EFI updates

[chris900f]

New and improved mount for the LP pump and test fit for the air-temp sensor mount.





A longer bolt passing through the battery box mount gives me a place to mount the relay. Since the relay is 1 amp (1 amp~12 watts), I nixed the idea of another separate relay for the o2 sensor.

[chris900f]

I took the pipe for welding. The guy is a talented fabricator, and his home shop is filled with pro machining equipment and some awesome bike projects. I underestimated how difficult the location would be: it's very tight and hard to maneuver the torch between the headers. The Suzuki welds look comically bad right next to Domingo's. *see last pic)

I should have marked the outside of the headers as well as the bung location for angle, as after the hole is drilled the area to be welded needed to be sanded. Without the bike there it was my call and I overestimated how much forward tilt was required to clear the frame crossbar. I decided to rotate forward about 2mm; which is what we could get away with keeping the holes lined up. It was unnecessary and actually reduced the clearance between the top of the sensor and the fender more than I would like.

I could get another sensor (eg. short-bodied NTK/Ford from Mustang/Crown vic), but instead added a little more preload to the forks to use the taller Denso without worry that I might clip the fender under hard braking. It's simple enough to revise this in the future, but for now I want to press on...

[Bluesmudge]

Really cool stuff! You are making quick progress.

[chris900f]

Thanks Blue, it's been a bit of a grind lately.

I rode the Honda to work today, and had to get a jump start to get home. Seems like I neglected my battery maintenance; it was really low on electrolyte so I hope topping it off will cure it. Anyhow I had to remove the seat to get the battery out so thought I would post a pic of the Megasquirt...good luck fitting that into the GS.


Ok back to business: here's the O2 sensor installed




Routing the sensor wire over the engine put the connector farther toward the back of the bike than expected.

I couldn't get hold of a matching connector, but the female terminals harvested from the GSXR harness fit nicely; I can install them manually since there is less of a chance of the elements reaching it in this location. I used a block of high density foam to secure the sensor cable between the throttle cables for now.




Here's where I started today. The cross member is connected to directly to the negative battery terminal. This allows me use it as a common ground block for both sensors and fuel pumps, this way I can minimize the number of wires running  through the narrow spaces available on either side of the bike to get all the way to the tail where the ECU will be mounted. The O2 sensor has four wires 1 heater power/1 heater ground, the output signal and its ground. so both O2 grounds can terminate at the grounded cross-member.

The bundle attached to the cross-member plate is the common positive terminal. It is fed 12v+ by one wire from the relay. The bundle supplies 12v+ to both pumps and the injector harness. The o2 sensor gets its own 12v+ directly from the relay, as the socket came wired with two outputs, and a separate positive will allow me to install a toggle switch in the future if needed. 



The air temp sensor mounted: one wire of the sensor can be grounded right at the mount.




The relay is mounted under the battery box. The left side through-bolt was replaced with a slightly longer one to create the mounting point. The relay socket has multiple wires: the pos/neg ring-connectors for the battery terminals, two 12v outputs, the relay trigger wire, and an inline fuse in its own sealed box.

In order to contain this potential mess I cut up some tubular handle covers I had in the garage. An old motorcycle grip would have worked just as well, but this is what I had, so...


I cut the ends off of two of the covers. The first fit tightly around the relay socket, the second holds the majority of the cabling and the third holds the little fuse box--I cut a little window in the end of the third tube-cover so the fuse can be accessed. The 3-piece design allowed me to use the gaps to exit wires at appropriate locations. After getting everything arranged, I used waterproof rubber stretch wrap to combine the three pieces into a single cylinder. Rubber stretch tape is the "bomb" it's waterproof, heat resistant and sticks only to itself without any adhesive.

I also used an inverted flange nut on the right hand battery-box mounting bolt as a support to keep the whole assembly level. The right side is simply zip-tied to the bottom of the battery box.

