I knew that the purpose of a vacuum line on the fuel pressure regulator was to help vary fuel pressure according to manifold vacuum. I just didn’t realize that it’s actually pretty important to do that.

Quoted directly from the MegaManual:

The vacuum referenced fuel pressure regulator is essential. It provides constant pressure differential between fuel at injector nozzle and manifold air pressure [port EFI] or atmospheric pressure [TBI]. This makes the injected fuel quantity solely a function of the injector open time.

If you were to ‘cap off’ the manifold vacuum port on the fuel pressure regulator, you are reducing the dynamic range of the injectors. This means you will need lower pulse widths at [idle] (giving less control over idle mixtures) and lower flow under boost (restricting the maximum horsepower).

So, in general, for port injectors, have the fuel pressure regulator connected to the manifold vacuum is a good thing. There is very little reason not to do it (though some have argued against it for individual runner port EFI set-ups).

Ok, ok, hint taken, especially paragraph two. It will be very easy to hook up, assuming there is no pressing reason not to allow the vacuum from just one cylinder to operate the regulator. hmmmm Logistically, it would work really well to T off the MAP signal; perhaps the regulator would serve as an accidental vacuum accumulator and smooth out MAP repsonse… In any case, I hope to have some wrench time on Buzz this weekend. I also very much hope to have some *riding* time on Buzz this weekend.

Speaking of fuel, I forgot and left the fuel valves on again. I’ve been out in the driveway at least briefly most nights this week except last night and saw nothing out of place. Low and behold, tonight I go out and find a big puddle of gasoline under Buzz. Of course, it’s dark and I had no time to troubleshoot, so I just turned the valves off. Next time I go out there, there will be no leak and no evidence of a leak, thus nothing to fix.

In other light reading, I am trying to find specs on this twin TB. As sold, it was described as being for a Kawasaki KFX450R ATV, but everything I can find on the KFX450 indicates that it is a single. After looking around in the parts diagrams on Kawasaki’s website, I’m pretty sure these are from either a Ninja 650R or a Versys, both powered by a fuel injected 650cc parallel twin. I haven’t been able to match the part number on the box (16163-0078) but the illustration of 16163-0084 is identical. Furthermore, looking up parts for the KFX450R (as the box was marked in Sharpie and as the eBay auction described it) reveals a single throttle body, not a twin.

I suppose whatever the throttle body was intended to fit, it changes my conditions very little. I still need to find out what the injector ratings are. Googling for the part numbers on the injectors (ETA287 on one side, 2870581 on the other) doesn’t find them. I also will still need to build a manifold for it.

On that subject, the intakes on the heads are so nearly exactly 1″ ID that it will have to do. That end of the TB is 2.2″ and Buzz’s current TB assembly fits in 5.25 inches between the cylinder head and the base of the filters, , not including the filters. Since two TBs will fit between the frame pieces that currently limit the space available to four TBs, I may be able to stretch that distance a bit. Also, these TBs may be a a quarter inch or so shorter than the ones currently installed, but it’s hard to accurately measure with them installed.

But basically, I will be building two kinda short blocky Y (really more like a J with a crossbar; I need to draw it out and scan it), with a 2.2″ pipe on one end and two 1″ pipes on the other. The engine ends will have flanges and the TB ends will hopefully have an O-ring friendly pipe. The two J/Y things will be joined in the center for both rigidity and to facilitate a bracket to attach the TBs to. Because of TB and cylinder spacing, there will be an almost straight shot from the TB into #2 and a short horizontal runner to #1; likewise with #3 and #4. It’s pretty clear in my head, so hopefully my fabrication skills will keep up.

More ideally, there would be two equal length runners from each TB to it’s cylinders. I could probably accomplish this by feeding #1 and #3 from one TB and #2 and #4 from the other. Again, I’ll have to draw it out and see, and accurate drawings will be much easier the the arrival of the spare cylinder head.

