With the fusebox mostly wired up, it was time to deal with the other wires. While I’m trying to “repurpose” wires in the stock harness when possible, that mostly helps when connecting up existing things like the lights. For the new sensors, I needed to pull new wires.

First step was to add an intake air temperature sensor. Fuel injection systems need this to calculate air density and hence mass airflow into the engine. I used a small NTC thermistor I got from Mouser and drilled a small hole in the airbox for the wires.

For now, the intake air temperature sensor was “mounted” by squeezing it between the velocity stacks. It weighs practically nothing, so it may even stay there. The wires exit through a small hole that was “sealed” with heat shrink tubing on both sides.

To close up the hole for the wires, I added two layers of heatshrink tubing on each side. It’s certainly not air tight, but there won’t be a lot of air going through. Since this all needs to change when the carburetors are removed anyway, I’m not too worried about making this permanent now.

The new wires for the harness. The tiny connector in front of the thermostat is for the intake air temp sensor, Behind it is the new 2-wire coolant temp sensor. Three wires continue up to the temp gauge. The thicker, brown wire is for the radiator fan.

Rather than worry about making a new harness now, I decided to just pull the new wires along the old harness. I can foresee that things are going to have to be tweaked, so until I know that everything works, I want the new wires accessible. Eventually, I’ll take the entire harness out, integrate the new wires, and remove obsolete ones. For the sensor wires, I’m using 20-gauge TXL wire. For larger sizes, GXL. This is automotive-quality wire rated for 125C.

I looked hard for suitably small, sealed, connectors when I was planning the wiring. The tiny, 2-terminal connector for the intake air temp sensor in the picture above is a Molex “Mizu-P25”, which seems quite promising. It’s big enough that I could crimp the terminals with my crimp tool, they’re rated for 4A with a 20-gauge wire like these, but they’re still very slim. (For supply-current sized connectors, I’m using Metri-Pack 280 which is rated for 30A but, as you can see in the previous post, those connectors are very bulky.)

The coolant temp sensor was replaced with a 2-wire model with a dedicated ground wire. (You don’t want to use the frame ground for sensor wires. To avoid interference from ground currents, the Microsquirt has a dedicated sensor ground pin, and all analog sensors need to go back to this pin.) The old temp sensor wire connects to the temp gauge, so that’s now obsolete.

Since the temp sensor now goes to the Microsquirt, I need a way of driving the temperature gauge. I’d already verified that I can repurpose the “fast idle” PWM output for this. The fast idle function normally connects to a air valve that admits extra air when the engine is cold, raising the idle. (Kind of what the choke does on a carburetor.) I’m not going to use this function, but I realized a while ago that since this is essentially just a PWM output driven by the coolant temp, I could hook the temp gauge up to it and tweak the lookup table until the gauge shows the correct temp. (The gauge reacts slowly enough that the only symptom of it being driven by PWM is a slight hum if you put your ear up to it.)

Anyway. To hook the gauge up, I needed to replace its power supply and ground wires so it connects into the Microsquirt’s power/ground. This is both to avoid the gauge being affected by ground currents, but also because the Microsquirt outputs all sink current and if they are powered when the Microsquirt is not, power can back-feed into it. All in all, I needed to run 3 wires all the way up front. (Actually, I’m going to need to run another, for the tachometer, too.)

Finally, because the Microsquirt will control the radiator fan, I needed to run a new supply wire from the relay I added down to the fan.

All the connections to the wideband controller/comm box are done. Looks quite neat with the Ampseal connector and the wire braid.

Moving to the back, it was now time to start connecting all the electronics. It’s quite difficult to know how long the wires need to be, but in most cases you can just cut them “long enough” and trim later. After a full afternoon’s work, the connector to my custom box is all wired up, and the result looks quite nice, if I may say so.

The Microsquirt wiring is still a complete rat’s nest.

Mostly, it’s now a matter of crimping and mounting all the terminals for the big 35-pin Ampseal connector for the Microsquirt. It may end up being quite tight in there, given that you probably want some slack on the wires. The final task is to connect the two oxygen sensor wires (the two thick, gray wires in the picture above) to the controller boards inside the box, and then I think I may be able to start it (using the stock ignition system.)

With the oxygen sensors mounted and the wideband controller box finished, I could finally start doing the wiring. While this seems like a simple task, I had a hard time deciding how to do it.

The big question in my mind was how to wire the power buss part of the fusebox. This requires connecting the incoming power to many fuses, so the wire somehow has to be split. The Metri-pack terminals in the fusebox will only admit one wire (and even if you managed to jam two in there, you’d ruin the sealing that makes the box waterproof) so some form of distribution circuit had to be created. The obvious solution is to crimp a bunch of wires together, but it’s kind of fiddly to create and it creates very thick and inflexible “bundles” of wires that are a pain to handle. However, any other solution would require some form of terminal block, which is exactly what I was hoping to avoid with the nifty, waterproof fusebox.

