Honda Insight Forum banner
761 - 780 of 830 Posts

·
Registered
Joined
·
741 Posts
Putting the IPU cover back on caused my OBDIIC&C to keep resetting due to interference on the H Line.
Noise really bounces around in there with the cover on.. :eek:

Anyway I cured it by adding a tiny 10nf ceramic cap between the HLine and GND at the OBDIIC&C pcb.
I have seen this sort of thing before with mods in the back involving re-routed and extended HV wiring..
We used to put in what looked like sheets of plasticized steel wool to cancel noise reflected along slab faced metal cans
YMMV
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #763 ·
Just got back from the Mid-Atlantic Insight meetup, plus a few days wandering around West Virginia. Trip notes:
-1500 miles with zero errors.
-I've now driven the various LiBCM prototypes ~6000 miles.
-West Virginia is QUITE mountainous. I charged/discharged my pack at least a dozen times in just a couple hours.
-I agree with Peter's/Isaac's/Natalya's statements that the default voltage spoofing algorithm isn't optimized for fuel efficiency. I don't hypermile, but I did notice that the pack voltage adjustment during regen causes a race condition, wherein the MCM dials back current (due to pack voltage rising), which lowers spoofed voltage, which then causes the MCM to dial current back up. I will further optimize this algorithm to remove that issue... later. Right now I'm focused on hardware... in a few days - after I order polycarbonate covers and RevC PCB - I'll switch back to firmware.
-For now, if you don't like the regen behavior, just change the profile (in config.h). The most OEM profile is "VOLTAGE_SPOOFING_DISABLE", which will cause LiBCM to always spoof 12 volts less than the actual pack voltage.
-With the existing algorithm the background charge tends to keep the pack topped off... Even when I hit the assist hard, within half an hour the pack is full again. I get that this background regen isn't ideal for hypermiling, but on the other hand it does allow the pack to stay charged during long trips.
-Based on my anecdotal observations, the MCM will continue charging the pack until it rises above a certain voltage (TBD)... even with the "disable regen" flag set. For example, when I was driving down a 9% grade for more than 1000', the MCM was still charging the pack even though the regen_disable flag was set. I just used Calpod... but certainly something we need to adjust with the firmware.

Overall, no new hardware issues to report... LiBCM's shortcomings at the moment are entirely firmware based.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #764 ·
In regards to noise in the IMA bay, I did notice that my Pegasus was acting up... a couple times my instrument panel froze up. It's possible we'll need to add some ferrite cores after all. I'm considering putting them on the BMS leads, so that the voltage stops jumping around. Not sure if that'll work, but I'll be testing it out next time I take out the pack (probably next month when LiBCM RevC arrives).
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #765 ·
Peter, are you seeing much heat from long pulls? I have a hard time going through one charge before thermal throttle takes over and I upgraded my fan again and this was in roughly 75 degree weather (22?c)
I pushed these batteries hard this weekend... I didn't feel anything noticeably warm, but note that I haven't programmed LiBCM's QTY17 temperature sensors yet... so it's hard to tell what the actual pack temperature is. It's on my list, but right now the beta testers are running blind as far as pack temperature.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #766 ·
If we end up needing to reduce the high current leads (to reduce noise), the optimal (shortest) cable configuration is with QTY1 18S+ & QTY1 18S- pack. That reduces cable length by several feet.

Of course, another option is to add ferrite cores.

...

In general, the increased noise in the IMA bay could be due to a higher 'Q' value... basically because the battery has less ESR, the current pulses end up having higher di/dt, which would radiate more noise. Overall, I'm not worried about the noise... ferrites will solve the problem, so ultimate it comes down to cost. That'll be something we figure out with the RevC PCBs... which have quite a few onboard noise reduction elements.
 

·
Administrator
Joined
·
13,097 Posts
Regen should be at around 50% of the peak Assist power although it is harder on the IGBT especially at low rpm and can trigger the FSC code safety protection.

With the voltage we have available in the 48S setup I would hope we should see over 10-12kw regen and about 20-25kw assist when it's all sorted..

A higher voltage pack will give more assist, but might actually give less regen at low rpm as the motor simply might not be generating enough voltage.. I seem to remember seeing this effect with my 200V+ HV packs.

There are also some other weird dead band effects at high or low voltage extremes with certain rpm's.
 

·
Premium Member
Southern California
Joined
·
896 Posts
In regards to noise in the IMA bay, I did notice that my Pegasus was acting up... a couple times my instrument panel froze up. It's possible we'll need to add some ferrite cores after all. I'm considering putting them on the BMS leads, so that the voltage stops jumping around. Not sure if that'll work, but I'll be testing it out next time I take out the pack (probably next month when LiBCM RevC arrives).
Whoa, that's really bad! First I've heard of that happening. How do you fix it? Key off and on?
Any other symptoms?
 

