moment BLS is deadbugging a wire that bypasses this connector.

Prior to figuring out the above, I recorded this rambly video:
Key points:
-I found another bug in the voltage spoofing firmware. Honestly I'm amazed the MCM is ok with it... but since it's not causing any problems, I don't have any plans to immediately update the firmware... I've added the fix to my backlog.
-After I ship these QTY29 LiBCM units I'm building right now, I don't anticipate building more LiBCM units until I can finalize the v1.0.0 firmware (which will remove the 40 day update requirement).
-If you want an LiBCM unit before September(?), you should order one soon... because as soon as these QTY29 units are gone, that'll be it for a while.
-After I build these units, my plan is to focus on firmware for the next couple months.
 

·
Premium Member
Joined
·
2,864 Posts
I'm fairly certain we figured out @BLS ' issue. I believe the issue is a poor mechanical connection inside the OEM BCM temperature sensor wire harness. At this moment BLS is deadbugging a wire that bypasses this connector.

Prior to figuring out the above, I recorded this rambly video:
Key points:
-I found another bug in the voltage spoofing firmware. Honestly I'm amazed the MCM is ok with it... but since it's not causing any problems, I don't have any plans to immediately update the firmware... I've added the fix to my backlog.
-After I ship these QTY29 LiBCM units I'm building right now, I don't anticipate building more LiBCM units until I can finalize the v1.0.0 firmware (which will remove the 40 day update requirement).
-If you want an LiBCM unit before September(?), you should order one soon... because as soon as these QTY29 units are gone, that'll be it for a while.
-After I build these units, my plan is to focus on firmware for the next couple months.
Wait I can buy one right now? I didn't know they were available to buy yet. I seriously NEED one, but I'll probably end up buying 2.

MUDDER I HAVE ALREADY BUILT ONE WHY DO I HAVE TO FILL OUT THIS QUESTIONARE AGAIN REEEEEE

[Edit]: that's it I'm trolling the QUESTIONARE
 

·
Premium Member
Joined
·
2,864 Posts
Ok real talk mudder, I do have an issue. I don't want fast chargers for both of these kits, just one. How do I separate that out?

[Edit]: I filled it out anyway and specified rapid charger, we can sort out the details later.
 

·
Linsight Designer
Joined
·
3,891 Posts
Discussion Starter · #678 ·
Once you've filled out the form at least once, you can just PM me with all future orders (for your own personal use).
That form is a legal shield to prevent your heirs from suing me if you win a Darwin Award.
 

·
Registered
2000 Insight (Red) Manual AC 150kmiles 59.8mpg for now....
Joined
·
30 Posts
We are getting there!!! The direct connection from B22 to the orange wire on the mcm seems to have worked. Installed the battery and flipped the switch, key on and the IMA light did not come on. CEL was still on on the dash and about 20 sec after turning the key I heard a loud "Click/ snap" from the back and the IMA light came back on. Turned the key off and when turning it back on I immediately get the click again. Got a P1445 code and P1576, The P1576 code will clear but the 1445 remains. If I cycle the IMA switch It will do the same process of no IMA light for 20sec then "click" and an IMA light.
Vertebrate Cartoon Organism Mammal Gesture

Also thank you so much Mudder for taking time out of your week to help me troubleshoot. It was a great experience I learned a great deal and it has renewed my confidence that it will work.
 

·
Linsight Designer
Joined
·
3,891 Posts
Discussion Starter · #680 ·
P1445 indicates the HVDC voltage on the PDU differs too much from the battery HVDC voltage. Verify you plugged both junction board relays back in (we unplugged them during troubleshooting). Also verify the VPIN_in (orange wire with white anderson) and VPIN_out (black wire with orange anderson) are outputting the correct voltage:
VPIN_in: should equal the actual pack voltage - as displayed by LiBCM - divided by 53.
VPIN_out: should equal the spoofed pack voltage - as displayed in parenthesis by LiBCM - divided by 53.

Report back with your findings.

FYI: The service limit for P1445 is huge: 37 volts or more for at least 2 seconds. So even with your 6% uncertainty DMM you should be able to measure within those limits ;).

...

We can conference again whenever is convenient for you.
 
661 - 680 of 796 Posts
Top