Finally, the simplest manifold may be to simply take a 2″ or so pipe about 16 inches long, tap two 2.2″ TB connections on one side and 4 1″ engine intake connections on the other side and see what happens. No bends, a cap in each end of the big pipe, 4 flanges and 6 pretty much standard fishmouth pipe welds.

By the way, the sub-throttles can definitely be left in place on this unit. One kinda weird thing is that the fast idle cam arrangement on Buzz’s existing set of TBs is kind of done on these as well, but the cam is connected to the sub-throttles. When the are forced all the way open, the cam engages the main throttle, opening it about 1-2%. It would almost be funny if the IAC stepper function of MegaSquirt could be adapted to run this motor. [note: see this note in the VW Trike blog for the answer to that question – Sluggy] Perhaps someday someone will write some support for them into MegaSquirt. I don’t think I’ll ever had the time!

Woohoo! I won a really cheap auction for a cylinder head to use for fabbing the intake manifold for the smaller TB!

FedEx Ground tracking seems to indicate that it will not be delivered before next Monday. Phooey. I’ll be out of town all next week.

I’ve put some brain cycles into this intake manifold thing I’m considering. I’d still like to experiment with fiberglass fabrication at some some point, but I think I can make an effective steel manifold with less trouble overall. One of the biggest things is that a steel manifold will be relatively heat, vibration and fuel proof without resorting to exotic resins and generally speaking will just be a lot simpler.

I’m watching a couple of eBay auctions for cylinder heads, one of which has a broken exhaust flange bolt. That head would probably require repair for use on a bike, so maybe it will stay cheap. The intake side is just fine and I can thus use that head as a jig for building my intake. If that auction falls through for some reason, I suppose I can still use my own engine.

I see construction from relatively simple steel tubing. I could probably use EMT conduit if I had to, but I’d rather avoid burning zinc. I wouldn’t require a lot of material, anyway, so even hardware store quantities of 1″ or 1-1/4″ tubing would be affordable.

I can braze it together if needed. My torch will get hot enough to weld steel, but I have secured the relatively open-ended use of a decent gasless MIG welder, as opposed to the $120 Harbor Freight model that I might otherwise buy. I think this kind of construction might be easiest with an easy to use wire welder.

I also have a line on the short term use of a fairly decent metal cutting chop saw, which would make pretty short work of the cutting requirements for what I see in my mind.

When I get home tonight, I will do some basic layout and measuring to get started with a parts list.

It was a cool, but not unpleasant day today in the carport. I pulled the exhaust system and wirebrushed the area where the O2 sensor would go.

I decided that placing the sensor in the center of the collector was more important than the risk of getting the sensor oily. I marked and drilled the hole. I found, as was somewhat expected, baffles inside the collector. As luck would have it, however, my chosen location was not in the center of a baffle, but rather adjacent to one. I used a die grinder to relieve enough of the baffle to ensure that sensor would have access to exhaust from all four cylinders. The die grinder action went a little wild so the hole turned out a little oblong. At least it’s was smaller than the bung.

I suppose it goes without saying that I ran out of oxygen for the torch halfway through the brazing job, and it being Sunday, I should have just gone directly to Home Depot instead of trying closer places first. Even so, I think it came out very well.

I put the exhaust system back on the bike and mounted the LC-1 just below the steering head. That is probably not the greatest place, but I think it was the best place where the fixed cord length on the sensor could reach. I fished the wiring under the tank without removing it, an easy enough task. Solder on connectors, hooked it up, blew up an LED, another trip to AutoZone to get an almost but not entirely unsuitable replacement LED (I could have found a more suitable LED at Radio Shack, but they were closed by then) and finally fired up the ol’ LC-1.

Following the very simple directions for the LC-1, I calibrated it in open air then reinstalled the sensor permanently (I hope).

I hooked up MegaTune to the ECU and started Buzz up. Immediately, AFR was reading just over 12. I didn’t really need the gauge to tell me that the mixture was rich, but it was cool to see an actual figure attached to it.

