[ttsgeb]

I told @hurricos to make this thread so we could ask questions as they come up and document what's been done, but he wasn't actually ever going to because he doesn't frequent this place yet and would never remember.

I gave the insight bug to my friend. First I put the EV project at his house, then I let him drive my car while I was using his CRV to get parts. Fast forward and he has his own now, but it has a laundry list of issues.

Brakes don't work right and it's a northern car, so I'm pre-emptively changing out the brake lines for stainless steel assuming everything goes alright.

There's a bit of a stumble when pressing the gas, possibly an EGR plate problem?

The shifter for the CVT feels very wrong. It doesn't have a solid detent anywhere and is super stiff to move. Maybe just regreasing? Maybe something more?

There's a hellacious exhaust leak seemingly from the upper cat. Does anyone know if the cats are the same between the earlier cars, or do the CVTs and MTs have different setups? Do we have a good source for these?

Hood latch needs to be rebuilt and the bracket for the release straightened.

The panel gap on the tailgate is BAD, seemingly from the factory.

That's all I can think of right now. I'd appreciate any advice.

Here's the victim behind my my EV conversion:

 

[ttsgeb]

Spent all day futzing with brake lines. Found at least one reason for the terrible brakes, the driver's front line was rotted out and wet with brake fluid.

We also found out that the kit we ordered was actually for a MT car despite being listed as universal on the website, so after talking to support for a bit, they were happy to work with us.
The CVT has a creep assist module between the master cylinder and ABS. This, naturally, means that the lines are different because they're cut in two. I carefully removed the original lines, which surprisingly hadn't rusted out yet, and wrapped them. I then carefully packed them in the box with the lines that were wrong for return and creation of a new set. Hopefully they'll arrive in good enough condition for the vendor to measure them out and have the appropriate kit as an option in the future.

We also found out that this car has Nissan calipers on the front, so that's a bit of a bummer. I'd be interested in knowing what car they came off of, though.

 

[hurricos]

Josh G., a mechanic at Fine Lines / SSTubes, was willing to hand-bend the right tubes if we sent in the old ones. He emailed me yesterday to say they'd be done today. We'll see about that.

 

[Joule]

So Fine Lines does not currently offer a set of ss lines that fits a CVT?

 

[ttsgeb]

So Fine Lines does not currently offer a set of ss lines that fits a CVT?

Not that anyone knows of. Hopefully they'll be able to take measurements from the lines I sent in, and won't just make the one set and then throw them out.

 

[jime]

Welcome to IC and good luck with your Insight. Looks like you have an able assistant and that will be handy. If you put "CVT shifter" and various terms like that in the search line at top of the page you may fine some threads which speak to your problem. Good luck

 

[Joule]

If you speak with them again, please let them know I may be interested in 2 complete sets for CVT.

 

[hurricos]

If you speak with them again, please let them know I may be interested in 2 complete sets for CVT.

Hey @Joule,

I called Fine Lines and spoke to Josh Grayshaw, a mechanic there. While he hand-bent the variant tubes to my creep controller, he has confirmed that he has set aside an identical set for his coworker to program into their system.

He will send me an email once the programming is done and a listing was available for the CVT variant of tubes. I'll post the listing once I get it, it'll be similar to this one, but with 9(?) lines instead of 7(?): CBK0101SS - 00-04 Honda Insight; Complete Brake Line Kit; Stainless

He was aware of how few of these cars were actually made, but I let him know that they will be seeing a lot of business in the short-term. Without even having a listing, we've already got half a dozen folks interested; so I estimated they'd get 20 or 30 more sets of orders for them in the near future.

 

[hurricos]

Lines were replaced on Friday. They are doing fantastically.

To commemorate, I took my Insight on a long drive last night. I knew from @ttsgeb's Insight what it looks like to "put your foot through the IMA" -- intentionally force pressure on it until the SoC sags, as a testing manoeuvre. After about 2.5 miles of forcing assist, the IMA decided it was time to Say No:

Dug into python-OBD this morning with @ttsgeb's OBDLink SX, and this piece of Python:

import obd
connection = obd.OBD('/dev/ttyUSB0') # auto-connects to USB or RF port
cmd = obd.commands.GET_DTC
response = connection.query(cmd) # send the command, and parse the response
print(response.value)

yields only sadness:

(pyobd) [[email protected] pyobd]$ python3 ./get-codes.py 
[('P1444', ''), ('P1449', '')]

I have the spreadsheet that shows all the code mappings ... - https://docs.google.com/spreadsheets/u/1/d/1bLUyuFXWywH0PIzTXYtT6UprBNkOtjnWh5PFKjDFylA/pubhtml

... so I read this as:

• P1444 (High Voltage Short Circuit) - P1444 IMA High Voltage Short Circuit
• P1449 (Honda Battery Module overheating, or Battery Cell Overheating, or battery module Individual Voltage) - P1449 Code Questions / Options

I need to do some paperclip magic to get the blink codes, but I'm going to do that only after I've cleaned out the EGR, because at 270K miles it's probably full of **** enough to explain the low-RPM lurching.

