Another festival has come and gone, Not as many photos as last year but I think that we all had a nice time. Special thanks to Paul, for the use of his garage. We have a new MPG champ this year, Jim Epting after running some practice loops did the route with 134.4 MPG, Mountain Driver got 110.
The IMA battery can be completely removed from the car, and the car will still function, if the DC/Dc converter can have the HV input it requires. The typical way to do this is to remove the whole connector/relay board from the pack, and use it to make those connections. If the fuse and connector from a used pack is available, a much safer and simpler IMA bypass can be construced. You remove the DC/DC connector, the 30A fuse, and make up some wires with ring terminals, and make a short bypass connector with those components. It is important to keep everything well insulated, and prevent ant possible shorts, so here is how I made mine. If you can leave the HV battery in the car, just pull the "A" connector of the BCM and turn off the HV battery switch location of BCM A connector
A typical discharge graph for one of the MaxIma sticks. The little wiggle in the trace is caused when labview stops recording the signal for a few seconds while writing a PDF file, and is not a problem in the battery. We see nearly an additional AH of capacity, a higher discharge voltage (less Internal resistance), and less heating even with the additional AH. All good.
testing to the cell level at currents similar to the car
I revived my labview based stick tester and will be using it to test the new MaxIma sticks. Here is a trace of a stock stick at the 75-80A load. Note that we see 5.8Ah and a temperature rise to over 120 F
Max IMA batteries show improved internal construction
Several people have picked up the ball that was dropped by Hybrid battery repair, and are offering new sticks for the civic and the insights, so it is time that we see just how well the sticks are built and how well they perform. I got 4 sets of 20 subpacks from Bumblebee, and am in the process of evaluating the performance, but we also know that the construction of the cells is a big factor in how well they can handle the currents involved with use in a Hybrid car. The first photo shows the inner cell construction of the MaxIMA cell more and longer weld points on this cell allow the current to flow with lower internal resistance, so in theory that should translate into less voltage drop under the same load as well as less heating.
Wanted to do some charger/discharger test on a pack that was pretty bad, so I took off the covers, and made this chart, which should be a nice road map for any battery explorers. ********Danger********* Adding the jumper across where the big fuse and switch was located as shown here removes any protection from shocks or short circuits, so proceed only if you know what you are doing. May be better to use a DC rated fuse in the jumper to maintain short circuit protection.
After several hours of trying to find the BCM connector that is the mate for the battery tap harness connector,I gave up. Using a pack that had no cells,so the connector would not have power. I took a small clear plastic bag, and carefully taped the cut up bag over the harness connector,and secured it with black tape near the base, where the wires come out. Next I inserted the pins pulled from one of the 104Pin headers used in the MIMA plug and play adapter, through the plastic into the harness connector female pins. Making sure that the pins were inserted fully,I covered the plastic covered connector with epoxy putty, allowing the pins to stick through the epoxy. I taped the epoxy with black tape to hold the shape. When the epoxy cured,I soldered a ribbon cable to the pins, making the order of the ribbon follow the taps from the - end to the HV + end in sequence. An aluminum duct tape dam that was adhered to the epoxy, made a cavity on the rear where the ribbon attached to the pins.The cavity was filled with hot melt, casting the wires and pins in place. I put a 12 pin .1" OC straight male header on the other end of the ribbon. The voltage taps can be read here, or I can plug in another female 12 Pin header, also attached to a ribbon, into a pill bottle(it was handy). I mounted a 20 position break before make dual pole rotary switch on the base of the pill bottle, and two terminals on the cap for the volt meter. I also made the conector for a civic pack, that plugs into the same female header on the rotary switch /pill bottle, so the same switch can do either. To use it, one unplugs the voltage tap connector from the pack, and attached a voltmeter to the two pill cap terminals, and I can monitor each 12 cell stick , in sequence, as well as the whole pack voltage. It took less time to build than I spent looking for the connector, and It can do both a civic and insight. The connector plugs in reliably and securely, after the plastic is removed. Not pretty, but it works well and is pretty rugged and safe.