[chris900f]

Better view if the relay tube thing:



The plumbing to include the low-pressure pump. I'm not that fond of the pvc tubing, but its flexibility
made it the "clear choice" for this one section.



While I was working, my neighbor stopped in to see how I was progressing. He owns a few classic SOHC Honda CB750's, a V45 Sabre, some dirt bikes and a fully done-up 70's Mach 1 Mustang--all of which are serviced by other people. As he was leaving he said "one of these days we're going to find you have built a time-machine" I chuckled, but was a bit distracted and didn't really get the joke. Later I wasn't sure if it was a complement or a diss

Anyway...this was my final mod for today

[Bluesmudge]

I think your neighbor, like me, is just jealous that he doesn't have the skills to do something similar. I'm really impressed with your work. How did you acquire all the skills needed to adapt fuel injection to a carb bike? What's your day job!

[chris900f]

I think your neighbor, like me, is just jealous that he doesn't have the skills to do something similar. I'm really impressed with your work. How did you acquire all the skills needed to adapt fuel injection to a carb bike? What's your day job!

Thanks again Blue, it would be a very long post since I'm gettin' pretty old here But it has nothing to do with my day job which was as a production manager in a mid-size embroidery/clothing imprint company (until covid-lockdowns killed the business), and my real passion was always music. I've been playing bass in bands since I was 17.

I only started working in the MC industry last summer, but my old EFI Honda had a lot to do with getting the job. It's very seasonal here, I'm just back to work as a parts guy as of last week.

As one of my favorite musician's used to say "I'm formally self-taught".

(search my moniker in YouTube and you'll see me and a drummer trying to be a whole band--pretty funny and I was still dressed in my work clothes due to scheduling errors--old vids from back in the early days of YouTube)
 

[chris900f]

One thing I don't like about Speeduino is the Molex micro-connectors. To do the job properly you need to buy an expensive tool. I thought I had got around this by carefully crimping with a mini-needle-nose pliers and std crimper; but as I I began to populate the connector I realized that my results were too dodgy for an actual installation. Also I ruined about 10 of the supplied connectors and only had a few left.




After digging into my electronics box I found theses larger terminals. They were too big to populate the female harness connector, but they plugged securely into the male connector already soldered to the board; and they were much easier to handle, without a special tool.


In order to prevent contact/shorting I used some heat-shrink tubing. I was able to label each terminal, by writing on the tubing before applying the heat. This will come in handy if I need to troubleshoot and I can unplug all the connectors to locate the ECU to its final location. without a lot of extra stress.


Left side progress.





The left side is complete and contained the majority of the connections.

The right side should be much easier: just the 3-wire TPS and power (4 grounds and 1 12v+.) The switched power will come from the taillight harness, and the grounds will be local to that area as well.

3-wire TPS


I'll be glad to get the battery installed and start testing the pumps, sensors etc. with the computer.

[chris900f]

Annoying, scary frustrating!!

Yesterday I wired everything up for some basic testing of the fuel pump relay. After several hours I turn the key...nothing....damn!

I'm thinking to myself that I must have made some error. The positive signal to the relay fires the pumps when connected to the positive battery terminal, so now I'm thinking that I have a problem with the ECU as the FP pin was something I hadn't tested. I check the FP pin with the multimeter--nada. After some pacing and disaster-izing, (I maybe I broke  something, should I bypass the ECU and just run the fuel pumps off of a toggle switch? no not cool.)

Around 2am, with work in the morning, I post my problem on the Speeduino forum. "I'm not getting any signal on the FP pin" and went to bed.

I wake up early and check the forum. "You won't see a signal on the FP pin; the FP pin is grounding"
Nice, well the Speeduino manual doesn't mention that little tid-bit. Basically you need to set-up the relay in reverse with the negative input connected to 12V+ and the "positive" signal lead connected to the ground of the relay...

I was worried about changing the wiring but it actually came out a little cleaner, with fewer ring-connectors on the battery ground.


