Honda Insight Forum banner
601 - 620 of 674 Posts

·
Premium Member
Joined
·
1,324 Posts
I encountered some interesting LiBCM behavior last night. Made a 380 mile round trip from my home to the Chicago airport, which is the longest trip I've done with LiBCM. First, a little background, I've got one of the MCMs that is causing background charging when driving with Calpod enabled. Mudder is aware of this issue and his short term solution has resolved that, but the short term solution is causing background discharging when Calpod is enabled. It is on his list of things to fix.

I drove to the Chicago airport with Calpod enabled. The LiBCM SoC was slowly dropping, as expected. After the SoC dropped to around 50 (on the 4x20 display), as I would encounter a longer downhill stretch, I would occasionally turn off Calpod and let LiBCM charge back up a bit, adding a few percentage points back to the SoC. Driving back from the airport, I did the same thing. Overall, however, my SoC was still dropping because I was primarily driving with Calpod enabled. Finally, when I got back to my home town, I pulled in to my gas station to fill up. At this point, my SoC was showing 35%. (I'll add here that my OEM dash SoC has always shown "full" ever since installing LiBCM.) I filled up my tank, and then I turned the car back on and as I drove away I noticed two things:

1. The 4x20 display bumped the SoC from 35% to 45%.
2. The OEM dash SoC had dropped from full to about half-way.

Any idea what happened there? Unfortunately I wasn't doing any debug logging on the trip so I don't have anything to upload.

I got home and plugged in my grid charger and this morning the 4x20 shows 73% which is typical for me. My OEM dash SoC is still pegged around 50%. I know that the OEM dash SoC doesn't work yet so I'm not bothered by that, but I did find it curious that it "moved" last night at the gas station for the first time since my install back in February. I suppose that it is related to the 4x20 SoC jump, but not sure.

-Bryan
 

·
Premium Member
Joined
·
2,684 Posts
I encountered some interesting LiBCM behavior last night. Made a 380 mile round trip from my home to the Chicago airport, which is the longest trip I've done with LiBCM. First, a little background, I've got one of the MCMs that is causing background charging when driving with Calpod enabled. Mudder is aware of this issue and his short term solution has resolved that, but the short term solution is causing background discharging when Calpod is enabled. It is on his list of things to fix.

I drove to the Chicago airport with Calpod enabled. The LiBCM SoC was slowly dropping, as expected. After the SoC dropped to around 50 (on the 4x20 display), as I would encounter a longer downhill stretch, I would occasionally turn off Calpod and let LiBCM charge back up a bit, adding a few percentage points back to the SoC. Driving back from the airport, I did the same thing. Overall, however, my SoC was still dropping because I was primarily driving with Calpod enabled. Finally, when I got back to my home town, I pulled in to my gas station to fill up. At this point, my SoC was showing 35%. (I'll add here that my OEM dash SoC has always shown "full" ever since installing LiBCM.) I filled up my tank, and then I turned the car back on and as I drove away I noticed two things:

1. The 4x20 display bumped the SoC from 35% to 45%.
2. The OEM dash SoC had dropped from full to about half-way.

Any idea what happened there? Unfortunately I wasn't doing any debug logging on the trip so I don't have anything to upload.

I got home and plugged in my grid charger and this morning the 4x20 shows 73% which is typical for me. My OEM dash SoC is still pegged around 50%. I know that the OEM dash SoC doesn't work yet so I'm not bothered by that, but I did find it curious that it "moved" last night at the gas station for the first time since my install back in February. I suppose that it is related to the 4x20 SoC jump, but not sure.

-Bryan
Funny you say this. On my drive home yesterday, I took the SoC of my LiBCM down to 14%. My gauge also read halfway for the first time ever. Also when I got home and keyed off, it said 14, then when I plugged the car in and the 4x20 display lit up, it said 21%.

The SoC change is due to pack voltage settling to a higher voltage at rest. The gauge itself is just wonky, that's the nature of the beast.
 
  • Like
Reactions: joeaax1j

·
Linsight Designer
Joined
·
3,611 Posts
Discussion Starter · #604 ·
I pulled in to my gas station to fill up. At this point, my SoC was showing 35%. (I'll add here that my OEM dash SoC has always shown "full" ever since installing LiBCM.) I filled up my tank, and then I turned the car back on and as I drove away I noticed two things:
1. The 4x20 display bumped the SoC from 35% to 45%.
When the key is turned off for more than 10 minutes, LiBCM updates the SoC based on the resting cell voltage. LiBCM can't do this when the car is on, because the resting cell voltage condition never occurs, hence the voltage cannot be used to determine SoC. So when the car is on, LiBCM counts the cumulative current into - and out of - the battery. This cumulative counting over time is a mathematical integration. You may remember that integrating adds a 'c' constant... in this case, that constant is comprised of three things:
A) accumulated ADC sample error caused by incorrectly counting the net charge that has flowed into - or out of - the battery, and/or;
B) battery degradation, and/or;
C) SoC misestimation due to cell imbalance, and/or;
D) your cells aren't reaching their resting voltage within the ten minute cooldown period (each time you turn the key off).

