Honda Insight Forum banner
241 - 260 of 977 Posts

·
Premium Member
Joined
·
2,949 Posts
I had read of someone using 72S but I would like to run sub-packs, but I'm not sure of the best arrangement and finer details. Any recommendations for this type of cell or in general?
Suggest stick with the sub-packs .. placed/assembled into the main pack one sub-pack at a time in the car .. It is much harder to work with one big pack .. At least that's been my 20/20 hindsight from working with my 48s pack (~25kg) .. I had originally intended on breaking mine into either six 8s (~4kg) or three 16s (~8kg) sub-packs .. and sometimes I've wished I had stuck with that original intent .. This applies even more so with yours at about twice the OEM pack weight ~50kg.

Also, what additional components in series between sub-packs could be added, particularly for electrical isolation & safety?
Thin layer of dielectric film can provide additional electrical isolation between cells / sub-packs .. I forget the type Peter used between his cells .. but I used a 1mil (0.001") thick Teflon Film .. 2,000v Dielectric strength .. even 73 sheets of that only add 0.073" (~1.9mm)
 

·
Premium Member
Southern California
Joined
·
896 Posts
Also, I haven't started my joystick solution yet, so for now the early adopters will need Peter's OBDIIC&C AND the joystick he sells that plugs into said OBDIIC&C. I've probably run out of time to juggle a joystick design, too. We'll see how much energy I have tomorrow; maybe I can knock out that design in a day or so.
That joystick can be connected directly to OBD-II, you don't need OBDIIC&C. It sends packets directly over the H-Line:
http://www.insightcentral.net/forums/906417-post264.html

In addition, I want Pegasus to do emulation of Peter's OBD-II joystick (as Pegasus will use an analog joystick with the same functions), so that's one more potential solution. That still leaves the question of what everybody else does, though.

Seems like Linsight should include its own joystick and LCD, with the user having the option of using Pegasus or OBDIIC&C if they happen to own one as well.
 

·
Super Moderator
Joined
·
9,229 Posts
Poor connection so can't read much, so don't know everything that has passed, but I vote for a progressive analog joystick for MIMA. That is one of the nice features of the Dabrowski MIMA. Progressive Analog has an intuitive feel that digital steps will never duplicate, for me at least.:)
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #244 · (Edited)
hmm, maybe I could have worded the question a little differently. You've been mainly planning on using the 18 Nissan Leaf modules, correct? A Leaf module contains 4 cells, for a total of 72 cells. The Linsight is slated to 'support' 36 cells "onboard". So my question is, have you also been planning on using the onboard 36 cell support to monitor 2 cells per module, or something else? For example, the passive balancing - are we balancing 2-cell pairs when we use 18 Leaf modules? Or are we expected to buy additional LTC thingys? Is there much of a problem/risk not monitoring every cell? Should/can/do we treat the 2 parallel cells as 1 cell?
This should clear up most of the confusion from my previous posts:
Whenever I say "n cells" (e.g. "36 cells"), I mean "n cells in series." I mentally combine any number of parallel cells into a single 'supercell', as there's no need to monitor each cell in parallel AS LONG AS all parallel cells are shorted together with large enough leads that the entire current pulled from the pack can travel through said leads without developing a large voltage difference between parallel cells (at that point, the cells aren't truly in parallel). For the Leaf cells, there's a gigantic copper bus bar that shorts each pair of parallel cells inside the 2S2P enclosure. Note that Nissan only allows three external voltage sense points: the external plus and minus, and also the center tap.


Apologies I've caused confusion multiple times in this post because whenever I've said "n cells", when I should be more specific. In all future posts, I'll clearly specify the number of cells in series AND parallel (.e.g. "36S2P", which means 36 cells in series, two cells in parallel).