In the end, I settled on crimping the wires together with the “open-barrel” splices I got from Cycle Terminal back when I replaced the burned-out alternator wiring. After a few failed attempts, I finally got a procedure that worked and spent some time making the necessary distribution wires.

These are the main power distribution wires finally crimped together, heatshrinked, and plugged into the fusebox. The green “knobs” poking out of some of the holes are plugs that seal up unused holes.

It turned out OK, but there’s always the question of how well the adhesive in the heatshrink flows into the space between wires when you shrink many together. At least these are easy to pop out and replace if they should turn out to have a problem in the future.

With the distribution wiring completed, next task was to get the battery and alternator wiring in. Back when I redid the alternator wiring, I added a fuse for the alternator in that wire. Since there will now be an alternator fuse in the fuse box, this was removed as was the stock fuse box and the main relay I added before. To be able to disconnect the main power, I added a 2-position Metri-Pack 280 connector for the incoming battery and alternator  connections. The stock wiring for headlights, tail lights, indicators, etc, will be retained. They were connected to the fusebox via a 3-position Metri-Pack connector. At this point, the whole deal looks like this:

The stock fuse box has now been removed and those circuits connected to the appropriate fuses using the two Metri-Pack connectors on the left. The small 2-position connector connects the main relay to the ignition switch.

The Metri-Pack connectors are kind of large but they tuck up pretty well under the metal bar so there shouldn’t be any interference with the seat. The inside of the fusebox looks like this:

The main relay and the fuses for the alternator and the lights that were moved from the stock fuse box are now in place.

That was pretty good progress for a day’s work. There will be three more relays in the fusebox, one controlling the radiator fan, one controlled by the Microsquirt to enable power to the fuel/ignition system, and one that cuts said power when the kill switch is off. There will also be a bunch more fuses for the fuel/ignition components and the fan.

I haven’t touched any of the electronics wiring yet, but there are also a bunch of wires that need to connect the Microsquirt to the other electronics box and to some other sensors. Plenty more to do.

In the last post, I described the electronics used to talk to the wideband oxygen sensor controllers. What remained to do was to physically mount the sensors themselves in the exhaust pipes.

The sensors themselves are quite large, so I struggled a bit with finding a good place to mount them on the famously cramped NC30. The exhaust system is a 4-2-1 type, and the final merge collector is so far back that with one sensor it would be out on the side, basically just in front of the muffler. This location is fully visible, hard to route the wire to, and perfectly positioned for getting whacked by the rider’s left foot. Not too good.

As mentioned in the previous post, I wanted to run 2 sensors anyway. The factory carb jetting differs between the front and rear cylinders, so it seems prudent to measure their air/fuel ratios independently since they apparently require different tuning. This also had the advantage that I could locate the sensors where the exhaust pipes wrap under the bike, which is less visible and (hopefully) less susceptible to getting hit by stuff, since it’s somewhat protected by the lower fairing.

The only tricky thing is that the pipes are very nearly side by side, so there’s not much room. The sensors also must be mounted with the sensor element pointing downward by at least 10 degrees. This prevents condensation from collecting in the sensor. There was no way to fit the sensors if they were mounted 90 degrees to the pipes, but luckily there are also angled bungs. Angling them mean they take less space and you can also fine-tune the position by rotating the bungs. After some fiddling, I arrived at locations that seemed to work.

Once the locations had been determined, the next step obviously was to make holes in the pipes. This turned out to be more of a chore than expected. I knew that stainless steel work hardens, so unless your drill bits are very sharp and keep cutting, it will quickly become impossible to get through. I had new drills, but it still took me forever to make a 2mm hole, and when I stepped up to a larger drill, the edge quickly cracked. Maybe the pipes already are work hardened from manufacture?

In the end, I had to resort to the Dremel’s cutoff wheel and carbide end mill. (This end mill is amazing, it cuts anything like butter, but in the process it also produces clouds of tiny metal needles that love to embed themselves in your skin. I constantly have to remind myself to only use it on a clean work table that can be meticulously vacuumed afterwards.)

The hole is mostly done. Getting through the stainless was a chore.

To get the bungs to fit on the curved pipe, they also had to be ground into a concave shape. This was done on the bench grinder. I also had to add some material with the welder to avoid grinding into the threaded part. Once that was done, they were tack welded onto the pipe.

The first bung is tacked in place. I also had to weld on some material where it was initially too short.