I already knew that Buzz’s idle pulsewidth was just about at the minimum and it doesn’t look like I can affect idle AFR that way; the pulsewidth just can’t get much shorter. On the other hand, the pressure regulator seems to have drifted up to about 25 PSI. I need to connect the vacuum line to the regulator. That should let fuel pressure run low at high vacuum and higher at low/atmospheric levels. The lower pressure has also affected starting, so I need to tweak cranking pulsewidth as well.

I tried playing with the autotune feature of MegaTune. As expected, it had not room to adjust anything at idle, but I held the throttle at a few fairly low RPMs and was slightly giddy at seeing it automatically adjust the appropriate cells in the VE table. So, there are a couple of 30-something cells surrounded by 50 and 60-somethings.

That lead me to try again to generate a new VE table. It’s very frustrating that the obvious tool to use seems to think that my realworld parameters are erroneous. It want an idle MAP of <=60, but I’m sorry, it’s 72; that’s as low as it gets at idle. If it’s less that 72, Buzz is spinning down from high RPM with the throttle closed, not idling. Furthermore, I can’t seem to make the table generator happy with the horsepower and torque numbers. I entered the published horsepower spec for the stock bike, 50hp at 9000 RPM, but I tried the actual 9000 RPM (no load) MAP in my log of 79. Nope, sorry, the table generator prefers MAP to be =>90 at peak horsepower. I put in 90 like a good little sheep and it makes a table full of big values. So, I guess the table will be mine to generate manually.

Of course, all this MAP trauma is probably because my throttle bodies are too big, which I have suspected for a long time. I might have to build that intake manifold after all.

It’s still fun, in a kinda masochistic way

My LC-1 arrived today!

The brain is a little bigger than I had imagined, but it occurs to me that I’ve never seen one adjacent to anything by which to judge scale.

Ok, it’s not really MegaSquirt related, but it *is* Buzz related. He’s gotta be able to stop, too….

I won the auction on a little nicer master cylinder. Ironically, its off an ’03 ZX636, like the throttle bodies I’m using.

My goal was to get a newer cylinder instead of rebuilding the old one again. I will need to see if my caliper will work as is. The issue is one of hydraulic ratios. If the diameter of the master cylinder is too much smaller than the stock cylinder, it may not move the caliper piston enough, although it would move it with great force.

I’d like to replace the brake lines with braided stainless steel lines, but I may *have* to replace the upper line to reach on the new master cylinder. Like many bikes, there is a line from the master cylinder to a manifold of sorts with another line running from there to the caliper.

I was a dummy and left my laptop at work. It’s not too far back to the office, but I didn’t wanna go back.

I did take Buzz on a little test ride.

He was a little hard to start cold. I will need to adjust cranking pulses and afterstart enrichment. Once started, I let him warm up while I went inside and geared up, sans wallet. He was idling nicely. The FIDLE light was still lit, but it was off shortly after taking off.

First I went down the street and just played with basic acceleration. He seems to be starving a bit for fuel. Not running lean, at least no so lean as to knock, but he just isn’t running at full power. This is no surprise since I’m running less than half the fuel pressure I was. On the other hand, he seems to have lost the majority of the surgey cruise. Similarly, he’s a much more predictable for low speed cornering. Before, the engine was either putting down power or it wasn’t. Now, rolling the throttle on does not induce a sudden jump as the engine transitions from braking to driving. All in all, street drivability is much better. I’m looking for forward to actually tuning on it.

As I had planned to top of the gas tank and see what my fuel consumption was last weekend, it came to my attention that, as mentioned earlier, I was without my wallet. I ran back home, retrieved said wallet then proceeded to the Chevron. It took 2.100 gallons to top off and mileage was 49.3, for an MPG of 23.47. The leak in the first tank definitely made a difference, but 23.5 MPG is still way below what I both hope for and expect to be able to get from this setup. After fueling up, I took a short run on the freeway to see how he gets along with others. I’m definitely missing the power, but I’m not missing the surge-y miss-y cruise. There is still a very tiny bit of sputter, but in my short ride, I had to look for it. Just gotta start actually tuning the engine.