As anyone that has pulled the battery pack on their insight can tell you, it is just heavy enough to hurt your back, and it requires you to lift in a very awkward position. Honda has a special lifting frame that the techs use, so I thought that with the requirement that one pulls the Insight pack to do a clean grid charger install,it was time to see if a quick and dirty version of the lifting frame could be fabricated. Randall sent me a photo of one he made, using some handles, and a piece of plywood.
I made one with long handles, that allow comfortable lifting angles, and full control of the pack while removing or installing.
Front end accident shifted IMA motor coil knocking it against rotor
I acquired a complete insight motor/transmission from an insight that had been in a front end collision.I was preparing to shim the IMA rotor so I could remove it from the engine,and discovered that the rotor was shifted sideways from the coil assembly to the point where it was rubbing on one side, and the other side had a big gap.
The mass of the coil assembly and severity of the impact seem to have shifted the coil on its mount. I could not see where the aluminum block or IMA housing had shifted, but do see some abrasion on the front of the block ? the problem is that if the casting was bent there the gap would have been on the other side???
When I get a chance I will eventually remove the coil assembly from the casting, and compare it with one that was not hit to see what actually moved?
The point here is that if you get in an accident, the coil can shift and rub the rotor causing some future failure. This should probably be inspected after a front end hit, but I really doubt that it ever is???
Since Hybrid battery repair has abandoned the pursuit of cell level rebuilding of the subpacks, and I have already built the welder, and other fixtures and test systems,that he does not want, I will use the equipment to dig deeper into determining how the subpacks look at the cell level.
A look at the traces below will show that many subpacks only have one cell that is a bad actor, others show overall deterioration that may or may not respond to charging. Time to pin down some of the failure modes, and with the new smart chargers, we should be able to troubleshoot a problem pack and determine which sticks are acting up, and then which cells within the sticks require replacement to get the pack back into a useable condition for the least cost. I grabbed all of the packs and sticks I have, and have assembled them so I can begin the testing. The automatic whole pack discharger is working well with the 6 100W bulbs, so I will run capacity recovery test on all the packs I have available to see what we see.
While most of the nice OBDII code readers can tell us which P code is causing a check engine or IMA light, it cannot tell us the subcode ( Blink code). The photo shows which pins need to be shorted together on the Insight OBDII plug to have the car report the blink subcodes. Short with wire and the codes are reported.
When reading the IMA blink codes, the IMA light blinks and if the check engine light is on, it will blink Example: a 74 IMA code would be 7 slow blinks, and 4 fast blinks, if there is more than one code, the blinking will have a long pause, and then blink out the second code.It will cycle through all the codes in a circular fashion. Say there was a 74, and a 78, and a 63. The IMA light would bilnk the 74, then the 78, then the 63, and then start back with the 74 and repeat until you release the short.
When shooting the MIMA video, I needed a way to hold the camcorder so it would shoot the dash display,with no hands, as I needed both to drive the car. I cut an oak board in a triangular shape with two soft rubber feet for the driver door side,and a square shape that fit into the steel square tube on the other. A locking screw was made by welding a nut over a clearance hole on the end of the square tube. It works just like a tripod leg. The telescoping arm is put between the two front doors and forced open so it presses against each, then lock it. A tripod head was attached so the camera can look at the dash while driving. The MIMA videos were shot using this technique.
Hybrid battery packs develop several problems as they age, which are aggravated based on temperature history, and length of time spent inactive. Inactivity allows the cells to self discharge, and since all cells are going to do this to one degree or another, and the rate of self discharge can be quite different from one cell to another, a typical pack that is setting codes like the 1449,1447,1433, will have cells that have fallen way behind others , and the IMA control system which works to keep the cells in the 20% to 80%, has no mechanism to rebalance them. All Nickle based batteries develop this self discharge because they can spend a lot of time sitting at a middle SOC, where Nickle dendrites (little whiskers of nickle) will grow and gradually provide a leakage path between the + and - plates, eventually leading to a shorted cell.