To answer your questions specifically:
You've been mainly planning on using the 18 Nissan Leaf modules, correct? A Leaf module contains 4 cells, for a total of 72 cells. The Linsight is slated to 'support' 36 cells "onboard". So my question is, have you also been planning on using the onboard 36 cell support to monitor 2 cells per module, or something else?
Linsight will monitor up to 36S cell packs. Said packs can have ANY number of cells in parallel, as long as ALL cells at a particular pack voltage are wired in parallel with leads thick enough to prevent a voltage differential between any said parallel cells.

For example, the passive balancing - are we balancing 2-cell pairs when we use 18 Leaf modules?
Yes, when using Leaf modules, Linsight balances two parallel cells per channel. Note that conceptually a lithium battery can be divided into N parts with 1/N the cell's capacity. For example, a 4 Ah cell is 100% equivalent to four 1 Ah cells in parallel, AS LONG AS the cells are truly connected in parallel. As noted above, the Leaf cells in parallel are truly in parallel because they have a huge copper bar and each voltage tap in in the middle of said tap.

Or are we expected to buy additional LTC thingys?
Only if you have more than 36 cells in series. You could have four cells in parallel (i.e. 36S4P) and Linsight would still manage all cells AS LONG AS they are truly in parallel.

Is there much of a problem/risk not monitoring every cell? Should/can/do we treat the 2 parallel cells as 1 cell?
The Leaf cells present zero problems/risk (in regard to this conversaion) when connected 36S2P to Linsight. A customer that created the following "36S2P" pack would have much risk: two 36S1P packs whose cells were then shorted in parallel to their matching neighbor with 22 gauge wire. Since the wires that combine both 36S1P packs together cannot handle the pack charge/discharge corrent, any imbalance that occurs on a single "unhealthy" cell will cause a voltage differential across the 22 gauge "parallel" wiring, which will cause the stronger cell neighboring the "unhealthy" cell to source too much current across the 22 gauge wire, causing the wire to overheat, etc. The Leaf cells don't have this problem because their integrated bus bar can handle a few hundred amps without developing more than a few mV differential between the two cells.
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #245 ·
[tiny print in image below apologies - it had to shrink to fit.]
You're right, it's too tiny for me to offer any feedback. I don't know what the full cell voltage will be, but 72S is almost certainly too much voltage for the insight's VCM and DCDC converter. Even at a low 3.6 V full charge voltage, 72S is 259 V! The maximum voltage before things start shutting down is ~210V.
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #246 ·
I just realized a mistake in my math here - OBDIIC&C updates the screen four times a second, which is what I was thinking of. But for each update, it requests 8-10 parameters (4 on H-Line and 4 on K-Line, plus one more on each line for some internal calculations I think, don't have my notes with me), so there's actually 16-20 parameters per second being fetched from each line. It is still about 50% of the bandwidth, though.
I think we'll have enough bandwidth because you moderate OBDIIC&C when it's piggybacking on your OBDII port. Linsight and Pegasus should plan on communicating via METSCI.

Pegasus should be able to do that. If you provide them as a PID that I can request over the H-Line, then it's super simple. The alternative I suppose would be to request them from METSCI. Either one is definitely doable.
But what if a user doesn't have Pegasus or OBDIIC&C? Do you have plans for a screen that connects straight to Linsight? Or will users just be out of luck (in regards to that feature) if they don't have one of those things?
If a user doesn't have a method to view the temperature sensors, they're still used behind the curtain to disable the IMA system if any temperature falls out of range. Linsight will require at least one temperature sensor (my kit will come with at least a few). Linsight also has three onboard temperature sensors (to ensure the passive balancing resistors don't overheat the PCB), but those sensors won't be able to tell anything about the battery temperature, hence my requirement that at least one external sensor also be installed.

