(Was) Honda Civic Hybrid goes 100,000 miles with no problems

[dnmun]

richard may know more than me, but i think the electric assist is actually added in between the individual combustion events and not continuous, so there is a little torque added to assist the motor 3 times per revolution on the insight, which is 3 cylinders. maybe luke knows.

 

[grindz145]

This is awesome info to have. I'm considering the civic hybrid as a next car for my girlfriend when her 97 finally gives up the ghost. However thanks to Honda engineering, it has 160k and still gets over 30mpg, and shoes no signs of stopping

 

[NickF23]

I think I can tell if I have a weak cell in or two in my 40 cell nimh ebike pack from the voltage 'bounce' after a full discharge. For example last week I discharged the pack down to 30 volts at c/20 and within 5 minutes pack voltage bounced back to 48 volts, indicating most of the cells are fine and just one or two dead cells is causing the voltage to collapse under discharge I'll have to see when I open it up, but it should be obvious where the bad cell is from the cell voltage.

 

[fechter]

richard may know more than me, but i think the electric assist is actually added in between the individual combustion events and not continuous, so there is a little torque added to assist the motor 3 times per revolution on the insight, which is 3 cylinders. maybe luke knows.

When the boost is active, it works just like any brushless motor. It doesn't turn on/off with the combustion events.

After the third charge cycle, it is definitely showing signs of improvement. I can make it to the top of the freeway hill without running out of boost now. It seems to be absorbing a lot more regen now too. It used to crap out on long downhills as the high cells topped out. I probably still have some weak cells, but I'll run it for a while and see how it does. My balancing charge cycles have been a little on the short side (12-14 hrs), as I still need to drive the car. Some guys report it can take 24hrs to fully balance the cells.

I'm planning to build a impedance tester that will be a high current load for the pack that switches on and off at something like 100Hz. While this is hooked up to the pack, I can measure across each cell module and look at the AC component of the voltage. Cells with higher impedance will have a higher AC voltage, which I can measure with my DVM or oscilloscope. The idea is I want to identify the weak cells as quickly as possible so I can remove the pack, swap out the weaklings, and replace the pack in one afternoon. It's quite a bit of work to take the pack out of the car and take it apart to the point where you can replace the cells. Lots of screws and wires to deal with. It's pretty straighforward though, and not really difficult, just time consuming. Honda designed the thing so it's pretty safe to work on as long as you know what you're doing. There is a main pack disconnect switch that's mounted mid pack, so as long as the switch is off, there is no high voltage on any of the exposed terminals. Once you get the pack out, you can unbolt all the jumpers on one side of the pack, which will render the thing pretty safe.

There are something like 200,000 of these cars out there and nearly all of them will eventually develop a problem with the main pack. I suspect most people are not going to want to spend $3k+ to replace the pack when it fails, so used ones with pack problems may be quite cheap at some point.

 

[fechter]

Progress report:
So far so good. The IMA light eventually went out by itself and the system is getting recalibrated. It takes quite a few driving cycles for the computer to get the pack calibration dialed in. I'm going to run it for a while without charging and see how it holds up.

 

[GGoodrum]

Too bad Lithiums don't balance this way. It would've saved us a couple years work.

My wife's Escape hybrid has basically the same setup as the Prius. The motor is definitely large enough to power the vehicle by itself. If you don't stick your foot in it, you can keep the ICE off all the way up to 40mph. I've done a couple tests, to see how far I could go on battery-only and the max is 4-5 miles, with some of that being downhill.

I've been waiting for the outfit doing the Prius plugin conversions to do one for the Escapes, but who knows when that will happen. I may eventually try to do a sealed LiPo pack add-on myself, similar to what I did on my GEM booster pack, but on a bigger scale. We'll see.

-- Gary

 

[grindz145]

Just talked to myfriend who had a first gen prius and he has 180k on it without having to touch the battery. The prius system might not complain about the capacity loss the same way that the civic hybrid does though.just another interesting data point.

 

[fechter]

When charging the pack, the thing gives off a lot of heat. Most of the Insight guys wire up the stock fan to an external 12v supply to cool the pack. That looked like too much work for me, so I took a cardboard box and mounted a fan on it. This blows air through the pack like the stock fan. The fan wire is routed from the trunk through a handy opening for the child seat attachment point. Pretty ghetto, but does the job nicely.


Here's the charger setup in action:

 

[grindz145]

That is so sweet Richard! What do you think the overall improvement is during your commute, when you start with a fully charged battery?

 

[fechter]

That is so sweet Richard! What do you think the overall improvement is during your commute, when you start with a fully charged battery?