Getting closer, still a little hesitant to button everything down until she runs.

Pumps prime for 1 second(software adjustable) at power-up, then standby for cranking instructions from the ECU

TPS is wired in and responding I need to finalize the switched power to the ECU. Next I need to find some lighter-gauge MAP line for a better fit to the TB spigots.

***Just thought of something: will my O2 sensor, running off of the second output of the relay still function; or am I in for more rewiring? I'll check that tomorrow too. The heater is non-polarized . It's late...

[Ghetto Garage]

Im blown away with this project. I cant wait until you get it running and all the tuning issues worked out. You have me considering doing this as well. If this comes out as well as I think it is, I might be bugging you a lot 

Cheers,
Mike

[chris900f]

Well I'm kind of exhausted. The last few days I have been determined to get the bike back together to get my garage back. I'll keep this one brief...

Snags:
I have been pushing the start button for a few days. but the gremlins have made themselves known.

The automatic priming of the fuel pumps was going off multiple times during each test, I had the engine running, but it would scream up to 5000 rpm and abruptly die. Multiple blind alleys, and adjustments later; I found a few solutions.

Fixes:
The RPM input from the tach output lead from the stock ECU was "microphonic" any movement. vibration or contact with the cable would register as a big rpm spike. This was causing the fuel pump prime to go off multiple times when trying to start and when the engine caught and ran it would only run at a high rpm, and that only for a short uncontrolled burst--as the engine died the prime would go off 2 or 3 times as well.
I was familiar with noise in guitar cables, so I dug around in my big box of instrument cables and found a shielded 1/8" cable about 2 feet long. A shielded cable has 2 layers, a regular plastic or pvc insulated wire carries the positive; this is wrapped with fine copper stranded wire connected to the ground.
I soldered a terminal to the inner wire and connected the stranded portion directly to the battery ground. I moved the "flux" capacitor as close to the output of the stock ECU as practical and ran this essential signal as a dedicated line, outside of the split-loom to avoid interference from the other sensor and power cables. Fixed, no more "noise" no more stray signals or RPM spikes.

Next was a fueling problem. I tried to feed fuel from the low-pressure pump to the EFI pump for a few days. Either the used pump I got at work is NFG, or it simply didn't get along with the petcock. It just wasn't delivering enough fuel and the main pump would starve out and the engine would die. I was never able to make the necessary adjustments to the fuel table as the bike would run rough and stall without constant tweaking of the throttle.
Finally I pulled the pump and it's plumbing and pulled apart the sub-harness I made to power both pumps(ouch!) I was relieved to find that the gravity feed seems to be sufficient to feed the EFI pump. This saves both electrical power and complexity so I hope it works under high demand.

All during these ordeals the MAP signal was so bad bouncing up to 20-30kPa per cycle--the graph on the screen was a zip-zag and impossible to tune. I needed to sync the TB's when the bike would barely idle. I ended up trapping a zip-tie with the bottom of the throttle-plate of the master TB and adjusting the slave TB so it's throttle-plate would just hold another identical zip-tie--sort of like using feeler gauges. Not really precise, but it worked well enough to allow me to adjust the fuel table.

Still plenty of tuning work ahead, but I think the mechanical issues are mostly solved. Once I get the TB's synced more precisely I can finally put the tank and bodywork back on.

Ok, so I said brief, and wrote a small book...anyhow here's the results for today. (Note the bluetooth connection to the laptop--totally worth the few extra few bucks)

[Bluesmudge]

Holy cow...that is so cool to see a GS500 running with fuel injection. To me it sounds like a better idle than even the most freshly synchronized carbs. Like the engine had just been waiting for this.

[Bluesmudge]

You probably won't ever get the bike onto a Dyno, but once you dial everything, I'm going to be forever curious if fuel injection does more for performance than all the carb mods/intake and exhaust mods that have been discussed on this forum for the last 20 years.