Storage and self discharge Another thing that happens when a cell drops down in the SOC relative to the other cells is it develops an effectively reduced capacity that is recoverable called Memory effect. memory effect Memory effect capacity loss is reversible by fully cycling the cell from 100% to less than 1V/cell.
So the code setting pack has this unmatched set of cells with widely different SOC, and wide spectrum of effective AH capacities, which reduces the effective capacity of the pack as a unit to a value that sets the code.
When the car charges the pack to full, (80% SOC) it stops when it thinks the pack is full, with the assumption that all the cells are the same, where in fact some cells are still only partial charged. The discharge limits are also skewed so the pack stops allowing assist long before the highest capacity cells are empty because one of the low capacity cells has dropped out. When that happens the rest of the pack discharges through the empty cell effectively reverse charging the cell. Cell reversal The IMA safeties kick in causing the SOC guage to drop to the bottom, and aggressive charging begins to pull the reversed cell back from sure destruction.
The longer the pack is used like this, the worse the pack gets. Another factor in this picture is heat. Once a cell gets to 140F the internal pressure can get so high that it vents potassium hydroxide and the cell will permanently loose AH capacity. No amount of charging can bring back capacity lost due to venting.
The charger and discharger system will first fully charge all cells to 100% SOC, for the first time since the pack was installed. This can take a long time, as we have to limit the charge current so the cells that fill up first do not overheat. When a cell is fully charged, all the charge energy becomes heat, which is why the cooling fan should always be used during charging.
Getting the whole pack to 100% will fix some packs, especially ones that have become unbalanced due to just sitting and self discharging, and have not developed any memory effect. Hypermileing where you do not use the pack much can cause the pack to develop the same self discharge unbalance.
Most packs have both memory effects and real capacity loss, as well as differing degrees of self discharge, so the discharger aspect of the system becomes a great tool to extend capacity on the memory effected cells by cycling the pack between full and the detection of the dropout of the weakest cells in the pack, The idea being that the weakest capacity cells are the ones causing the problems, so since they will be the cells that get the deepest discharge, if the issue is memory effect those cells will recover capacity with each cycle, and the minimum voltage reached during the discharge should get lower and closer to the 1V/c level which would be 120V on a 120 cell Insight or first gen civic. A well balanced pack even if at low capacity will discharge uniformly so the discharge would stop at near 1V/cell. In the real world that is seldom is seen, since all packs will have several of these issues happening at the same time, so there will be a point where the discharge will repeatedly stop at the same elevated voltage, like 138V. At this point the cell that is dropping out is likely one that has lost capacity due to venting or simply age, and to get that pack back in balance, we look at the other cells when that cel drops out to see if they are also ready to drop, or if the majority of them are still quite good. At this point, if the car is still setting codes, we need to take the process from the car to the bench. Using a procedure that we are still tweaking, we can use the charger/discharger, and datalogger to discharge the pack with the ends open so we have access to the ends of each stick, to make voltage measurements. If the other cells are still well over 7V replacing the stick that dropped out with one of higher capacity, would then allow the cycles to continue to reduce memory effect and increase capacity of the rest of the pack.