Hard to say this soon. I need to run it a while to get some mileage figures. The total energy in my pack is not that much, so it can only do so much for my commute. If you had a shorter commute or one with fewer hills, the mileage improvement might be better. Guys on Insight Central have a hack for the IMA that allows a higher amount of assist that would allow you to use more of the battery and less gas, sort of like the plug-in Prius setups. With one of these, and daily charging, you might see a noticeable improvement.

Right now, I'm just running it like normal to see how it behaves. I think the pack improved quite a bit, but I suspect it is still below the rated capacity due to bad cells. Next step will be to weed out bad modules and replace them with good ones.

 

[grindz145]

Cool. Not sure high-voltage medical current limiting supplies are in the toolbox of many mechanics, so I can see how it can be a bit of an undertaking.

You've got me on this hybrid kick now. I found a kit for 3 grand made by a company called "Engineer" that comes with a 3 kwh LiFePo4 battery that makes it a plug-in hybrid, and suppos edly bumps the fuel economy to 80-100mpg (in the right situation anyway) that's got me looking for old Priuses...

I really hate four wheel vehicles to be honest, but getting a giant battery for my car gets me just a little bit excited about it...

 

[fechter]

OK, after about a month of driving without any charges, the IMA light came back on indicating deteriorated battery module again.

I had two modules in the garage, so next step was to find the weakest two and replace them. Most guys take the individual sticks out and test them separately, but that would take like a week for me and I don't want to down the car for that long. So how to find the weakest modules quickly?

So, I came up with this plan: Do a full equalizing charge first, then drain the pack and see which stick runs out first. The ones that run out first have the lowest capacity. Charging was easy since I had that already wired up. Dishcharging shouldn't be too hard either. Normally if I just let my wife drive the car, it comes back near zero. Well, I didn't want to wait for her either, so I drove the car up some hills on the freeway, then put it in neutral on the way down so it didn't regen. I repeated this until the SOC indicator was down to 1 bar. This should be nearly empty and done under actual operating conditions.

Next step was to pull the pack and open it up so I could probe the individual modules. It's alot of screws and bolts, but nothing really tricky or hard. Once I had the ends exposed, I wired up a 120 VAC recepticle to the pack ends, and placed a jumper across where the pack switch goes. Pack was reading around 156V at this point. I wanted to put a good load on it, but couldn't find my old 220V stove top element, so I tried using an old 1000W hair dryer. When I turned on the dryer, it made a snapping sound and some really stinky smoke came out of the handle. Guess it didn't like DC. Oh well, next I tried a 100W light bulb. I figured it might burn out pretty fast, but I had a pile of them I pulled out when changing to compact fluorescents. It lit up OK and was really bright. While this was discharging the pack, I constantly measured all the 20 cell modules. I figured when the first one hit 6.5V or so, it was done. The light ran for about 2 hours, so there were quite a few watthours left in the pack even though it was too dead to assist the car.

While waiting for the cells to die, I took apart the hairdryer for failure analysis. It seems all the guts were OK, but the switch arced and did something nasty inside. After togglig the switch a few times, it started working again. With the switch already on, I then plugged it into the pack. Wow, that thing was like on turbo overdrive. It really sped up the discharge process as well.

Finally, the winners (or losers) were becoming obvious and I concentrated the voltage measurements on those. Unfortuantely for me, it was a 3 way tie, with another close behind and I only had 2 new ones. Oh well. I replaced the lowest two, then reassembled the pack and reinstalled it in the car. If I get more modules, I already know which two are the next weakest. The weakest ones would were in positions that I figure had the highest temperatures on average. Not on the outside edge and closer to the top.

After installing the pack, I did another long equalizing charge to make sure all the cells were up to full and balanced. We'll test again and see how it goes...

Pack removed from car-

All the relays and sensors are on one end. This stuff all comes off with 10 bolts.

Looks like this underneath: View attachment 2

Ghetto test setup.

Removing the other end of the pack exposes the ends of the sticks and makes for easy replacement:

 

[fechter]

I'll have to drive the car a bit more get better data, but so far it looks very good. It seems to have gotten out of the 'reduced capacity mode' and I'm getting full boost and regen again. Not bad considering there are two more weak modules in there. Going up the big hill on the freeway, it still runs out before the top, but it's close. I notice the SOC stays higher now and I get a lot less charging during travel on flat ground, which should improve the gas mileage.

I took one of the bad modules and ran it on the Cadex battery analyzer just for fun. Surprisingly, it came back with 84% of the rated capacity (at 4 amps charge and discharge). It did give me a failure code that translated to "charge current low", so I assume the stick voltage got too high during charge. This is consistent with my observation that the IMA faults happened at the end of a long charge period.