Linsight also has a USB port, so it can display nearly anything to a user when plugged into a laptop. As a fun bonus, the USB port supports host mode, so Linsight can actually act as a computer and receive input from a keyboard, mouse, joystick, etc. I'm thinking it would be cool to have a separate joystick to control Linsight's various modes, but that won't be in the initial software. I love software because it's field-upgradable!
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #247 · (Edited)
That joystick can be connected directly to OBD-II, you don't need OBDIIC&C. It sends packets directly over the H-Line:
http://www.insightcentral.net/forums/906417-post264.html
I know nearly nothing about Peter's joysticks, as I've done zero research on them. Peter mentioned via Skype that his joystick wasn't able to transfer data from the K-line to the H-line. That prevents me from getting data (brake pressed, throttle position, etc). Linsight will operate without K-Line data, but many of its features won't work (i.e. "regen only when braking", "Lean Control", etc). If Pegasus is installed, Pegasus could give Linsight said data. In that case, Peter's joysticks that plug directly into the OBDII port will work.
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #248 ·
Poor connection so can't read much, so don't know everything that has passed, but I vote for a progressive analog joystick for MIMA. That is one of the nice features of the Dabrowski MIMA. Progressive Analog has an intuitive feel that digital steps will never duplicate, for me at least.:)
Yes, whenever I get around to creating my own input device, it'll have a progressive joystick, as I've outlined in post 88.
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #249 · (Edited)
2016JAN25 update:
I took a couple days off after ordering the PCB, which really means I only spent 4 hours per day ;).

To recap the original questions I posed in this thread:
1: Would you pay $1000 for <Linsight>.
2: Would you pay $900 for a PCB that only supports 36 cells?
There were multiple responses of "I'd pay $600." After calculating Linsight's component cost today, I know that I can't sell Linsight for less than $700. Even if I buy parts in quantity (i.e. I build 50 Linsight modules), the parts to build each PCB exceed $250! I then have to reel these parts into my manufacturing equipment (each time I run a batch), run each board through a reflow oven, and then I'll spend another couple hours soldering, programming, testing and troubleshooting each PCB, and finally I'll probably spend at least an hour helping/transacting with each customer. My minimum consulting rate is $100/hour, so Linsight just isn't worth the effort to sell for less than $700, particularly since I expect some customers will install Linsight improperly and thus require further support.

And thus I propose the following social contract (and am soliciting your candid feedback):

(A) I will 100% test every single Linsight I sell before shipping it to you.

(B) I will sell the Linsight PCB for $750*, which includes up to one hour phone/email/PM support. Additional one-on-one support is available at $75/hour.

(C) Additional support is available for free by posting into my "Linsight support" forum post (to be created), as this benefits the community (and not just that single customer).

(D) I will provide an exhaustive youtube video series on how to properly install and use Linsight in an OEM insight. I will create a post linking each video with any additional information I find useful for the install process.

(E) I will not create a text&picture manual, but will answer support questions in the forum.

(F) If you break Linsight by not following my directions, I will determine what you damaged, then repair the board for my actual cost, plus $200 for my effort. That means if you somehow destroy your Linsight PCB completely, I will charge you the actual cost to send you a brand new PCB (~$250), plus $200 for my efforts building you a new one.

(G) If I send you a Linsight that fails because I did something wrong or it failed for a reason I can't pinpoint on you, I will repair or replace your Linsight module for free, minus shipping costs to me ($10 USPS).

(H) I will require you to sign a contract absolving me from all responsibility if Linsight drives through your garage door, slams into that car in front of you because you were riding its draft, or runs over your dog. Note: Linsight has numerous hardware failsafes. For example, Linsight cannot possibly drive the motor if the ignition key is in the off position ... the drive circuitry is disabled three different ways in hardware when the key is off, so even if the firmware makes a mistake, there's zero voltage at the IGBTs, zero voltage at all six IGBT predriver pins, and the master uC hardware pin disables the drive uC in hardware. Whenever the key is on and the transmission is in gear, the person sitting behind the wheel is responsible for insight's motion. The brakes will always overpower Linsight.

(I) I will send prototype Linsight systems to a select few customers and solicit their feedback prior to releasing Linsight for sale to the public.