[chris900f]

Holy cow...that is so cool to see a GS500 running with fuel injection. To me it sounds like a better idle than even the most freshly synchronized carbs. Like the engine had just been waiting for this.

Well I tried to sync the TB's today, and I think I went backwards a little. If you can keep the "bounce" to about 10kpa you're golden.

The Idle you heard is about as good as it gets for a twin. I think I'm a little biased having only tuned a 4-cylinder before where you always have an induction on one cylinder. Still it's a pretty good start; and now that there is a blog to refer to, anyone else attempting this will be able take a much more direct route and skip the drama. I would suggest anyone thinking of this find a set of 2001-03 GSXR600 TB's as they are getting pretty rare on eBay.

The lads at the shop are pretty curious, so I may be able to get on the dyno...but I need to get it running well enough to ride it down there in traffic. Right now I need to get it together just so I can go get gas. I went through at least 1/2 gallon of regular just trying to improve the MAP at idle...60ml at a time:)

[chris900f]

Time for a Sunday update:

If you watched my video, you can hear me gripping a little about how I didn't believe TunerStudio's  10:1
AFR at idle. Well, unlike the previous MegaTune software that I was using, a lack of voltage signal displays as a high rather than low AFR. The sensor was not getting any power due to the necessary rewiring I did to the Fuel pump relay.

Luckily Suzuki provides a convenient switched power source in a handy location: the positive terminal of the stock idle solenoid on the carbs. I rewired the O2 sensor and immediately I was able get proper readings.


Another issue was the power consumption of the O2 sensor. As was anticipated, the O2 sensor is the biggest draw in the EFI system. Switching on the sensor (it feeds from the RUN switch power while the computer feeds from the ignition switch power) the battery power display drops by nearly 2 volts. As the sensor heats up it will draw less power; but obviously simply hardwiring the the sensor into the system is problematic.

To get around this I wired in a power switch, so the engine can be started and warmed without the sensor pulling power and only being switched on when tuning the idle or when datalogging a ride.

I had already planned on making this cover to hide the fuel pump. The cover is actually an ankle guard from an unknown sportbike. It is held to the frame by 3 powerful magnets. I enlarged one of the existing holes and mounted this nickel-plated switch--I think it looks pretty good.




Finally, here is a look at the datalog from my first ride. The ride was a little sketchy, as the un-tuned engine is running from a VE Table that is a guess at best. Traffic and taking off from stoplights were no fun...

[Armandorf]

I imagine you can salvage the minimolex connectors from computer PSUs, i have a few laying around. Im pretty convinced that if you search on repair shops/online you can even ger them for free from dead Chinese bad quality PSUs.

How much is a Dyno run these days?

Really nice project, keep it up!

[chris900f]

I imagine you can salvage the minimolex connectors from computer PSUs, i have a few laying around. Im pretty convinced that if you search on repair shops/online you can even ger them for free from dead Chinese bad quality PSUs.

How much is a Dyno run these days?

Really nice project, keep it up!

Thanks Armandorf

Re: minimolex I did a work around by using slightly larger terminals and plugging them directly to the board connector. My advice for builders would be to solder your leads directly to the board and run them out to an automotive 24 pin connector and build-out your harness from there.

I need to allow the engine to breathe a little more easily to prevent over-carburation. Currently I'm refurbing a V&H SS2R oval exhaust-can almost exactly like what I have on my Honda. I got a pair of them locally for cheap ($60.00) on Kijiji--but they need some work. I've got the bike to where it cold starts and idles well (enough).

Next I need to get the drivability good enough to get about 40km across the city to my fabricator who will cut/weld the pipe and mount the V&H can. Once I have the exhaust set-up in its final form I'll road-tune a little more--dyno time is $115.00/hr so I want to be really close before shell out for the dyno.

[chris900f]

Well this will be the last EFI update...at least for this season

My progess tuning has been slow lately. The fluctuation in the MAP signal at idle and below about 4000rpm was not an insurmountable hurdle, but it made road testing both difficult and sometimes extremely dangerous.