 

[*sean*]

NICE.

So PyOBD works on the diagnostics port that the MCM hangs onto. Awesome.

It sucks that Pi Zeros are currently unobtanium, because building an IMA code reader for about $30 in parts would be awesome and stupid-easy.

I am replacing my radio with a touchscreen backed by a Pi. Adding this interface is now on my to-do list.

 

[hurricos]

EGR plate was almost completely caked:

Tried washing it with alcohol as per

...

... then gave up and went with a screwdriver, dental pick and HOT soapy water (dried out with a hairdryer) before purging it with alcohol.

The result:

Reinstalled everything, booted the car. Idle is down to 895 RPM (did not measure the idle beforehand but at no engine warmth did I idle below 1K). Lurch isn't gone, but I think that's because I have no IMA -- what is gone is the skipped "putt" when bumping the gas while in neutral.

 

[hurricos]

Went and pulled the IMA flash subcodes, following the paper-clip instructions from this thread:

How to read IMA (and other) diag codes without a scanner.

I've been playing the "free new battery" lottery for a while and finally got the IMA light to come on. Not sure if this would result in a new battery or $50 charge to clear the code, I read the service manual about reading the IMA codes. I talks about grounding the SCS line in the ODBII...

www.insightcentral.net

I have IMA flash subcodes 58, 59 and 78.

 

[hurricos]

I cleared codes with

import obd
connection = obd.OBD('/dev/ttyUSB0')
cmd = obd.commands.CLEAR_DTC
response = connection.query(cmd)

... then restarted the car, which pulled back in only the P1449 code from the MCM/BCM.

I reset my IMA by pulling and replacing fuse 18. IMA light is gone, assist and charging work again. There's no doubt in my mind that the IMA is not healthy, but at least I can confirm the failure isn't catastrophic (yet).

I plan to piecemeal rebuild the pack. I have:

• An order of 6 less-than-perfect-but-tested sticks coming (listing: Single Tested OEM Honda Hybrid Battery IMA Stick w/PTC - for Civic or Insight | eBay)
• Recollections from demonstrations by @*sean* et al. at the Frederick Insight meetup last month RE: battery variance (don't worry, I don't plan to measure my IMA pack voltages from WITHIN the BC ...)
• HV gloves from Grainger
• A Fluke 12E+
• This rebuild guide: Hybrid Battery Pack Removal, Disassembly & Rebuild Guide
• A healthy dose of caution from @ttsgeb

My goal is to pitch the worst 6 sticks so I can really put my foot through the IMA without it dying, and wait for @mudder to get the LiBCM demo I saw at the Insight meetup ready for beta-time (if it isn't already).

Question: Does anyone have recommendations for what state I should put the battery in to make it easiest to tell bad sticks from good ones by their resting voltage? Bonus points if the answer is "put my foot through the IMA again", ultimately discharged cells pose less danger.

 

[hurricos]

OK, I've pulled my IMA pack and tested it. I'm a bit baffled, at 18-bars SoC, all sticks read 7.83 +/- 0.03v, but even after drawing the pack down to 145V, the sticks are still balanced!

I dug around the forum for sources on stick testing, and found only this thread:

Battery Pack Rebuild

I am on a budget with my Insight and the grid charger didnt fix it, Is the battery rebuld guide on Ebay worth getting? and if it is worth getting, how much can I expect to spend on parts to rebuild it? what all is involved? is it mostly just battery sticks?::(

www.insightcentral.net

It sounds like the pros perform high-current testing, so that's what I plan to do.

The other day, I drafted a design I call the IMA/BAP (IMA Battery, Assisted Profiler) -- see Martin Kennedy / IMA Battery Assisted Profiler -- and with no options left to test the batteries, I'm going to implement it.

