2000 Honda 5spd Insight + its modifications


1890 lbs., with A/C


0.995L aluminum head and block, motor-balanced offset-bore 3-cylinder

12V SOHC VTEC-E with extreme lean burn (up to 25.8:1 air:fuel ratio)

Tuned-port sequential PGM-FI works with Integrated Motor Assist (IMA)

10 hp permanent magnet synchronous 3-phase motor bolted to crankcase and crankshaft

Air blower cooled IGBT invertor

1kWh 100A out, 50A in 144 – 167V string of 120 Panasonic NiMH D cells


Stainless steel w/ resonator

Manifold integral with head casting, includes coolant passage

Primary standard convertor bolted to manifold/head, heated O2 sensor

Secondary nitrogen reduction convertor with wide-band dual O2 sensor


Hydraulic clutch with wide-ratio (thick rod-like) cable-actuated 5-speed


Aluminum alloy unibody, thermoplastic front fenders and door skins.


2-seat hatchback


0.25 coefficient of drag ('85 CRX HF: 0.32, '96 EV-1: 0.19, '10 Civic Si: 0.25)


Front strut, rear early Rabbit-like reversed floating tube with passive rear steer


165/65HR14 Potenza RE92, which actually has decent enough traction and is sturdy enough to take 60 psi in normal use without any issues, wear or otherwise

Compares to

CRX DX, but with a silent rechargeable source of real low-end torque and no vibration at idle or, well, generally ever really. No noticeable chassis flex either. Uses “silent torque” instead of lots of stick rowing, with rpm generally staying in 1600 – 2500 range being perfectly reasonable.

There's a somewhat higher hatch deck height, since there's a 1kWh power station back there under a sturdy dual-layer aluminum sheetmetal floor. Nil wind noise, less than even in a CRX, and coasting continues seemingly forever.


Braking (maximum recharging of 100A), regenerative engine braking (50% recharging rate) with amperage decreasing in lower gears,

plus home-made 2-stage grid-powered home charger


2-stage grid charger:

A pair of simple “Debrowski chargers” in parallel, one set to slightly higher voltage than the other

Each consists of 3 Meanwell RS-25-48 constant voltage (48VDC adjustable) power supplies and a LS-350-20 constant current (350mA) 48VDC power supply, with the 115VAC inputs all in parallel and the outputs all in series for each “Debrowski charger”. One is set to 168VDC maximum, the other to 176VDC maximum; as the cells approach the threshold between full normal voltage and a balancing top-off charge, current ramps down from 700mA to 350mA. There is also a 12VDC wall brick used, to power the battery pack fan via the 3-level thermostat circuit I made to replace the (in my Insight's case) unreliable stock fan power supply, which is also designed to needlessly shut the fan off during idle Auto-Stop. That 12VDC wall brick power supply I also have connected to the battery condition monitor (BCM) power input leads so that it can track the power being fed to the battery pack as it is charging so it can do its coulomb counting during the charging, negating any need for a state of charge positive recalibration event (a “+recal”). Avoiding a “+recal” is good since a +recal requires the battery pack to maintain at least 180VDC for a few minutes (170VDC for the earlier original BCM). The reason 180V is high is that it's the maximum voltage the pack can continuously receive for charging (120 cells x 1.5V maximum during charging). Giving the pack much beyond that risks damage, but that's what happens during regenerative braking during or soon following the positive SoC recalibration event. 170V is much gentler, since the maximum rated output voltage for the pack is 120 cells x 1.4V max for NiMH cells = 168V. My car has the newer BCM, with its 180V requirement. Allowing the pack to maintain excessive voltage is an excellent way to grow oversize crystals in the electrolyte, reducing effective available surface area and possibly damaging the separator membrane in the process.

I use a console-mounted LED pack voltmeter, nice and bright and easily read at night vs. a backlit LCD. Near the battery pack I use a pair of high resistance resistors and a potentiometer to reduce the voltage to a tenth of its original value and eliminate any risk of shorting damage via excess current to the meter assembly. Across the meter is a capacitor rated at 200VDC to help smoothen the reading. It's a 20V assembly from allelectronics.com, works well. I had used an LCD panel, but it only worked right when the IMA (charging/assist) was off, no matter how much shielding and capacitor isolation I gave it. The LED unit has no such issues, allowing me to see the difference colder weather makes in giving the cells difficulty in keeping up with current levels with going into an undervolt or overvolt condition. It also helps greatly in helping me decide how much pre-commute charging to give the pack, and very clearly shows any self-discharge issues.

I made oak spacers with longer but still high-strength seat bolts added to get the very low driver's seat up a bit.