With the bungs tacked on, I put the exhaust pipe back on to check the fit. Unfortunately, the sensor to the front cylinder pair was uncomfortably close to the bracket holding the suspension linkage, so it had to be moved a bit.

Time to check the fitment. The top sensor ended up a bit too close to the suspension bracket for comfort.

After grinding off the tack welds and moving it slightly further down, it fit much better.

After moving the bung down a bit, both sensors fit without interference. It might look like the bottom sensor doesn’t have the requisite 10-degree down angle, but I measured it with the digital level and it’s fine. (The bike also now sits on the center stand so taking it off will lower the rear end a bit more and improve the angle further.)

Then it was time for the welding. I was quite nervous about this, since I’ve never welded stainless before and an NC30 exhaust system is pretty much “unobtainium” in Hawaii. When welding stainless pipes, you also have to use a purge gas on the inside of the pipe, otherwise the weld will oxidize on the back side and lead to premature cracking. To do this, you cover the ends of the pipe and route another Argon bottle into it.

The exhaust system is ready for welding. The aluminum foil is to keep the purge gas contained. The Argon hose was mounted at the muffler end and that side taped off. A couple small holes in the aluminum foil give the existing air a place to exit.

I’m happy to report the welding worked fine. It was a bit awkward to access the side of the lower bung that faces the other pipe, but the weld wetted out easily and I didn’t burn any holes in the pipe! I think it looks pretty good for a first attempt!

The welds are done. Not too bad for the first time welding stainless!

I’ve been anxious about this welding step since I first started thinking about the project, so I’m happy that’s done. Now I can proceed with routing the wires to the sensors and then it’s finally time to do all the wiring between the fuse box, the Microsquirt, and the electronics box.

As described in the first post in this series, I’m working on converting the NC30 to fuel injection. That post was 2 months ago, since then I’ve mostly been working on a box of electronics that aren’t exactly required but that I think will make things a lot better.

When designing the whole Microsquirt system, there are a couple of loose ends that need to get tied up:

• Pressure sensors for barometric and manifold pressure need wiring up, with dedicated analog lines. You can use automotive sensors for this, but they’re a lot more expensive than the ones you can get from Mouser.
• For tuning, you need to connect the ECU to a computer. Normally this is done over RS232, but I wasn’t psyched about having a serial connector somewhere on the bike exposed to the weather. Instead, you can use a Bluetooth serial bridge and connect to the bluetooth on the laptop. However, you need to put the Bluetooth modem in some protected location.
• Finally, you need to hook up wideband oxygen sensors. Most consumer-grade ones output an analog voltage, which means you have to be very careful about signal interference. However, there’s a DIY one you can buy as a kit from 14point7 that can also be read digitally over I2C. You can then send this data to the Microsquirt over the CAN-bus connection, eliminating any analog part from this very important signal. This seemed like an obvious use for an Arduino…

At first, I meant to accomplish this in an as simple as possible way, using a breadboard and breakout boards and DIP chips. However, it quickly became clear that this would not be much easier and would be really clunky. Instead, I decided to design a small breakout board with surface-mount Atmega microcontroller and the CAN-bus controller/transceiver. I’ve never used any of the on-demand circuit board businesses, but it turns out that small boards are really cheap. I used OshPark, and the small double-sided board ended up costing $6 for 3, including shipping! And they’re purple!

The front side of thee OshPark board after soldering on the SMT components. The large chip is the CAN-bus controller.

The back side of the relay board. The large chip is the Atmega 328 and the small one the CAN-bus transceiver.

I had already tested this with an Arduino and DIP versions of the CAN-bus components, so I knew I could send data to the Microsquirt and have it show up as fuel/air ratios.

With the complicated circuiry out of the way, I just needed a simple board to mount the pressure sensors, the Bluetooth modem, and connect to the wideband controllers. This was sufficiently simple that I decided to make a single-sided board myself using toner transfer.

Here’s the single-sided carrier board after etching. Not perfect, by any means, but it’ll work. The super-wide traces in the lower right is for the power supply to the wideband oxygen sensor controllers. Since the sensor has a heater, they actually use up to 5A per sensor.

I haven’t done any etching since the wine refrigerator project back at the beginning of this blog, so it took a while to remember all the subtleties. And while this worked fine, when you take into account time spent doing transfers, etching, drilling, etc, it’s not at all clear that it’s cost effective compared to ordering from OshPark. This board has a lot of area, so it would have been about $30, but it also took me quite a few hours to do by hand.

This is the fully assembled carrier board. The pressure sensors are at the top (the MAP sensor is the one with a barb, the barometric pressure one obviously doesn’t need one.) The red boards are breakout boards from Sparkfun with the Bluetooth modem and RS232 level converter. The two MicroFit connectors in the lower left are the power and I2C connections to the wideband sensors. The large connector is the external Ampseal connector.