More details:
A) Any offset on the current sensor measurements will cause a slow integration error. For example, if the current sensor is off by +200 mA (it thinks there's more assist than regen), then the battery SoC will fall 4% per hour driven (200/5000*100). To minimize offset error, LiBCM re-calibrates the 'zero-current' offset value each time the key turns off. This isn't ideal, because if the key is turned off when the IMA bay is hot (or vice versa), then the offset value could change once the bay has cooled down. Unfortunately, there's no other valid time to measure the zero-current condition... we have to do it when we know the engine isn't running and the contactor is open. We could probably make LiBCM recalibrate the current sensor every ten minutes when the car is off, but that will increase LiBCM's keyOFF current consumption. It might be worth it, though, particularly if your car's offset error (versus calibrated error) is larger than I've measured.

A) Note that the issue where the battery slowly discharges (or charges) when Calpod is on is a different issue, which I've described previously. The current displayed to you (and sent to the MCM) is calculated differently (less accurately) than the current value used to integrate net charge into - or out of - the battery during each keyON session. Specifically, that more accurate current value lives within SoC_integrateCharge_adcCounts, and uses units 'uC'. Even after I fix the "Calpod discharge" issue, these two current calculations will remain separate... there's no reason the user ever needs to know that 5000 mAh is 18000000000 uC... it's too large conceptually, but allows me to use integer math without losing resolution.

B) Right now when the key is on, LiBCM always assumes all cells are exactly 5000 mAh (specifically: the value 'stored' in STACK_mAh_NOM). So the SoC value displayed to you is always a percentage of that value, not the actual pack value (e.g. if your battery is at 3789 mAh, that's 75.8% SoC, regardless of the actual 100% SoC value). That will eventually change. For example, if your worst cell is actually 4789 mAh, then the resting cell voltage will drop faster than the estimator anticipates (e.g. if you pull 3000 mAh from the pack, LiBCM's keyON estimator will have reduced the SoC by 60% (3000/5000), whereas the actual SoC reduction was 62.6% (3000/4789). Then, when you turn the key off, LiBCM will correct for this SoC error.

C) keyOFF SoC is presently based off of the lowest voltage cell. If the cells are unbalanced - and the lowest cell voltage actually has more capacity than a higher-voltage cell (because it is smaller), then LiBCM will have misestimated the SoC... and will correct for it during the next key off event where the cells are in fact balanced.

D) It's possible LiBCM needs to wait longer after each keyOFF event to reset SoC...

2. The OEM dash SoC had dropped from full to about half-way.
The OEM dash SoC doesn't mean anything right now. I still haven't figured that out... I keep wanting to, but then I get busy with other things. Such is life. Based on what I've seen, the SoC value updates one time just after keyON, and then never updates again for the duration of the drive... others have mentioned that the SoC might update once or twice after hours of driving, but I've never seen that.

It can be figured out... I have tons of data on it... just haven't fiddled with it enough yet. It's not a top priority.

I got home and plugged in my grid charger and this morning the 4x20 shows 73% which is typical for me. My OEM dash SoC is still pegged around 50%. I know that the OEM dash SoC doesn't work yet so I'm not bothered by that, but I did find it curious that it "moved" last night at the gas station for the first time since my install back in February. I suppose that it is related to the 4x20 SoC jump, but not sure.
Nah, just the MCM sticking to whatever the last packet it thought was valid was... even across key events.
 

·
Linsight Designer
Joined
·
3,611 Posts
Discussion Starter · #605 ·
Catching up on @BLS' issue:
Last week I sent BLS a known-working MCM. He put it in his car and the issue persists.