I use the Bridgestone Potenza RE-92 tires that were designed specifically for it, so sturdy that 60 psi is just fine for them. Another Insight driver ran over a significant concrete rise in the highway surface at speed without a blowout while at 60 psi, plus after 22,000 miles the tread wear is even inside vs. outer, not worn extra on the inside as one would expect. That's good enough for me, so I keep the fronts at 60 psi and the lighter-loaded rear tires at 50 psi. I'm used to CRXes and prefer cars that let me know what's going on at the contact patches, so some extra tire noise and pebble sensitivity is just fine with me.

To help the engine warm up in very cold weather I installed an IMA deactivation switch that doesn't just work in parallel with the clutch position switch, a very commonly added modification, but only does so when the throttle isn't closed via a heavy-duty lever-arm microswitch carefully and sturdily mounted at the throttle body. That way both regenerative braking and regenerative engine braking are preserved, but neither IMA assist nor background charging will occur. It's also perfect for those frequent times on fairly level highway when the car's speed would otherwise seem to need some IMA assist only because background charging had been slowing the car down...

In cold weather I use a 12V seat warmer, basically a thick seat cover with heating elements in it. I haven't checked its conductivity, but I think it developed an open circuit during storage last summer... But it does, when it works, delay the need for cabin heat long enough to help the engine warm up quicker.

I used to use a form of stove heat to help it warm up in cold weather, but now with lots of highway driving I don't need it anymore and it got in the way of fluid level checks. Below about 20 to 25 deg. F the intake air itself seemed to keep the engine from fully warming up, even without any cabin heat being extracted. So I routed the air past its primary catalytic convertor before it could reach the throttle body... problem solved.

A final modification to help with cold weather is a layer of black felt cut to exactly match the shape of the off-white hatch area carpeting, including the slits for opening the storage bin cover. Cold but sunny days aren't so cold in the car, as long as the sun can reach the black felt (not hard with the right car aiming, given the large glass area directly over it).

Pack Self-Discharge

How much the pack voltage drops while sitting at work for 8.5 hrs:

The dark line is pack voltage upon my arrival at work in the morning, the green line voltage upon departure, and the red line is 20 X total voltage drop. The first data point was after one balancing top-off charge followed another by only about four days, allowing lots of oversize nickel dendrites to form. The minimum voltage drop corresponds to the minimum voltage upon arrival. The lower the voltage upon parking, the less the dendrites are an issue. Dendrite formation is the downside of topping off the cells with a long balancing charge – the cells that overcharge while waiting for the lower-voltage cells to catch up develop dendrites. You want to stuff as much grid power into the cells as possible, but you don't want to go from 165.1V upon arrival at work all the way down to 156.0V as happened on 6/21/11 from chronic overcharging of lots of cells. If you try hard to dissolve the dendrites with letting the voltage, one can run into a negative SOC recalibration, as I have increasingly have happen.

I need to make my own MIMA to allow lower pack voltages with sustained assist; it would go very well with grid charging for mini-PHEV duty, as the negative recals are basically the car having a fit from the engine being kept from recharging the battery pack, overruling the driver in making sure the engine is used to recharge the pack. If you want to instead use grid-powered charging, not petroleum, overruling the ECM would help. The BCM calls the final shots though, preventing polarity reversal in individual cells in extreme discharge (sustained assist at 0 bars).

Other stuff:

I have the original modules (“sticks”) the Insight was built with (no longer used, replaced awhile back), and have found most modules to have one bad cell in each, a few with two, but some modules are all good. After setting them up in 2S10P configuration with my 12.2 – 14.2VDC thin film solar-charged piggyback home power system that most, depending on weather, powers my (LED) lighting, netbooks and occasional laptop use to get to see at least some cycling, they're back in the Civic pack case my Insight's previous cells were originally in recharging from the MkI 350mA single-stage charger I used until recently. (I'm able to have plenty of spare RS-25-48 power supplies from having nabbed a bunch off an eBay liquidation sale.) I'm plotting the voltage as they charge and then self-discharge to see how worthwhile it would be to use them as a secondary buffer pack, only adding recharging to the main pack via an industrial diode and the full pre-charge resistor, contactor etc. that's part of the Civic pack.

How the pack's doing so far, starting from a static voltage of 125V:

Wps = watts consumed by the power supply (Watts, power supply). The known-bad cells are surely mostly responsible for it taking so very, very long to exceed 1.4V per cell, and for the quick voltage drop-off upon stopping charging. Self-discharge is really bad, and could power a collection of CFLs at about 1.0A +/- 0.2A (they don't mind DC, and seem to self-regulate to maintain a fixed power) for only a few hours no matter how much I charged the pack ahead of time... No good.

I've printed out for bedside reading the three parts of the PICAXE hobby microprocessor programming / use manual, with one use of them being to relieve me of the need to monitor voltage readings over time (such as when I'm sleeping for instance). The ancient TRS-80 Color Computer was my main introduction to computing, and the PICAXE 18M2 can do most of the stuff the “Trash-80 CoCo” could do but using only three or four AA cells and a few ICs on a small board.