 

[fechter]

Progress report:

With the two new sticks installed, the car is running almost like new. I think it will be up to snuff if I do two more sticks. The IMA system seems to be back to full boost/regen now and the SOC consistently stays higher. Gas mileage is better now too.

 

[peterperkins]

RU interested in assisting with the Civic IMA control and Battery Gauge project?
I've started it on IC and am looking for help in data gathering

http://www.insightcentral.net/forums/other-honda-hybrids-discussion/19788-inside-1st-gen-civic-mdm.html

You would get a free pcb/kit if we can make it work :wink:

 

[fechter]

RU interested in assisting with the Civic IMA control and Battery Gauge project?
I've started it on IC and am looking for help in data gathering

http://www.insightcentral.net/forums/other-honda-hybrids-discussion/19788-inside-1st-gen-civic-mdm.html

You would get a free pcb/kit if we can make it work :wink:

That sounds cool. See you there!

 

[TylerDurden]

So, I'm wondering if the "toolpack mentality" will apply to these high-mileage Hondas:

• Pack looks duff, so owners dump em cheap?
  We fix em cheap?

Or, do owners suck it up and buy entire new packs?

 

[fechter]

I don't think so.
Do a search on Yahoo Autos for used Civic Hybrids or Toyota Priuses.
I was 'shocked' at the prices (pun intended).

Makes me think most owners were replacing the packs when they die. A car with a bad pack should be cheap, but maybe there just aren't any around yet. Some of the older ones that failed early had packs replaced under warranty, so Hondas with bad batteries may be extremely rare at this time. This will change soon I think. There were hundreds of thousands of them sold. http://www.electricdrive.org/index.php?ht=d/Articles/cat_id/5514/pid/2549 Way more Toyotas are around.

I don't think there are many places to get your pack serviced other than the dealer either. I could do it in my driveway. One day service in most cases. Perhaps a business opportunity that's developing. Most people end up paying $3000 to $5000 to get the pack replaced at the dealer. I could do it for *much* less.

 

[fechter]

For those that want to try balancing or testing Honda batteries and can't find a cheap elecrophoresis power supply, here's a power supply you can make yourself. This has the advantage of being much simpler to operate.

4 MeanWell constant current power supplies in series. The supplies are available from a number of places. This place seems to have the best pricing: http://www.trcelectronics.com/Meanwell/lpc-20-350.shtml $16.57 each

This will make a 192V, 350mA constant current supply, which is perfect for balancing the Nimh pack.
For Toyota batteries, you would need a couple more supplies as they run a higher pack voltage.

The diodes are not critical. Almost anything with a 600V, 1A or higher rating should work. The pack blocking diode should be mounted close to the pack + connection. This will prevent bad things if the wire shorts out or touches you by accident.

WARNING: the output of this supply can easily KILL YOU. Don't attempt to build this if you are not familiar with high voltage safety precautions.

 

[fechter]

Well, after replacing two more sticks with ones from a 'donor' pack, performance seems to have actually gotten worse and the IMA light came back on. When testing the donor pack, it appeared ALL of the sticks were weak, but somehow I didn't want to believe that all of them could be bad so I went ahead and used some of the better ones anyway. Looks like I need to find a new donor. I'm not sure what could have happened to a pack that would make all the cells go bad. Maybe storing it at an elevated temperature for a long time or something?

It also appears that during long downhills that give an extended high rate charge, the pack is overheating or perhaps going overvoltage on a pair of sticks and the charge current folds back toward the bottom of the hill. After driving on the flat for a while, it seems to cool off and the charge/assist rate goes back to normal.

In testing the packs, I was going by the first one to drop out during high rate discharge. Perhaps a better indicator is the amount of 'reverse sag' you get during high rate charge, or a combination of discharge sag and charge reverse sag. Good cells have the least amount of sag.

It would be nice to find a source for new replacement sticks instead of relying on questionable used sticks from junkyard wrecks.

 

[liveforphysics]

If you put new sticks in a NiMH pack, its just as bad as putting in old sticks.

 

[fechter]

If you put new sticks in a NiMH pack, its just as bad as putting in old sticks.

I don't know, not if I replace all of them. :wink:

Even if I did a partial replacement with grossly mismatched sticks, I think as long as the new sticks are paired so that the pack monitoring connections are across them, it should prevent them from driving the weaker sticks into the dirt (cell reversal). Of course the pack is still limited by the weakest cell in the string, but having some stronger ones in there shouldn't necessarily cause a problem as long as the BMS does it's job. Mismatches in self discharge are not going to be easy to avoid, so periodic balance charging may be needed to prevent problems due to self discharge.

