Honda Insight Forum banner
141 - 160 of 977 Posts

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #141 · (Edited)
Status update:
I've dimensioned the OEM MCM enclosure and laid out the entire BMS. The three yellow planes are isolated high voltage areas, one for each LTC6804-2:
Green Yellow Parallel Technology Font

Each plane contains measurement and sinking circuitry for 12 sequential cells (e.g. cells 1:12, 13:24, 25:36). This is called "bank isolation."

If you look closely, you'll notice there's ZERO electrical contact for at least 3 mm around each of the three BMS subsystems. I'm using an isolated transformer to communicate with each BMS subsystem. This is called galvanic isolation and it prevents the high voltage lithium battery from shorting out into the 12 V system if some component fails or (more importantly) if the user connects cells in the wrong order. Note that there is zero reverse voltage protection, so the bank isolated section of the board will be destroyed if you connect any cell backwards (e.g. you accidentally flip two leads, e.g. you wire C1->C2->C4->C3/C5/C6/C7/C8/C9/C10/C11/C12). I simply didn't have enough room for additional protection components. Note that I couldn't install series resistors to limit current because that would prevent accurate voltage measurements and would also limit the balancing circuitry's functionality. You've been warned ;). Fortunately, if a user accidentally fries a particular subsection, it's fairly easy to replace the damaged components... I've made the traces wide enough that the components will fail first, so it'll be easy enough to fix with a hot air knife, new transistors, & a new LTC6804-2.

Since we have three galvanically isolated 12S measurement systems, a user could theoretically monitor three separate banks, each at ANY relative potential. For example, a user could monitor cells between -20&24 V (e.g. -20/-16/-12/-8/-4/-0/4/8/12/16/20/24) on one bank, while simultaneously monitoring cells between 100&144 V on another bank, while also simultaneously monitoring cells between 200&244 V on the third bank. Pretty cool, right?

One cool thing for those using the Chevy Volt cells: the Volt's packs are 12S, which means you don't need to do any fancy math to remove the voltage drop on the leads between each pack.

One nonideal (but trivial) thing is that users must manually connect cell12+ to cell13- AND cell24+ to cell25-, because this is where each galvanically isolated boundary is on a 36S pack. Note that 12+/13- & 24+/25- are the shorted together on a 36S pack, but since they aren't shorted together on my PCB (for the reasons explained above), the user will need to physically connect a wire between 12+/13- AND 24+/25-. Otherwise, the BMS will measure incorrect voltages (around zero volts) for cells 13 & 25... this will make the Linsight board disable the IMA system (because it thinks two cells are at zero volts).

In short, the above paragraph means you'll have one extra wire jumping from connector A to B, and another from B to C. Doing this on the board (e.g. with jumpers) is a huge safety nono in UL's eyes, as an inexperienced user might not realize the jumpers are installed and then electrocute themselves.

...

The BMS monitors each cell's voltage and will disable the IMA system whenever ANY cell is too high or low (user selectable setpoints for each cell). There's also a logic-enabled 15 Ohm resistor to passively balance full cells. A Nissan Leaf cell is nearly full at 4.1 V, and thus each resistor sinks 1.12 W (V^2/R) when enabled. Balancing multiple cells simultaneously will cause the PCB to heat. Balancing EVERY cell simultaneously (there's no reason to do this conceptually, as the entire stack is full, i.e. the charger should be disabled instead) will cause the PCB to sink 40 W, which is a lot. To prevent the PCB from overheating, I've installed three onboard temperature sensors that will automatically cycle through the cell balancing proceedure as efficiently as possible without causing a meltdown. This is another safety feature, but a user can disable it if they know what they're doing (I can't imagine why they would).

...

Here's the BMS portion completely routed:
Electronic engineering Electronics Microcontroller Circuit component Computer hardware

Here's the silkscreen text:
Electronic engineering Electronics Microcontroller Circuit component Technology

The BMS portion only required 114 square cm, which is MUCH less than I thought it might. I've still got 88 square cm of room left for the uC and everything else! That's plenty of room based on what I know today.

...

As Mario alluded to, I've chosen a much more powerful microcontroller: Atmel's AT32UC3C. The automotive grade version runs up to 50 MHz, which is more than three times faster than the OEM uC (16 MHz). The additional processing power will make the BLDC chopper virtually silent, as the chopping will occur at ~30 kHz (people hear up to 20 kHz). The OEM chopper runs much slower, in the audible range.