The photos and graphs below were done with my cell level test fixture and Labview graphing program for my pc. This is a graphical analysis of my Insights Subpacks after I got an IMA and Check engine light. The pack never did recalibrations, and it set the IMA code several days before setting the check engine code. P1447 set the IMA, P1449 set the check engine. A system reset would not clear the codes. Each subpack was charged most of the way, topped off, then discharged while watching the temperature and voltage across each cell. I looks to me like subpack 10 and 12 are the guys that set the code. I hope to be able to replace the cells that are bad, cycle and rebalance the rest and try the pack back in the car. Please feel free to comment on the graphs with your opinions about what they show, my comments are very preliminary since I have only put them through a single cycle so far. I charge @ 5-6A, and can do a high rate charge of 30A. The discharge is at 30A, and can be increased to ~70A. Read More
Ok I can test the subpacks down to the cell level, so the bad cells can be identified, now how do we get the cell replaced. Grabbing the stick with plastic strips in the milling machine vice, I milled off the welds being careful to not machine into the stainless steel case. Once the welds were removed, the cup just slid off the underlying case with hardly any force required. The case only needed a light sanding to make it like new. The two end caps also were quite easy to remove and they were left in a reusable condition. All of this is good news. With a proper fixture to hold the sticks, I could separate the end caps and cells on a whole stick with everything ready to become part of another subpack. No new parts required. Time to get on the welding machine.
Ok now that I have a good electrical test fixture and have identified a cell that was shorted, recovered, but not back to full capacity, it is time to take a look at the insides of the cell. The first thing that jumped out at me was the really nice welds. they are deep, show little discoloration in the heat effected zone,with no surface oxides, so they were probably done in an inert gas. The welds were so strong that the sheet metal tore rather than the weld separating. I made a test fixture to see how much force it would take to "crack" the stick at the cell to cell boundary. Amazingly I was not able to break the welds, even up the the 300 lb max that the scale would take. I had to bend it back and forth 4 or 5 times to get the welds to break out of the metal. I sanded of the weld stubs, and put the cell in a vice to drain out all the energy. I turned the cell in my lathe and cut off the top. The top inner welds were also solid, but I was a bit surprised to find that the top terminal and bottom terminals were simply pressed into the spiral wound plate.The cell must have been assembled in a press to give adiquate contact pressure. I carefully unwrapped the perforated nickle strips which measure about 24 inches. I looked at the plates and separator blotter material, and saw no signs of holes or burned areas(Capacitor Zapp), so this somewhat supports that the capacitor discharge did not blow any holes.Very interesting.
Finally got a look under the black and clear covers of the PTC strip. It turns out it is really 6 discrete PTC devices that are connected by stainless steel strips, so the active temperature sensitive area is located right in the center of the cells.Very interesting, I will run a resistance to temperature profile on one, and see what the actual characteristics are.
Protecting against a short like a self resetting fuse
Several years ago I did some experiments to figure out what the red thingies on the battery packs were, and what they are for. I hit one with a heat gun, and saw the resistance shoot up from ~ 160 ohms to over 600 with a quick blast. I explained my findings on Insight central, but someone else tried the test in an oven, and was not able to duplicate what I found. I finally found myself with a loose Insight side terminal plate in my hand, and decided to put my original theory to a different type of test. I set up my lab supply with 30V, and the current adjustment was set to the 1 A maximum for the supply, so no power supply current limiting would be involved. I clipped two probes to the thermistor leads, and used one of the banana jacks as an on off switch. On first application of the voltage, the current pulsed up to just about the expected 185mA, but rapidly over only a few seconds the current dropped to only 25mA and the thermistor was at 250 degrees F. I held it for a minute, and the thermistor stayed at the 250 f. If the battery tap cable was crushed and all wires shorted, all that would happen would be for the battery end plate to get warm. Nice safety. Swap crushed cable, and the battery pack is good to go. A self resetting fuse.
If a larger capacity battery pack is used in the Insight, we must "fool" the BCM into allowing the full AH capacity of the new pack to be used. I setup a BCM and junction board with current sensor, to determine the best way to fool the current sensor into thinking that only 1/10 of the actual current is passing through it. After a thorough analysis of the system, I determined that the Insight battery current sensor was outputting a current loop signal with ~.20mA/A, and since it is a current loop, a simple current dump resistor to BCM ground should do it.