(J) I will thoroughly test each "general release" firmware package in my insight - or another insight tester's insight - before releasing the firmware package to Linsight owners. I will test the initial release for at least three weeks before sending any prototype units to testers.

(K) I will sell Linsight under an LLC ("Linsight, a Texas LLC") to protect my personal assets. This is because I don't want to waste time defending myself from any lawsuits that may arise from Linsight if anyone craftily works around the contract I outlined in section H. If anyone decides to sue me, I will immediately cease production, destroy all remaining Linsights I have in inventory, and hand you the keys to Linsight, LLC. For those interested, Linsight, LLC is currently worth -$4583, as that's what that shell company has presently spent in parts to make Linsight a reality. I will charge Linsight, LLC a per-unit royalty fee equal to any profit I derive from selling Linsight. Thus, Linsight, LLC will never be worth more than a few thousand dollars.

(L) Unless I so desire, I will not provide one-on-one support to customers that do not use 2S2P cells from the Nissan Leaf, if said support concerns said non-Nissan-Leaf cells. I will provide instructions - via a post on this website and a video on youtube - showing how to change the full/empty voltage setpoints and nominal battery ampacity.

(M) I will make every effort to ensure Linsight meets the capabilities stated in the forum post "Introducing Linsight" (to be created).

Hopefully the above terms aren't too scary. Feedback greatly appreciated.

...

*The Linsight PCB (only) will cost $750. Linsight is BYOB ("bring your own battery"). I will also offer a "Linsight install kit", which will include the following additional hardware:
-spacer washers for Leaf cells (I'll explain their use later)
-200 W grid charger (controlled by Linsight)
-all cables required to connect Leaf cells to OEM insight
-external battery temperature sensor(s)
Kit price is TBD, but I expect it'll be ~$175.

The above kit won't initially include any method to manually control the IMA system. Initial customers will need to use Peter's joystick, as mentioned previously in this thread. I will eventually offer for sale the user controls I outlined in post 88. Price is TBD, but I expect it'll be $75.

Thus, a complete Linsight system - including a grid charger and manual controls, but not including a battery - will cost $1000. Or you can pay just $750 and bring your own wires, grid charger, and controls (e.g. Peter's joystick, or no controls at all).
 

·
Registered
Joined
·
7,843 Posts
....Linsight will monitor up to 36S cell packs. Said packs can have ANY number of cells in parallel, as long as ALL cells at a particular pack voltage are wired in parallel with leads thick enough to prevent a voltage differential between any said parallel cells....
Thank you very much for this detailed answer. Much appreciated...
 

·
Premium Member
Southern California
Joined
·
896 Posts
I'll admit that $600 was definitely a lowball offer and after seeing what's going into the project, I would now gladly pay $1k for the whole package. $750 for just the PCB is pretty reasonable.

Question, why not have somebody else assemble your boards? Do you have no local PCB fab/assembly houses?
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #252 · (Edited)
Question, why not have somebody else assemble your boards? Do you have no local PCB fab/assembly houses?
All the boards I've quoted out in the past came back way too high, hence I now own my own assembly line. In fact, our manufacturing facility is directly across the street from Silicon Hills, Austin's largest PCB assembler.

Linsight requires 54 separate reels and has nearly 500 components, several of which are through hole connectors (requiring wave or hand soldering). Thus, I suspect the per-build setup fees are around $1500. I also suspect the per-unit cost is ~$250, plus ~$25 for etest and programming. Thus, if I pay someone to assemble Linsight 10 units at a time, I'm looking at paying someone else $425 per PCB to put parts on it.

Since I've already got my own pick and place machine and five-zone reflow oven, I'd rather build the boards myself (or actually pay one of my employees to do so).