BLS then measured the MCM'E' connector to see the "keyON-but-not-starter" and "keyON-after-START" values. While the general waveform shape looks correct, the amplitude is way wrong: 30Vpp, whereas it should be closer to the actual pack voltage (e.g. 180 Vpp). Therefore, I suppose one of the following issues is present:
A) The MCM'E' connector isn't connected properly to the 8-pin HVDC screw terminal block on LiBCM. @BLS previously verified the wiring, but probably worth checking again. Use the following procedure to verify everything is connected properly:
-Turn car off.
-Turn the IMA switch off, then unplug the MCM'E' connector from the MCM.
-Turn the IMA switch back on. DANGER: YOU ARE INTERACTING WITH HIGH VOLTAGE IN THE FOLLOWING STEPS! DON'T BE SLOPPY! IF YOU SHORT OUT THE LEADS YOU WILL VAPORIZE COPPER INTO PLASMA!
-Set a voltage meter to at least 200 volt range.
-Place positive meter lead to LiBCM 'HVDC+'. This is on the 8-pin connector with wires sticking up out of the top of the clear plastic cover.
-Place negative meter lead to LiBCM 'HVDC-'.
-You should measure the true pack voltage (e.g. 171 volts). If not, then the HVDC wiring from the junction board to LiBCM aren't connected properly.
-Leave the negative lead in place, but move the positive lead to the RED wire on the MCM'e' connector. You should measure the same voltage (e.g. 171 volts). If you measure zero volts, then there could be an open wire in the MCM' harness, or the HVDC fuse on the LiBCM PCB is blown out. If you measure a negative voltage (e.g. -171 volts), then the MCM'e' connector is connected backwards.
-Leave the positive lead in place, but move the negative lead to the WHT wire on the MCM'e' connector. You should measure the same voltage (e.g. 171 volts). If you measure zero volts, check the MCM'e' wire harness for continuity.

B) The MCM'E' spoofing circuit on the LiBCM PCB is damaged. We can check as follows:
-Turn IMA switch off and disconnect the MCM'E' and HVDC wires from the 8-pin screw terminal header.
-Set a resistance meter to at least 50 kOhm range.
-Place positive meter lead to LiBCM 'HVDC+'
-Place negative meter lead to LiBCM 'MCMe+'.
-You should read 10 kOhm.
-Move positive lead to LiBCM 'MCMe-'.
-Move negative lead to LiBCM 'HVDC-'.
-You should read 10 kOhm.
for this by measuring from HVDC+

C) The waveform was measured incorrectly. @BLS sent videos of his waveform measurements... looks like things are connected properly. One possibility was that the probes are set to one gain (e.g. '5x'), whereas the scope is set to another gain (e.g. '1x'). This would cause the scope to display 20% of the actual voltage.

D) If the measured pack voltage is actually 30 volts, then it's possible that one of the 18S modules is connected antiparallel, such that the voltage between the two packs is cancelled out... leaving only the 12s module... but it would have to be almost completely discharged (2.5 volts/cell) to measure 30 Vp.

Other thoughts:
I 100% test the MCM'e' circuit on LiBCM... so if there is a PCB issue, then it happened after I shipped it (or possibly during shipping).

I'm certain we'll get to the bottom of this.
 

·
Premium Member
Joined
·
1,324 Posts
Thanks for the through reply, Mudder! As always, a great read.
 

·
Registered
2000 Insight (Red) Manual AC 150kmiles 59.8mpg for now....
Joined
·
27 Posts
-Place positive meter lead to LiBCM 'HVDC+'. This is on the 8-pin connector with wires sticking up out of the top of the clear plastic cover.
-Place negative meter lead to LiBCM 'HVDC-'.
-You should measure the true pack voltage (e.g. 171
Ok sorry for the obvious question but when you refer to HVDC+/- these are the 2 that go back to pack +/- on the LIBCM correct? Is it worth taking the pack out to get better access? Thank you for the thorough response looking forward to troubleshooting.
Edit- nvm you stated it in your video. Mcm harness is good. Pulling pack to measure resistance.
 

·
Registered
2000 Insight (Red) Manual AC 150kmiles 59.8mpg for now....
Joined
·
27 Posts
Ah might have found the issue. On step B and I have 10k on pos but not on HVDC - to MCM -
Passive circuit component Circuit component Scale Motor vehicle Hardware programmer

Product Circuit component Motor vehicle Passive circuit component Hardware programmer

(I used a better dmm for the readings but it displays a code in place of OL so I just used this cheap one for the pics). For the record wore ppe for all checks just took it off to use phone for pics.
 

·
Linsight Designer
Joined
·
3,611 Posts
Discussion Starter · #609 ·
Post a picture of this part of the PCB:
Passive circuit component Circuit component Light Hardware programmer Electronic engineering

With the IMA switch off, each one of those blue resistors in the picture should measure 10k. I suspect R72 is either damaged or missing. I'm not sure how that could have passed my several visual inspections - and especially my automated tester that I run on every single PCB - but anything is possible.

If you can read a schematic, here's what you're looking for:
Rectangle Slope Font Line Parallel
 

·
Registered
2000 Insight (Red) Manual AC 150kmiles 59.8mpg for now....
Joined
·
27 Posts
Circuit component Passive circuit component Hardware programmer Electrical wiring Computer hardware

I can try to get a clearer photo if need be. Won't be able to get to it tonight but I will check the resistance and report back.
 