Years ago when I converted the 4-cylinder Honda CB900F the idle came together very quickly, and I was able to achieve better than stock off-idle and low-end throttle response in just a few days. After those essential steps, road testing was not particularly difficult; datalogging on low-speed-limit side roads and gradually working up to highway speeds and beyond. Even though the tune was never perfect, it was always good enough to maintain a margin of safety.

This twin does not pull enough vaccum at low speeds. The resulting "spiky" MAP signal translates directly
to the injector pulse-width... I was getting a pretty good handle on it in my last few test rides. The idle and sync seemed good, but I couldn't find any take-off power without adding a ton of fuel in the top-most sections of the fuel table in this low-rpm range. eg 85kpa at 1800rpm so the engine would hesitate and then catch with a lurch forward...roll off the throttle too soon when this happens and you tend to stall. This was happening at stoplights and intersections. I finally just got used to winding the engine up to around 3000rpm and letting the clutch out fairly slowly.

Once above about 3000-3500 rpm the bike pulled really well, even with a fuel table that was still in the guesswork phase. I was encouraged by this and did more datalogging and revisions and felt I was really getting somewhere.

Problems and more Problems.

So I was stoically forging ahead when two more ugly issues came up. One, once the engine was hot and the engine manifold boots got soft and pliable even a slight misfire back though the intake would unseat the TBI's from the manifolds, causing air-leaks at the intake. The EFI registers the air leak as higher MAP and compensates with more fuel--flooding the engine.


Two, and this one is the real killer. Once the bike had been running for about 10 minutes the fuel pump began to puke fuel from its overflow. This happened mainly at low and idle speeds...so I'm riding along and everything is working so-so/ Then I come to an intersection and by the time the light changes I have a pizza sized fuel spot under the bike. This problem got progressively worse. The pump is a cheaper copy of the Kohler modular fuel pump...I had the choice of buying another cheap chinese copy and according to online reviews, encountering the same problem again, or spending around $400 for a genuine Kohler pump.

I just don't have enough time or money to continue with this project, I have other things to do and other more essential vehicles that need money and attention so after much painful deliberation I decided to pull the plug.


Silver Linings:

One of my design goals when I started this project was to be able to put everything back to stock
incase of failure and to preserve resale value. As such no hard mods were made to the bikes original wiring or fuel tank etc.

So I unplugged the TBIs and the pump from the harness. But I saw no reason to pull the computer and all the wiring out just yet, and frankly the thought of undoing all of that work made me sick to my stomach. As I retrieved the carbs and airbox from storage, it struck me that I had a unique opportunity--one that i never would have pursued had it not been for the prospect of fuel injection.

The TBI's only had two plug connections, one for the injectors and the other for the TPS. The MAP hose simply disconnects from the TB spigots. I had just purchased a new O2 sensor, since the used one I had been using was consuming and inordinate amount of power--nearly 3amps--enough to pull a fully charged battery from 14 to 12 volts. A quick test confirmed that the new unit (downstream sensor for a 2003 Honda civic) was only pulling a fraction of the juice..and as a bonus, the male connector from the used Denso, matched perfectly with the female connector from the new NTK. So I harvested the cable and connector from the denso and rewired the new sensor to get switched power from the now unused fuel pump relay.

While I had the TBI's off I had a look at the fit of the manifolds to the head. The factory install was very sloppy. I deduce that the assembly method is to install the manifolds loosely to the head, then plug the carbs in to quickly set spacing, then tighten the manifold screws in whatever position allowed easy access to the screws. Both manifolds were slightly offset, so that airflow from the carbs ran straight into the metal crescent visible between the manifold and the head. Want to get more power, an better idle and generally better performance--next time you remove your carbs check and re-center your manifolds to the ports to eliminate this nasty "step" that robs you of 3-4mm of your carb diameter and obstruces flow.