To keep things easy, I'm actually going to leave the batteries in-situ, and work with pairs of sticks; the reasoning being:

- I haven't been able to find reasonable high-current 7V loads, though at >12V there are more options

- single-stick loads require the end-user (me) to tear down both sides of the pack

- High-current 9.5V power supplies are difficult to find; high-current 19V power supplies are plentiful (high-power Thinkpad bricks do 12A @ 19.5V)

- It is ultimately faster to test pairs of sticks at once

- I get a stick cooler for free!

 

[hurricos]

Update: I drained the pack down to 135V using a 60W bulb overnight. Some worse-off sticks were beginning to come out of the works.

I ran tests both under load and after load, and ran those tests in both directions down the pack so that the loss of charge or rebound during testing time could be counteracted by just taking the average. (I was able to test voltage under load by measuring pairs of sticks.)

The results are here:

2001 CVT IMA Potentials

Load, not-load, pair+tap test Load and measuring order,06/09, S/R,12/15, Q/P,18/19, O/N,16/13, M/L,10/07, K/J,04/01, I/H,02/05, G/F,08/11, E/D,14/17, C/B,1, H,3, T,4, I,20, A 6,18,3,9 Off, forward,9.84,12.6,13.58,12.36,13.28,13.92,13.72,12.92 Off, backward,10.24,13.24,13.65,12.42,13.35,13.93,13....

docs.google.com

I wasn't able to find a proper diagram which indexed the batteries, so I created this index (the sheet is defined against it). From the right hand side, facing the fan:

.. 03 06 09 12 15 18
01 04 07 10 13 16 19
02 05 08 11 14 17 20

 

[hurricos]

I've pulled and replaced six sticks (I had only ordered 6). There were some others that might have benefited from replacement, but I purchased pretty cheap minimally-tested sticks, so I don't even know the quality of those (though I suspect, and will confirm with a load test after grid-charging, that they are better than the sticks that I'd just pulled).

 

[hurricos]

To note, I am intentionally skipping high-current testing for expediency. I really want to see how much better my pack works, in particular whether I can do a full charge cycle :^)

 

[hurricos]

Grr. I've been impatient. I grid-charged for two hours (from 146 -> ~155V) and tried to start the car before fully re-assembling the rear cover and all -- it started but threw codes (P1568, didn't check the blink code). I reset that, but now I just get P1449, blink 74 and 78.

Then I realized what's going on. When I rebuilt the pack, I combined sticks at 8.0V with sticks at 6.9V. The BCM must be seeing this and saying, "wow, your batteries are ****ed". Whence the IMA telling me, yes, in fact, I am assumed to have battery degradation because the voltage difference is too high.

Oh, I also realized my battery fan is dead. I should fix that.

One other thing to note ... I didn't see a green temperature sensor wire. At least, I don't remember seeing one. Looking through, someone has definitely messed with this pack quite a lot, though the sticks were very clearly all original.

 

[hurricos]

Reassembled this morning after 12h grid charge to 168V, IMA good (recals to working on start, no light).

I push it hard, immediate fallthrough. P1449 @ 78.

I reset, push it gently through the neighborhood. After about 1 mile, falls through, P1449 @ 78.

This is substantially worse than the original pack's performance, but as the saying goes, the battery is only as good as its worst stick.

I'm going to pull the battery and do another bench test, and see where the replacement batteries stand. I should have been patient enough to test this at the time with the full pack in series; sadly I was not.

The sticks I purchased: Single Tested OEM Honda Hybrid Battery IMA Stick w/PTC - for Civic or Insight | eBay

I think I know which one it was :eyeroll: ... All of the sticks had apparently been calibrated to 8V exactly, and then, according to the listing, bench-tested under load. The sticks had two numbers written in sharpie on the side, probably voltage after 15s and 30s of 80A load. However, one stick was marked lower than the others -- 5.12 / 4.66 -- and the voltage across was 7.89 while the others had stayed up around 7.98 +/- 0.01V.

It remains to be seen how all of these sticks perform during light load.

I've purchased all of the parts needed for my IMA/BAP; if sussing out the bad sticks under light load fails, we'll move to that.

 

[hurricos]

My new sticks are still substantially (+0.20V) better charged than my old ones. That might be what's failing my pack; it's just out of balance. That must show up pretty strongly to the IMA when it tries to pull a lot of power.

I'm gonna go get a higher-wattage bulb and drain the pack to 0, then grid-recharge over another full day, to see where we get.