Let me know if you have an assembler that can place 500 parts on a PCB and program two microcontrollers, then etest (all in low volume) for less than I've estimated... it's been a while since I've even quoted a board, as I've just been assembling everything in-house. Here's the part placement position file if that helps:
View attachment Linsight-top.pos.zip
 

·
Administrator
Joined
·
13,083 Posts
Pricing

There were multiple responses of "I'd pay $600." After calculating Linsight's component cost today, I know that I can't sell Linsight for less than $700. Even if I buy parts in quantity (i.e. I build 50 Linsight modules), the parts to build each PCB exceed $250! I then have to reel these parts into my manufacturing equipment (each time I run a batch), run each board through a reflow oven, and then I'll spend another couple hours soldering, programming, testing and troubleshooting each PCB, and finally I'll probably spend at least an hour helping/transacting with each customer. My minimum consulting rate is $100/hour, so Linsight just isn't worth the effort to sell for less than $700, particularly since I expect some customers will install Linsight improperly and thus require further support.
People will often expect a rock bottom price and something for nothing. Things take time and considerable effort and expense to develop. I applaud your efforts on this project. You should price your product at a point where you get a fair reward for your efforts. I do.

OBDIICC&C and IMAC&C cost what they do as they took ages to develop, and they provide part of my retirement income stream. People often don't think of the negative consequences of competition or sharing of info. It becomes a race to the bottom and kills R&D etc. If I gave away the OBDIIC&C code and pids etc etc then I would destroy my income and ability to invest in time for more Insight R&D. Don't let yourself be pressured into selling it too cheaply as that will also erode your enthusiasm when things go a bit awry.
 

·
Premium Member
Joined
·
3,661 Posts
Just to re-iterate (from November):

- ...............

More capacity, ability to assist for hours vs minutes, p&p, and a boat load of features my dinosaur self is trying to digest and learn about; this sounds really good. I'm ok with costs up to what I've already paid for a replacement NiMH pack. Would be some time though before I would need to replace a pack.

I'm one of those folk who would need some pointy/talky with pictures to assure correct installation.

Awesome stuff John. :)

btw: 100 + mpg is a strong attention getter for the severely diseased hypermiler. ;)
 

·
Administrator
Joined
·
13,083 Posts
I just realized a mistake in my math here - OBDIIC&C updates the screen four times a second, which is what I was thinking of. But for each update, it requests 8-10 parameters (4 on H-Line and 4 on K-Line, plus one more on each line for some internal calculations I think, don't have my notes with me), so there's actually 16-20 parameters per second being fetched from each line. It is still about 50% of the bandwidth, though.
IIRC I settled at 4hz as the MCM and ECM would not respond if you pinged them much more often than that. So the theoretical bandwidth on the H/K Lines baud may be much greater than the actual rate at which the units will respond to data requests.

There was def some dead time after requests which is why I alternate them, so that one unit is resting while the other is providing data and vice versa. H,K,H,K,H,K,H,K etc etc MCM,ECM,MCM,ECM,MCM etc etc

You could/can requests chunks of memory upto $10 bytes IIRC rather than individual parameters, but that needs careful work and mapping of the memory for each unit. You then have to store that data and manipulate it internally.

It's fairly easy to test the dead time, just keep sending the same pid request at a faster and faster frequency until the MCM or ECM stops responding.

There is also some dead time after a request is recd by a unit and when the unit responds with the data. It's around 10ms for the MCM parameters. The ECM has much less dead time after a request is recd and IIRC it responds almost instantly. <1ms
 

·
Premium Member
Southern California
Joined
·
896 Posts
IIRC I settled at 4hz as the MCM and ECM would not respond if you pinged them much more often than that. So the theoretical bandwidth on the H/K Lines baud may be much greater than the actual rate at which the units will respond to data requests.
Well that's unfortunate, that makes me less optimistic that I'll be able to get full refresh rate for both devices when using OBDIIC&C and Pegasus at the same time. I'll do some testing myself to see how fast I can push it.