·
Engine-Off-Coast
Joined
·
2,126 Posts
-With the existing algorithm the background charge tends to keep the pack topped off... Even when I hit the assist hard, within half an hour the pack is full again. I get that this background regen isn't ideal for hypermiling, but on the other hand it does allow the pack to stay charged during long trips.
-Based on my anecdotal observations, the MCM will continue charging the pack until it rises above a certain voltage (TBD)... even with the "disable regen" flag set. For example, when I was driving down a 9% grade for more than 1000', the MCM was still charging the pack even though the regen_disable flag was set. I just used Calpod... but certainly something we need to adjust with the firmware.
Changing SoC, in my experience, has been way more reliable (100% so far) at controlling regen. If SoC is 60 you're getting background regen. If it's 72 you're not. I feel like SoC is the magical lever to push.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #771 ·
Whoa, that's really bad! First I've heard of that happening. How do you fix it? Key off and on?
Any other symptoms?
I haven't reported it yet, because I haven't looked into the root cause. Yeah, basically the entire instrument panel freezes (speedometer, etc), and doesn't unfreeze until I keyOFF/ON... the Pegasus joystick is non-responsive.

I'll let you know more when I research it more (later). The problem is almost certainly on my end. I don't actually think it's a noise issue, because I was able to reproduce it with Calpod on.

I wouldn't worry too much about it. We'll cross that bridge when we get to integration hell (again) in a few weeks.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #772 ·
Just ordered the RevC PCB... my goal had been to order it prior to leaving for the Mid-Atlantic meetup (last week), but it took a bit longer to make everything fit. Alas, OCT01 thru OCT07 is a Chinese holiday, so that's going to add another week to the PCB production time. With these delays, I suspect the PCB will arrive OCT15. Ultimately this won't actually delay the project much, as I've got plenty of other things to do... (e.g. building the manufacturing line, firmware work, designing new polycarbonate cover, figuring out various battery configurations, designing an 18S->12S bracket, etc).

...

RevC ended up having QTY694 components, which is QTY71 more parts than RevB. All the free space leftover on RevB is now stuffed with components:
Rectangle Art Urban design Font Electric blue


Key features that made it onto the final RevC PCB (as previously discussed):
-LEDs are visible from the top (with IMA cover removed)
-LiDisplay connector is accessible from the top (without removing clear cover)
-Latching, right angle ribbon cable connectors
-Tons of lowpass filtering (to reduce noise)
-QTY3 new temperature sensor to directly monitor each battery
-QTY17 total temperature sensors
-Several power savings improvements
-Added 'isCoverPresent' switch (not an interlock)
-Buzzer
-LiBCM now powers up when IMA switch turns on
-Improved current sensor accuracy under heavy assist/regen
-Improved voltage spoofing accuracy
-Reduced QTY unique part numbers
-Numerous minor hardware fixes
-Added QTY4 GPIO pins, to allow LiBCM to control other things (e.g. GRN/BLK wire to DCDC)
 

·
Premium Member
Joined
·
2,036 Posts
Nice! So this is the final revision right? I believe I have read that a couple times in the past.
 

·
Premium Member
Southern California
Joined
·
896 Posts
I haven't reported it yet, because I haven't looked into the root cause. Yeah, basically the entire instrument panel freezes (speedometer, etc), and doesn't unfreeze until I keyOFF/ON... the Pegasus joystick is non-responsive.

I'll let you know more when I research it more (later). The problem is almost certainly on my end. I don't actually think it's a noise issue, because I was able to reproduce it with Calpod on.

I wouldn't worry too much about it. We'll cross that bridge when we get to integration hell (again) in a few weeks.
If you can reliably reproduce it, try with the OBDII connector unplugged. Wondering if noise on the H-Line is doing something funky, or if the ground loop might be putting too much noise into the board.
I'm pretty confident this won't be an issue on the new version (it's all-around better designed, and with isolated OBDII communication, so no ground loops), but it would be good to narrow down what is happening.
And the firmware will have a watchdog, the prototype firmware doesn't.

-Latching, right angle ribbon cable connectors
-QTY3 new temperature sensor to directly monitor each battery
Much appreciated. :)

And nice Nietzsche quote (I had to look that up though)
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #777 ·
Nice! So this is the final revision right? I believe I have read that a couple times in the past.
Too early to tell... I previously thought the RevB PCB could be the final version... and was quite wrong ;).
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #778 ·
If you can reliably reproduce it, try with the OBDII connector unplugged. Wondering if noise on the H-Line is doing something funky, or if the ground loop might be putting too much noise into the board.
I'm pretty confident this won't be an issue on the new version (it's all-around better designed, and with isolated OBDII communication, so no ground loops), but it would be good to narrow down what is happening.
And the firmware will have a watchdog, the prototype firmware doesn't.
Pegasus works just fine with the OBDII connector unplugged. Note that the H-Line isn't connected to LiBCM (at the moment, although technically it is supported if the H-Line is routed to the LiBCM PCB). I can reliably reproduce the hang whenever Pegasus is plugged in... it wasn't always like this (with LiBCM), so I'm wondering if the data I'm sending on BATTSCI is 'good enough' to keep the MCM happy, but not properly formatted for whatever Pegasus is expecting?

Again, I haven't looked into this enough to formally file a bug report. For now I've just unplugged Pegasus.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #779 ·
Goodness wood is expensive! It's more than doubled in four years (since I bought the materials to build our house).
Anywho, giving that insight away last week freed up room in the garage for an SMT manufacturing line:
Wood Interior design Kitchen Window Floor


I also ordered all the new parts (that weren't present on RevB) for QTY100 LiBCMs.
 

·
Registered
2003 Honda Insight CVT, 2001 Honda Insight MT, 2000 Honda Insight MT
Joined
·
131 Posts
Looks great. Even clean and organized...thanks again for the presentation at the Maintenance Meet in Frederick.
 
761 - 780 of 830 Posts
Top