After reinstalling the carbs and the usual fight with the airbox, I installed a T between carb#2's MAP/Vacuum spigot and the stock tangle of hoses and connected back to the computers MAP sensor-I even reconnected the stock air-solenoid for the high-idle at start up. Finally I mounted the air-temperature sensor to the one of the screws of the air filter intake, reconnected the sensor and tided up all the wiring.

Of course the bike fired up and runs like its old self--maybe even a little better with the re-centered manifolds.

So here's the silver lining: I now have a stone stock, low-mile (3111km's) bike fully wired for MAP,RPM, O2/AFR, Cylinder head temperture and intake air temp.

I've just datalogged a long ride with the new set-up with some surprising results. I think I will start a new thread for this phase as it applies to all stock bikes, and may provide some tuning insights.

[Armandorf]

you can still datalog carb operation?
should be good enough info to understand how the carb reacts and maybe emulate that behaviour for a v2 EFI.
it also provides an excellent way to  fine tune the carb.
i understand you have: MAP,TPS, o2 sensor,RPM

Really hoping you can make it.
using the stock airbox with the EFI system is an option?with a single filter? this maybe can help you but im only supposing..

my thoughts in the air, maybe its nonsense:
i imagine is the diaphragm/throttle slide operation of the carb is different and difficult to emulate, a way to overcomne this could be being able to control the throttle valve, maybe a servo and throttle by wire, that way you can  somewhat control vaccum from the speeduino
what about engine load sensing found in cars?

what are the differences with the honda that could be in the way with the GS?

[chris900f]

Hello Armandorf

Yes I can use the onboard Speeduino and sensors to datalog with the carbs and airbox. The first thing I discovered was that the bike runs richer than expected.

Check out this histogram (compiled/averaged) view of a 40km ride. The darker green areas have more "hits" so they show where the engine was operating most of the time, yellow areas have fewer hits and the white background cells show are where the engine was operating for a very short time eg. at the highest RPM. More hits equal better accuracy

This is the view with with data below 180 degrees F filtered out. During the cooler on-choke operation the AFR is 10.6:1--so extremely rich-(makes me wonder about the idea of needing a richer pilot jet)



It would have be helpful to have this data before moving to throttle bodies. This table can be exported and then imported to the AFR target tables in both the MegaLog viewer tuning software and TunerStudio operating system. This would have made it easier to zero-in on the correct AFR's

You can see that the whole table shows rich AFRs except in the cruising zones where it drops to the high 13"s and low 14's...almost stoicometric.

So two observations: The common wisdom "these bikes are tuned lean from the factory" doesn't really appear to be the case, as under heavy load (high-MAP) the AFR's are quite rich and the idle sits in the low 12's even with the engine fully warmed up. If anything the little GS come from the factory tuned like a high-performance race bike--at least in terms of AFR.

Airbox: My first datalogs with the pods were done on cooler days around 65F and the intake air (behind the engine) reached as high as 115F. This latest datalog was done on a hot day, around 80+F and the airbox intake air temp only got to 91F. So I wonder if pods or lunchbox are really providing much of a performance improvement over the K&N drop-in filter. Unfortunately, the EFI set-up requires the space taken up by the airbox to mount the fuel pump; and the TBI's are too short to reach the airbox boots--so there's a few problems.

You are correct that having a second set of servo operated butterfly valves to emulate the diaphragm action of the CV carbs would help stabilize the MAP signal. Also the MAP signal IS the load sensing on MAP based EFI.

Finally, the Honda is a 4 cylinder bike. The 4 tbi's each have a MAP spigot and all 4 are ganged together. Since a 4cyl always has an intake stroke occuring, the MAP signal is much "flatter" and more even/consistant, with only a few KPA of bounce at idle.


Check out the AFR when pulling hard during acceleration--wow. (white=MAP, red=RPM, yellow=AFR)


Typical idle AFR


Super-rich on the Choke (10.6:1)