I'm still considering a dedicated 20 MHz ATtiny uC for the actual BLDC sine drive, as that's the most time-critical element. A dedicated BLDC uC allows 100% BLDC processing time. I'm a couple days from determining if this is required, but I might just do it anyways, as it's a safe option as long as it fits on the PCB.

I've worked out all the signals and how I'll interface them to the processor, so it's just a few more days until I order the first (and hopefully only) PCB. Mario, before you even ask, I'm still shooting to get the first units out the door lateFeb/earlyMarch.

...

That's all for today... got to work on another PCB (contract work).
 

·
Premium Member
Southern California
Joined
·
896 Posts
Mario, before you even ask, I'm still shooting to get the first units out the door lateFeb/earlyMarch.
Hey, I wasn't going to say anything! Rude. :)

Looks fantastic so far! And great job getting the HV bits routed.
The 32UC is a good choice for a microcontroller.
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #143 ·
In regards to charging, I previously mentioned the DCDC is permanently connected to the Lithium cells in my design. This allows the Linsight board to enable the DCDC converter to charge the 12V system even if the car is off. This also allows us to run the fan that cools the DCDC converter and BLDC driver with external air. Because those components are in near proximity to the lithium batteries, we can exhaust some heat from the otherwise thermally sealed system.

The DCDC consumes 9.15 W at idle (measured disconnected from insight), so it'll only be enabled as needed to run cooling fans. Leaving the DCDC on 24/7 would empty the lithium battery in 40 days, but that's plenty long if you leave your car in the sun for a couple weeks while you're on that Tokyo business trip.

The Linsight board is also on 100% of the time. Thus, it can balance cells whenever it wants to. Note that the balancing circuitry is disabled if the IMA compartment gets too hot (because the balancing circuits place more heat in the IMA battery bay.

The Linsight board will constantly pull ~2 mA from the lithium pack (300 mW). A Leaf pack is 63,000 mAh, and thus it would take 3.5 years without charging to empty the pack. I suspect the self-leakage would empty the battery first ;).

The IMA motor is still routed through the OEM relays, so it is disabled when the key is off. This is a safety feature... a user would rationally expect the car to be OFF whenever keyed OFF.
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #144 ·
Hey, I wasn't going to say anything! Rude. :)

Looks fantastic so far! And great job getting the HV bits routed.
The 32UC is a good choice for a microcontroller.
You mentioned you're an EE. If you'd like to review my PCB as it exists today, message me your email and I'll send you a copy. I'm using KiCAD (free, open source), but I suspect whichever program you're using can import it for review. I could even send you the gerbers if you just want to verify the isolation barriers.
 

·
Premium Member
Southern California
Joined
·
896 Posts
You mentioned you're an EE. If you'd like to review my PCB as it exists today, message me your email and I'll send you a copy. I'm using KiCAD (free, open source), but I suspect whichever program you're using can import it for review. I could even send you the gerbers if you just want to verify the isolation barriers.
I'd be happy to. I use Altium, though it might be able to import KiCAD. Better send gerbers as well just in case. I'll send you a PM.
 

·
Super Moderator
Joined
·
9,229 Posts
Looking great!

I realize it is very late to be asking questions, but just gotta do it anyway. Will the design work with the Genesis MIMA? That system uses on P&P adapter at the MCM.

Also, what happens to the stock BCM & MCM plugs at the front of the pack? As I "think" I understand, the BCM/MCM are eliminated along with the stock battery. You are using some of those signals, right? In case you are planning to plug these into something, I think I recall from Mike's MIMA work that the sockets are unavailable.
 

·
Premium Member
Joined
·
3,661 Posts
Looking great!

I realize it is very late to be asking questions, but just gotta do it anyway. Will the design work with the Genesis MIMA? That system uses on P&P adapter at the MCM.

Also, what happens to the stock BCM & MCM plugs at the front of the pack? As I "think" I understand, the BCM/MCM are eliminated with the stock battery. You are using some of those signals, right? In case you are planning to plug these into something, I think I recall from Mike's MIMA work that the sockets are unavailable.
Good question Jim, I've been wondering about this also. Peter's IMAC&C would be another question too.
 