To that end, I finally bit the bullet and ordered an Innovate LC-1. I have been chasing them on eBay for a while and every time there was an auction for a decent price, some yahoo (lower case) would bid it up to retail price in the last hour. Note to bidders: If you’re willing to pay retail price, STOP BIDDING UP cheap ones and just go buy one already.

It is the standard Innovate 3769 (LC-1 with sensor, bung, cables and CD; no gauge) for $159 with free shipping. This model is $199 retail, and I was getting tired of chasing used ones for several days, only to have someone bid it up to nearly new price, so this was the deal to make. As of this writing (1/19/09) they have 20 left at this price. I wanted a gauge and I’ve found the Innovate 3795 (same thing, but with blue digital gauge) as low as $199, but then it gets bid up to where bid + shipping => retail. Stop it!

I imagine installing the sensor will be the biggest hurdle, and really sensor *placement* will be the bigger difficulty. At this point, I plan to install the sensor in the front of the exhaust collector, relatively unseen, with the body of the sensor parallel to the ground. I think the biggest risk here will be getting oil on it when changing the filter. Installation should be a relatively straight forward process of cleaning the area, cutting a hole (with a holesaw instead of a drill, I hope) and brazing the bung in. Buzz may one day sport some form of aftermaket exhaust, so I may get to revisit the issue then

Before I get too involved with tuning, I need to set the valves and synch the throttle bodies. I’ll need to check again, but I’m pretty sure there are pilot air screws in the bottoms of all for TBs.

On a sort of unrelated note, I’m also shopping for an aftermarket brake master cylinder. This one is worn enough that the rebuild kit I put in it isn’t really doing much good. The brakes are still spongy and I’m pretty sure it’s well bled out. Anyway, nice Brembo cylinders tend to be about $300 and while one’s life should not be priced below $300, it’s still hard to fork over dough on something so generally taken for granted as brakes.

One eBay seller had a very nice Brembo master cylinder and in a separate auction, a matching hydraulic clutch cylinder. The clutch was running at $38, the brake at $270. Guess we know which one everyone wants!

I want’em both. I’m not opposed to converting the clutch to hydraulic. It’d be trick!

I have done a little tweaking of the fuel pressure, so far to good effect.

With the bike warmed up and idling, I reduced FP from 45 to 20 psi. Idle is a bit smoother and though I do not yet have any sort of EGO sensor, my built-in unburned hydrocarbon detector (nose) seems to indicate that it’s not nearly so rich. No-load reving reveals MUCH less black/gray smoke as well. I was able to create a rather James-Bond-y smoke screen by reving up against my fuel cut rev limiter. Not so much, now.

I hope to get in at least a short ride this afternoon, if for no other reason than to see how Buzz rides now.

The theory I’m working on is that a required fuel setting of 1.8 for (4) 245cc/min injectors is just too rich and the pulse width probably can’t get any shorter reliably.

By reducing the pressure, I have essentially changed my injector flow rate to 166cc/min. The injectors may not atomize the fuel quite as well at this lower pressure, but the observed effect on Buzz seems to be an improvement.

It’s been a busy week, so this ride report is really about riding Buzz around *last* weekend.

The ride back to work was fine. I had a BTW meeting to attend, so I took Buzz there, too. I took a very short jump on the freeway, but it was a very windy evening and I don’t like getting shoved around so much. A few people there have been loosely following the progress. They were pleased to see Buzz on the road, although they would be more pleased to see the VW trike on the road.

After the meeting, I met my wife at work and we took a slightly long way home, by way of the local watering hole. Buzz ran reasonably well the whole way.

Overnight, I forgot to turn off the fuel valves, which would not normally be a problem, but apparently something was leaking. I turned it off, but it was a while before we would be going anywhere, so it was all dry by the time I got to troubleshoot a bit. I’m about 90% sure it was the inlet to the fuel filter, and since tightening that clamp, there does not appear to be any other leakage.