The battery packs of all hybrid cars have a very special fuse to protect the circuits and the pack from short circuit and over current conditions.An Insight owner came over so that I could help him diagnose his IMA problem, and we found tht the main 100A fuse was blown. He had already replaced the inverter stage, which must have been the cause of the fuse failure. Opening a high voltage high current DC circuit requires a special type of fuse. It is called a semiconductor fuse, but this name is misleading. There are no semiconductors in the fuse, the name describes it's ability to blow extremely quickly so any semiconductor devices it is powering will be protected, and this means rapidly quenching the hot arc that forms when a high power DC circuit is opened. The metal element is similar to a regular fuse, but the fuse is filled with an insulating glass like powder that rapidly melts and quenches the arc.I opened the blown fuse so we could see how it is constructed.
Thanks to Jim Alger at Find My Insight who sent me a stripped out oil pan, I believe I have a better solution to the expensive replacement that many people have faced. This may have actually been part of the design, as I cannot see any other reason they would have put that strange O-ring retainer in the stack of washers. I carefully looked at the pan, and saw that there were at least 4 complete threads in the rear of the stripped out magnesium hole. I quickly realized that if the crush washer, and O-ring plate were gone, and I cleaned the remains of the old threads out of the hex bolt, that it threaded in fully, and actually came out the other side with the full 4 threads to hold on to. I got out the trusty gasket punch set, and made a nice 1/8" thick rubberized cork washer. I screwed the bolt back in, and found that from first resistance(initial crush), to where I felt I was going to squeeze out the cork (not likely)that I had nearly a 1/4 turn. This should seal easily between the hex plugs polished underside, and the nice flat surface of the oil pan. The only question that remained was could it work loose over repeated hot cold cycles. I decided that since there was already a nice blind threaded hole and bolt left over from the O-Ring washer, that I may as well make a hex plug anti rotation retainer, I dug up a thin aluminum sheetmetal plate, and drilled a hole that was just a smidge larger than the distance across the hex head flats of the hardened plug. This hex is 17MM, so as a backing cavity I used a 6 point socket, and pressed the hex head through the aluminum with my drill press. This cold forms the aluminum to make a nice snug fitting hex hole to fit the hex head of the plug. I cut out the aluminum and made a narrow tab, then carefully bent the tab to slightly less than the wide base of the hex plug and the cork washer so it would not bottom out when applied. Finally I drilled a clearance hole in the tab for the retaining screw. I tightened the plug with the retainer in position till the hole lined up, put in the screw, and believe that this oil pan will work as good as it did before stripping out, all without the need to do any machining or removal of the oilpan, and with only the addition of a single sealing washer. Cork rubber is good in oil, and is rated for -40F to 180F. I just happened to have a piece, but a better washer material may be Aramid/Buna-N, which is rated for -40 to 700F You can buy a pack of 5- 5/8" X 1" X .062 Aramid/Buna-N washers from http://www.mcmaster.com/ Part # 93303A284 for $8.10
As any hard core Insight owner knows, the Insight will buck and hesitate under low throttle when the EGR valve either gets clogged, or develops a bad feedback potentiometer.The exhaust gas passes into the cylinder intake by passing through an aluminum casting. Each of the three channels that are fed by this channel has a steel flow control orifice in the path. When we removed the intake manifold on the green machine, this channel was totally clogged, and we needed a screw driver to clean it out. Probably should be cleaned whenever the EGR is serviced.
Even before the rear MPG display, I played with using video cameras as a way to potentially replace the drag inducing rear view mirrors. I had acquired some LCD video displays and color cameras at the MIT electronics flea market. The cameras had a rather narrow field of view, so I set up four of them looking out the lower rear hatch window. I built a joystick operated 4 camera switch so I could easily select the camera. The display worked well except for when the sun shined directly on it, so I had a removable sun shield. The cameras worked very well during the day, or under street lights, but even with the automatic appture control, did not have the low light sensitivity to see much more than headlights of approaching cars. Of course to totally replace the rear view side mirrors, I would have needed two more cameras. With the correct choice of cameras,field of view, and a bright monitor, two side cameras should easily replace the side mirrors. Several years later,at Hybridfest, I saw Bill Kinneys Insight that was set up with exactly that system, which seemed to work very well.