You could/can requests chunks of memory upto $10 bytes IIRC rather than individual parameters, but that needs careful work and mapping of the memory for each unit. You then have to store that data and manipulate it internally.
I was actually wondering myself if the PID requests are more like direct memory reads/writes, just hadn't had a good opportunity to try it out yet. Good to know, I will definitely experiment with that. Maybe there are enough parameters close enough together in memory to make this a worthwhile strategy.
 

·
Registered
Joined
·
1,558 Posts
Lithium Titanate cells 2.4V

You're right, it's too tiny for me to offer any feedback. I don't know what the full cell voltage will be, but 72S is almost certainly too much voltage for the insight's VCM and DCDC converter. Even at a low 3.6 V full charge voltage, 72S is 259 V! The maximum voltage before things start shutting down is ~210V.
Luckily, the Lithium Titanate LTO cell's rated at ~2.4V nominally. [ 2.3V or 2.4V listed among various suppliers]

Taking the higher value voltage for the Lithium Titanate battery pack of S72 would have a nominal pack voltage of ~172V.
Using manufacturer specs* maximum charging cutoff is 2.7V at 1C.
Maximum pack voltage following cutoff charge would then be only ~192V.

*fwiw: Other battery sources for LTO cells, list specs indicating 2.75V max cutoff voltage. [1C] [xS72Cells=198V].

Also, there's a charging algorhythm reported listing max cutoff voltage as high as 2.85-2.9V. Using that max cutoff voltage would push the battery pack to a maximum pack voltage of 208V using that highest maximum charge cutoff voltage found for LTO in searchs.
___________________________________________
:biggrin:So, I'm encouraged that running S72 LTO 2.4V cells will be within the operating range for the IMA system.
___________________________________________
Here's the specs for the 'Padre' LTO Lithium Titanate battery cells.
[Their company lists 2.4V 10Ah, 15Ah, 20Ah,26Ah,50Ah,75Ah cells. ]

2.4V 20Ah Lithium Titanate LTO cell specs [Padre Electrical]

Model: TKLD-2.4V/20Ah
Nominal capacity: 20Ah
Internal resistance(AC): < 1 mΩ
Nominal voltage: 2.4 V _________xS72cells = 172.8V
Weight: 685±5g _________xS72cells = 50kg
Dimension 228×108×12.7(mm)
Charging mode: CC/CV
Charging cut-off voltage 2.7V _________xS72cells = 194.4V
Charging current Normal charging 7A
Maximum charging current 60A
Discharging cut-off voltage 1.5V ___________xS72cells = 108V
Maximum discharging current 100A
Operating temperature: Charging -10 -55℃ Discharging -30-55℃
Storage temperature: < 1 month -20 -60℃ ; < 2 months -20 -45℃ ; < 1 year -20 -20℃
Energy density: 70wh/kg 153wh/L
__________

Thanks again for all you are doing; this project is a game changer for the vehicle.
 

·
Super Moderator
Joined
·
9,229 Posts
And thus I propose the following social contract (and am soliciting your candid feedback):
The "social contract" you propose, as well as the costs, are acceptable to me. I think your information more than justifies your price point.

Re MIMA, as I read your permalink 88 your proposal of a one-axis, return to zero manual control is optimum for me. If I interpret correctly that would be a progressive type control with more "push" resulting in more assist - just like I like it.

I still don't want to be an early adapter, since my electronic skills are not current:(
 

·
Linsight Designer
Joined
·
2,647 Posts
Discussion Starter · #259 ·
Question, why not have somebody else assemble your boards? Do you have no local PCB fab/assembly houses?
After my initial response to your post last night, I sent the board out for quote. The lowest quote I've received thus far is $223/board (including setup fees) if I build 20 boards per batch. That's half what I expected, so I'll see what the other quotes are and then I'll consider my options.
 

·
Premium Member
Joined
·
636 Posts
Price seems reasonable to me considering all the time and effort you have put into it, this is a bargain. If it seems too expensive, they can opt for a regular replacement battery which would be around 2K, really a no brainer.
 
241 - 260 of 977 Posts
Top