·
Linsight Designer
Joined
·
3,611 Posts
Discussion Starter · #612 ·
All I can tell is that the parts are all there... need more detail to see if they're connected. You could also use the schematic I posted in #609 to test out continuity between R69, R72, HVDC-, HVDC+, MCMe-, MCMe+.
 

·
Registered
Joined
·
218 Posts
Captured an IMA code today occurred at WOT, just as the AC compressor engaged. P1440(57) w/ CEL blink 69.
SOC mid 50s


Also noticed ABS 61; EPS 22/23 - codes for another day. Have always had ABS pump kick on on larger bumps in the road while stopping. 4ga welding wire grounds, <6 month old Everstart Lawn & Garden 340CCA.

PS abnormal cell imbalance issue completely resolved by not grid charging above 75%. dV < .005 throughout the SOC 40 to 75 range.
 

·
Registered
2000 Insight (Red) Manual AC 150kmiles 59.8mpg for now....
Joined
·
27 Posts
All I can tell is that the parts are all there... need more detail to see if they're connected. You could also use the schematic I posted in #609 to test out continuity between R69, R72, HVDC-, HVDC+, MCMe-, MCMe+.
Blue Product Circuit component Passive circuit component Hardware programmer

Circuit component Hardware programmer Electronic engineering Font Electronic component

Hopefully these are a bit better. I've been swamped with work the past couple days but I'll go out tomorrow and check continuity.
 

·
Linsight Designer
Joined
·
3,611 Posts
Discussion Starter · #615 ·
R72 looks in-place and connected. Can you double check the resistance between MCMe- and Pack-? It should be 10 kOhm. Note that there's a (nonconductive) conformal coating applied on top of all components. It's possible some got onto the flat head screw on the green screw terminal connector, so try touching the probes to the gates that the wire would go into instead.

One other thing to verify is that all eight screw terminal pins are soldered to the PCB.
 

·
Registered
2000 Insight (Red) Manual AC 150kmiles 59.8mpg for now....
Joined
·
27 Posts
R72 looks in-place and connected. Can you double check the resistance between MCMe- and Pack-? It should be 10 kOhm. Note that there's a (nonconductive) conformal coating applied on top of all components. It's possible some got onto the flat head screw on the green screw terminal connector, so try touching the probes to the gates that the wire would go into instead.

One other thing to verify is that all eight screw terminal pins are soldered to the PCB.
So I stuck the leads in the gates making sure they were touching the sides but it still read open. Flipping the board over all 8 pins are soldered with a good connection. I checked the resistance on the back side of the board across MCMe- and pack- and got 10k. I find it hard to think that there would be a break in the 8 pin for just one but I guess it's possible.

Edit: It was bugging me so I cut some new wires and threw some terminal ends on them, running out of the two ports. Got 10k with a little bit of shifting around so there might be something off set a little bit in there. Gonna throw so terminal ends on the mcm leads and try it again.
 

·
Linsight Designer
Joined
·
3,611 Posts
Discussion Starter · #617 ·
Gonna throw so terminal ends on the mcm leads and try it again.
Any updates? Is it working now?

...

FYI: I still haven't received the production PCBs. They've been held up in customs for a while now. First the shipper didn't include documentation and wouldn't respond to my emails/messages. Next customs claimed the declared value was too low. Now they're having me fill out "Aluminum Extrusion Worksheets" (for the stencil frame) and other silly things.
 

·
Registered
2000 Insight (Red) Manual AC 150kmiles 59.8mpg for now....
Joined
·
27 Posts
Not giving up but no luck so far. The resistor bank appears to be good after more testing. Went to get the waveforms with the probes set correctly this time and got a lower reading than before seen here Round 2 Voltages

Here's a video of the wiring Battery Wiring
 

·
Linsight Designer
Joined
·
3,611 Posts
Discussion Starter · #619 ·
The wiring all looks correct. Try measuring the same test points with a digital multimeter set to at least 200 VDC range. When the key is off, you should measure the actual pack voltage across VPACK+ & VPACK-. You should also measure the actual pack voltage across MCMe+ & MCMe-... all the way to the MCM connector. If that's the case, then something about your scope configuration/settings is wrong.
 

·
Registered
2000 Insight (Red) Manual AC 150kmiles 59.8mpg for now....
Joined
·
27 Posts
When the key is off, you should measure the actual pack voltage across VPACK+ & VPACK-.
Just for clarification, I'm assuming this is with the mcm plugged in but what about the VPCK+/-? Am I hooking everything back up but leaving the two HVDC wires going into the 8 pin board out like in the previous resistance check or is this a full assembly?
 
601 - 620 of 674 Posts
Top