·
Premium Member
Joined
·
2,949 Posts
The 12S pack you propose would only be ~48 V. As a BLDC motor spins, it creates a "Back EMF," which is proportional to the motor's "motor constant", which is a function of the motor (not changeable). The faster a motor spins, the more back emf it generates. When the driving input voltage (~48 V in your hypothetical) is less than the motor's back emf, then said driving source is unable to drive said motor. The insight would be able to charge the battery, but wouldn't be able to assist.
Mentioned Chart of the IMA Back EMF , attached.

There are corner conditions where the above isn't true.
Which I would still suspect are not worth the effort.

- - - -
Sorry if I previously gave the wrong impression about it not being worth doing last time.
 

Attachments

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #149 · (Edited)
Looking great!
I realize it is very late to be asking questions, but just gotta do it anyway. Will the design work with the Genesis MIMA? That system uses on P&P adapter at the MCM.
If the Genesis grid charger can charge lithium, then it could theoretically work. I suspect since it's programmable it'll work without a problem electrically... we'll just need to figure out how to hook it up electrically. I have a user-configurable serial port for talking to chargers, so yes, it could work if all of the other above items are possible. I think it will work, but it's not the focus of my research at present.

Any MIMA system that doesn't internally modify the MCM PCB should still work, but keep in mind my PCB natively supports MIMA, too. The only reason people will want to use their existing MIMA systems would be because they like the user input method. Systems that externally modify the input PWM waveforms will work. For example, Peter's modified MCMs won't work, because it internally modifies the MCM. However, his joysticks that plug into his OBDIIC&C will work, but those joysticks he sells that plug directly into the OBDII port WILL NOT WORK, as they lack a vital connection between two data lines (that is present in the OBDIIC&C). Peter and I discussed his system last month and we're both VERY confident his joysticks that plug into the OBDIIC&C will work (it'll basically come down software support on both our ends).

Another way to conceptualize my system:
You are a Honda Insight ECU.
You see a black box.
You can send and receive signals from the black box, but you can't see inside the black box.
Inside the black box is the complete IMA system, either the OEM system or the Linsight replacement.
You don't know which one is there, because they both respond exactly the same to your input commands.

Also, what happens to the stock BCM & MCM plugs at the front of the pack?
The two BCM plugs that go to the NiMH battery are removed with the NiMH battery (it doesn't go anywhere else). The third BCM plug is no longer used; it just sits there looking pretty.

The MCM has 5 plugs, one on the back (pack plinus), and four on the front. The one of the back goes away with the battery; the Linsight board doesn't have this connector. The four front MCM plugs are all reused. They plug directly into my Linsight MCM board. No wires are cut. The connector exists and is orderable online... I ordered one a few weeks back from Mouser and physically have one in my possession (hat tip: Mike!):
Computer hardware Electronics Electronic engineering Technology Network interface controller

Thus, there's only one plug (to the BCM) that isn't reused. It isn't needed for anything. Everything routes through the single Linsight PCB, which reuses the OEM MCM enclosure.

As I "think" I understand, the BCM/MCM are eliminated along with the stock battery.
Yes, the OEM BCM/MCM/NiMH are removed entirely.

You are using some of those signals, right? In case you are planning to plug these into something, I think I recall from Mike's MIMA work that the sockets are unavailable.
Yes, I'm reusing 46 signals, plus the relevant power/ground leads. This is all transparent to the user... they just plug the existing 4 connectors into the Linsight MCM PCB.

Zero wires are cut. Insight wiring harnesses remain 100% virgin. The main goal of my 'drop-in' design was that it simply dropped in... tough to make it happen, but I've got all the signals figured out. The last piece dropped into place as soon as I realized the battery current sensor (which is more accurate, but plugs into the BCM (which is removed)) used the same connector as the MPI current sensor (which is horribly inaccurate, but plugs into the MCM). The sensors are wired differently, but I take care of that inside my PCB... still 100% plug-and-play to the user.

The only wiring the user must do is connect the battery sense leads to each cell. I'm using screw terminals, which means a user strips each wire, inserts it into the terminal, and then screws a flat head screw down, thus securing the wire. I considered using the plug that comes with the Nissan Leaf BMS, but didn't because I want support for any lithium battery. Thus, the user will need to manually connect 37 leads, one to each side of each cell.
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #150 · (Edited)
Mentioned Chart of the IMA Back EMF , attached.
Right Click -> Save As.