A common problem with the MT Honda Insight, and to a lesser degree the Civic hybrids is a condition called recalibration. The most common version of this condition shows up as a rapid drop in the battery SOC gauge from some intermediate value to empty. The car goes into a forced charge to bring the SOC back up, which if it happens frequently, can take its toll on the MPG. One of the possible reasons for this condition is an unbalanced SOC between the different cells in the pack. The warning label from a civic battery pack warns to run the car for 30 minutes a month. It is likely that the self discharge characteristics of the cells is not predictable, so some subpacks will discharge to a lower level than others, and there does not seem to be a rebalancing mechanism in place to correct the imbalance once developed? The civic hybrid battery packs made 2005 or before are a good source for replacement 6 cell subpacks with the PTC temperature strips attached.
Calpod asked that a regen based brake light activation be designed for MIMA equipped Insights, as the car can slow down pretty fast if full regen is applied. Without the brake lights to warn the guy behind you that you are slowing down, you could get rear ended if he is not awake. The software is already working, and the Violet MIMA aux wire will pull low when the regen is greater than 15A. The problem was how to interface to the car. If the brake switch was shorted by a relay, that would work, but as soon as the brake light is activated, the car will try to apply full regen, as that signal feeds to the MCM. In looking at the schematics, I chose the connector in the passenger side B column as the best place to tie in. This spot is where the drive signal for the high mounted brake light is located. With this circuit, only the high mounted brake light will activate when MIMA or the car is over 15A of regen. Any small 12V relay with SPDT contacts should work.
With the many projects I hope to work on, some extra hands are always welcome. Check here to see if their will be a workshop if you want to come up and lend a hand. Present project: Design, build, and test an IMA-EV mod to give full electric mode to the gen 1 Insight
Back in 2001 when I first started being active in the Insight on line community, I was concerned about the life of the battery pack, especially the issue of recalibrations. Our understanding of the reasons for the "recals" have not advanced much since then. I designed a battery monitoring system that used relays for isolation to read each of the 14.4v monitoring points in the pack. The connection to the pack was accomplished with a special spring loaded back probe that just clipped onto the battery. A Labview based data acquisition card and software scanned the taps and recorded the voltage graphically. I fully expected that my pack would eventually start having the recal issue and wanted to be ready. I am now nearly at 100K on my car, and still have not experienced a single recal. Honda says that gradually accumulating errors in the coulomb counting SOC software is the reason that recal's happen, but I ask what causes the errors? My intuition says that one cell in a subpack is of lower capacity, and since the pack is one long 120 cell series string, that one cell will cause the whole pack to have less capacity than the system expects. When the BCM senses the weak cell dropping voltage as it becomes depleted, the system stops assisting and starts charging to prevent the cell from becoming reverse charged which is instant death. A smart system would have the ability to balance the pack by charging that cell more than the others. Unfortunately since the cell is in a 6 cell subpack, that is not possible with either the Honda or Toyota systems.
The Insight pack is protected by a 100A fuse, and high voltage DC switching relays. The current into and out of the battery is measured by a hall effect current sensor, and the Current into the MPI is measured by a second current sensor.
The insight battery pack is made up of 120 D cell NIMH batteries, grouped into packs of 6 cells. You can see that the air path for cooling the pack is not very open, therefor the cooling of the pack is not very agressive. The pack has a strip of thermal sensing material that is in intimate contact with each cell in the pack. The D cell subpacks are terminated with a large stainless steel nut that can handle the 100A output. This is the special part of the pack, as we have not found a D cell that has similar high current contacts, so cell replacement with new subpacks is not an option at this time. The civic packs up to 2005 are drop in replacements, so you could get one of those much more readily available used packs and rebuild your Insight pack with the civic subpacks which are direct replacements with a more robust cell structure. The 2006 and up civic packs are built differently and are not direct replacements.
Genesis One, LLC
If you would like to get involved or support any of these projects, please contact me at (860)935-5569.