 

[liveforphysics]

This Honda pack has no active BMS, it can only observe and choose to do a "re-calibration" if enough mis-balance triggers it.

This is when it drains the pack down to zero volts (or 0.7v x2 since it uses IGBT's in the controller and they have a Vf) If you have some strong cells, this forces reversals on the weaker cells during the re-cal.

When charging, if some cells are stronger than others, they won't crest the voltage hill at the same time and drop in voltage with the other cells to show the BMS that the pack is nearly charged. This peak-watching is the only way it knows when to terminate charge.


The only way Honda's pretty damn crummy pack can work, is if the cells are matched extremely closely. Sticks of better or sticks of worse both cause substantial problems.

 

[fechter]

As far as I can tell, the processor looks at the voltages on the 10 pairs of sticks and will terminate charge or discharge if any pair gets too high or too low. The pack is never allowed to drain all the way down. In addition to voltage monitoring, each pair has an air temperature sensor and every individual cell has a PTC to detect overheating. There are also 4 or 5 random sticks that have thermistors taped directly to the cells (not sure what good that does).

During charge, the charge current can be high enough that the voltage depression at end of charge will get masked by the 'reverse sag' of the cells, which is significant. They have an awful lot of sag during discharge too. I think either voltage or temperature can limit the charge or discharge.

Recalibration occurs when a module voltage hits the limit before the SOC coulomb counter thinks it should have.

I was amazed at how much sag there is during discharge. It would run a lot better with a bunch of Lipo cells in there :wink: I'm not that brave or that rich to try it though. A123 cells might be a bit safer.

 

[MitchJi]

Hi Richard,

The fan wire is routed from the trunk through a handy opening for the child seat attachment point. Pretty ghetto, but does the job nicely...

Ghetto test setup.

Nothing ghetto about what you've accomplished. Excellent work!

It would be nice to find a source for new replacement sticks instead of relying on questionable used sticks from junkyard wrecks.

OTOH using old sticks from junkyards obviously isn't a great solution. Since its easy to get better quality Cells than OEM using new better Cells is obviously the best way to go.

An example of using better than OEM cells (8ah vs 6.5ah with lower resistance):
http://www.insightcentral.net/forum...cussion/19950-ima-betterbattery-revealed.html

There is nothing magical about these batteries. They are high-build-quality replacement cells with the following three advantages:

• 1. They are higher capacity
  2. They have a lower internal resistance than the stock cells
  3. They are brand new and have not been sitting in a warehouse for years

#1 results in a longer period of time before battery depletion. It also means that under the same load, these cells will be less stressed than stock cells. Why? Because the cells are run at huge amperages, and a 32.5A discharge is 5C for a stock cell, but only a 4C discharge for these cells. That is less work for the cells. Why is the capacity higher? The plates are larger in these cells. More surface area = more Ah.

#2 has wide-reaching effects. As you can see from Eli's discharge tests, the voltage doesn't drop as far under load. This increases the usable range of the pack. It also means that the pack responds more uniformly under varying levels of discharge. Honda's BCM seems quite capable of noticing this fact and adapting to it. Why? Because as cells age, they increase their internal resistance (the reverse of these cells). Honda obviously designed the BCM to be able to adapt to what it measures instead of forcing it to follow a rigid set of parameters.

** #3 has the least effect. There is anecdotal evidence that replacement packs are lasting for shorter periods than they used to, but nothing concrete. In addition, there is no way to know the actual build date of the cells in Honda's replacement packs. I can tell you that the BetterBattery cells are fresh off the assembly line.
** [Mitch disagrees with the "least effect" statement, should be least proven effect.]
-----
In the past my speed has dropped to 40 or so on the uphills in the "slow vehicle" lane in third gear. Not this time! My speed dropped from 80 to about 75 going uphill while I stayed in fifth the whole way (and no MIMA at all). Naturally at those speeds the MPG was low, but I did average 64mpg.

Better Cells are easy to find. How difficult or expensive would it be to have six high quality NiMH D Cells welded, or otherwise fastened together in a satisfactory manner?

I was amazed at how much sag there is during discharge. It would run a lot better with a bunch of Lipo cells in there :wink: I'm not that brave or that rich to try it though. A123 cells might be a bit safer...

Guys on Insight Central have a hack for the IMA that allows a higher amount of assist that would allow you to use more of the battery and less gas, sort of like the plug-in Prius setups. With one of these, and daily charging, you might see a noticeable improvement.

With a plug-in booster pack, properly configured I bet you could get a lot of the benefits of the hack (which AFAIK doesn't work with the Civic) of the hack. IE not having the motor start charging the pack going up hill, at least until the booster pack was depleted.