The plan was to stack this board and the two wideband controller boards and mount them in a weatherproof enclosure. The stacking worked out quite well and everything fit perfectly into the Hammond enclosure.

The two Sigma Lambda OEM wideband oxygen sensor controller boards from 14point7.com along with my board, ready for stacking.

This is the assembled board stack, ready to go in the box.

Getting the boards into the enclosure turned out to be a bit more work than expected. The problem is that the enclosure is quite deep, and the Ampseal connector quite large.

The Ampseal connector has a seal on the inside that will seal against the inside wall of the enclouser, so the hole in can’t be too large. This means it’s tricky to get the board in.

After careful filing I managed to make a hole large enough that the connector can be jammed through, with only a little flexing of the circuit board, while keeping it smaller than the seal on the flange. As you can see in the picture above, I failed a bit on locating the screw holes, but they are all on the “external” side of the seal so it’s only a cosmetic matter.

In the end, the enclosure is quite large, so there’s really only one place on the bike where it fits: in the tail along with the other electronics. There’s not really any extra space, but it works.

This shows how the enclosure fits perfectly between the two metal bars in the tail. The blue silicone hose is the MAP sensor connection. The connector fits outward because it would be really tight against the back of the fuse box on the other side if it fit the other way. The tail fairing is plenty wide there to accommodate the connector.

So that’s where it’s at right now. I’m currently in the process of welding bungs for the oxygen sensors onto the exhaust pipes, once that’s done and I know how long those wires need to be, I can mount this box permanently and start the job of wiring everything up.

While working on the fairing, I’ve also been planning a much more invasive project: Converting the NC30 to fuel injection!

This one may deserve some explanation. I mean, what’s wrong with it as it is? Nothing, really. This is strictly done in the interest of education… What semi-rational explanation I have is that one of the things I’m interested in doing with the engine for the airplane (long-term project as that may be), is to use an aftermarket turbo/fuel injection conversion. Since an airplane is a fairly dubious starting project for learning how to do do engine work, seeing as you can’t pull over when the engine decides to stop, I figured it might not be a bad idea to try a somewhat lower-risk project.

So here’s the plan: The NC30 will have its carburetors and the old ignition control unit ripped out, and they will be replaced with a Microsquirt engine control unit, and throttle bodies with fuel injectors. The list of changes to pull this off is pretty long, so I’m going to do this in two stages:

Stage one is to use the Microsquirt to control ignition, which is comparatively easy as there are no major hardware changes necessary. This is mostly electrical work, as the engine controller needs to be wired in along with the necessary sensors and wiring.

Stage two is then to switch out the carburetors, which is quite a bit more complicated. This entails adding throttle bodies, fuel injectors, fuel pump, pressure regulator, and return line to the tank. The throttle bodies need to be made to fit the intakes as well as the airbox. I’ll discuss this part more later.

The electrical hardware is going under the tail, where luckily there’s quite a bit of space that can’t be used for anything else anyway. An electronic fuel injection system has a quite a bit more complicated electrical system than a stock early-nineties motorcycle, so the old 4-fuse box also isn’t quite adequate.

The new fuse box is a sealed Metri-Pack 280-compatible matrix of connections that accommodates both fuses and relays.

The new electrical system will use a number of fuses and relays, and they will go in this super-flexible, sealed, connection box that uses Metri-Pack 280 connector spacing. This is the spacing used by ATM fuses, and there are things like relays and bus bars that use the same spacing. This means you can mix and match and design your own layout, and it’s all waterproof. I mounted this on some aluminum brackets I added to the unused passenger seat rails.

The Microsquirt case is mounted right in front of the tail light. Here, I cut away the existing plastic “wall” in the undertail and added threaded nylon mountings that I “Plastifixed” to the plastic. The Microsquirt has a watertight Ampseal connector that’s facing forward, which should make it fairly easy to add the necessary wires between the fuse/relay box and the controller.

Here’s the Microsquirt mounted under the tail. The “hole” right below it is where the old ignition control unit was located.

Still needing to be added are the controllers for the wideband oxygen sensors (one each for the front and rear cylinder pairs), a home-made board that will contain the Bluetooth modem that will make it possible to connect the laptop to the engine controller without hooking up a wire, an Arduino that will send the oxygen sensor data to the engine controller, and the pressure sensors for barometric and manifold pressure. There’s also an ignition module that switches the high current to the ignition coils, but it needs to be mounted to a heat sink, so I think it should go closer to the ignition coils.

There’s quite a bit of stock wiring that needs to be replaced, but rather than do a hack job, I’m going to do a proper job of disassembling and modifying the stock wiring loom. It’s going to take a while to see this through, I think.