As you can see, the pack constant is in fact a constant (i.e. the graph has a constant slope). Note the Y axis is line-to-neutral, which is unlcear... in a Wye three phase system Vrms is sqrt(3)*Line-toNeutral voltage. However, the insight uses - to my knowledge - a Lambda three phase BLDC (i.e. the BLDC motor is lambda, not Wye), so I'm not sure what to make of the Y axis. My first thought is that the label "line-to-neutral" means "pack plus to pack minus." However, this doesn't make sense because that would mean you'd have to have very high RPM to charge the IMA. I suspect the data gathered in that particular instance is a "loaded" measurement, which means the IMA motor phases remained connected to the IMA system during testing. A proper measurement would have left the IMA motor completely disconnected, so as to read the open circuit voltage generated at no load. Without further context, I can't comment further. If "average voltage" is actually "RMS voltage", then "line-to-neutral" is in fact pack plinus, but again, that testing probably wasn't performed with an open circuit motor.

Where did this picture come from, Ian?
 

·
Super Moderator
Joined
·
9,229 Posts
Thanks mudder for answering my question. Now, all I have to do is read the answer an couple more times and I think I've got it;)

This pack, with a MIMA, is going to bring a whole new meaning to MPG competitions. We'll be seeing numbers like 152.4, 163.1 and perhaps a 200, depending on course length.
 

·
Premium Member
Joined
·
2,949 Posts
Where did this picture come from, Ian?
DOE 2006 document #
ORNL/TM-2006/535

Link for a pdf copy in my Google Drive.

It is technically a test of the ~12kw 2005 IMA motor used in the Honda Accord .. not the ~10kw IMA motor used in the Insight ... soo there are no doubt a few differences .. but I suspect the back EMF is very similar .. and still doing a very similar-ish line / progression .. thus an example of the kind of issue you wrote about previously on using a battery voltage that is too low.

As for the 'average' ... the above pdf includes the following about that as well.
 

Attachments

·
Administrator
Joined
·
13,083 Posts
Comments.

The tiny micro analog joystick I am using now for the IMAC&C P&P is very neat and much more compact than the MIMA joystick.. It is documented in the IMAC&C thread IIRC.

I think my little IMAC&C joystick board might be made to work directly with this project as well as via the OBDIIC&C, it's might be just a case of software changes or a small add on board/transistor hack.

But as this project supports manual control anyway? Then it may be academic.
Perhaps we can retain the HLine and Mudder's System can listen for IMA control signals as per my IMAC&C P&P protocol, when they are present it can overide the normal IMA in the same way as my internal MCM pcb does.. Is that too onerous?

Ima motor back emf was a concern for me, which is why I went for a pack nearer the oem voltage and then went higher voltage when increasing the power output. Also when wanting max motor output at high rpm then the voltage has to be high enough to allow that.. Back emf at high rpm 5k+ is probably significant, but I have no idea how that would translate into % reduced motor output with a lower voltage pack. (Regen potential would be better though, so pack voltage is a double edged sword.)
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #154 ·
Peter, per our conversation last month, IIRC your joysticks that plug directly into the OBDII port lack a K-Line connection. I need certain K-line parameters rebroadcast onto the H-line that runs to the MCM. Your OBDIIC&C can do that, so joysticks plugged into the OBDIIC&C will work.

FYI: I'll offer a plug-in converter for those that don't have either of Peter's devices. It'll plug into the OBDII port. It won't be necessary if using any of Peter's supported devices.
 

·
Premium Member
Joined
·
2,949 Posts
depending on course length.
Fully charged .. the energy replacement equivalent of about ~2/3 of a gallon of gasoline .. The fastest the IMA system could probably use it up would be about ~45 Minutes or so.
 

·
Super Moderator
Joined
·
9,229 Posts
Fully charged .. the energy replacement equivalent of about ~2/3 of a gallon of gasoline .. The fastest the IMA system could probably use it up would be about ~45 Minutes or so.
That would be great. I see some magical MPG numbers in my future;)
 

·
Premium Member
Joined
·
2,949 Posts
Measuring only mpg is pointless if you are also using additional power sources for propulsion.
All sources of energy would need to be accounted for, including electron consumption.
Information on both sources will be available .. the MPG FCD .. and the Linsight system energy usage ... Sense they are considering having Pegasis Dash Mod and the Linsight Drop in be compatible and able to 'talk' with each other .. the final combined all sources MPGe might eventually be an option to be able to have displayed on the dash .. :D
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #160 ·
Ian, repost eMGP in Chris' (Mario's) thread... he'll like that excellent option.
 
141 - 160 of 977 Posts
Top