Especially after the leak, the tank was pretty low. It took 2.8 gallons to top him off, and with only about 50 miles run, that works out to 17.5 mpg. An undetermined amount of fuel went on the ground and I still have not done any tuning. We rode some on Saturday and some on Sunday, including a few miles on the freeway and a few minutes of twisty maneuvering in an exceptionally smooth parking lot. Buzz is a lot of fun to ride, overall, although at very small throttle openings, he runs pretty rough. I haven’t had the chance to top off the second tank yet and it doesn’t appear to be leaking anymore and even with a really rich mixture, I expect his mileage will go up.

Getting to ride Buzz and having him behave pretty well really helped because, wise or not, I began adding up the money I’ve spent thus far.

I had an envelope that had a few receipts in it. I consulted my eBay and PayPal records for more. Last night I dug through a stack of receipts at the house. I have another stack to audit from a desk drawer at work.

I have not found every receipt, but I have most of them. Some of the figures I have, especially ones I gleaned from my bank statements, are for the full purchase amount for transactions where it’s quite likely that not everything was for Buzz. I have dithered back and forth one whether to count tools and/or supplies bought during the course of the project. Arguably, I would not have needed them all without working on Buzz, but they are still usable for non-Buzz tasks. Likewise, do I count things that I bought for Buzz, but either lost/damaged and had to replace or just didn’t end up using? There are also a couple of things left to buy, most importantly some form of O2 sensor, preferably wideband.

I will post an itemized list or two later, but for now I can say that converting Buzz to EFI, before O2, ran somewhere between $915 and $1925, depending on how you add it up. More than half of that total was split between DIYAutotune, Del City and K&N Filters. The auto parts stores and the hardware stores got the majority of the rest of it. Ironically, the critical components like the throttle bodies, fuel pumps, etc were fairly cheap parts. It’s the little stuff that adds up. In reviewing other people’s MegaSquirt projects, I expected about $1000, but not $2000. By that last figure, the EFI system is worth twice what the bike is…

In the full analysis to come, I think I will be able to show that it *could* be done cheaper by cutting corners in certain ways, such as using solder and tape splices wherever possible rather than connectorizing everything, or even by using cheaper connectors. I could have used cheaper air filters, worked around the neoprene reducers, etc. Organizing my shop would help keep me from having to purchase the same materials twice due to loss. Haste makes waste, too. If I’d not rushed to ride, I may not have forgotten the filter and with that, would not have destroyed the first fuel pump. Similarly, “gotta have it now” costs more than shopping around.

The experience I’ve gained will greatly benefit the next project, most likely the VW trike. Some of the issues that apply to a bike may not apply to the trike at all, saving me all kinds of money and time. Plus, I already have a lot of tools…

Things can’t be too bad; I took Buzz to work today without incident.

Of course, I had to show him off, so three trips outside and at least one more planned. Since the car wash was closed last night (and since the ECU isn’t actually waterproof at this point), I’m glad I spent 5-10 minutes using the Armorall Exterior Detailing Wipes to clean him up a bit. Made the shiny bits shiny. He’s been parked unwashed for the better part of a year.

On that subject, it came to my attention that my first blog entry on this subject was January 10, 2008. Today marks one year, pretty much to the day, since I got serious about the project. Now, I didn’t start taking Buzz apart until March 20, but the actual project, from “yeah, I’m gonna do it” to riding a fuel injected Buzz to work was one year.

There have been many interruptions along the way and gaps during which little tangible progress was made. I have re-engineered a few things along the learning curve. Knowing what I know now, once the parts were gathered, a conversion like Buzz has turned out to be as of today, would probably take me about a week of evenings.

I do not consider the project finished yet. Buzz still runs rich, especially at low RPMs. I still need to adjust the valves and bolster the ignition system. Since I’m planning to Squirt the VW trike, I may order a MicroSquirt and put it on Buzz and use this ECU for the trike. MicroSquirt is ready out of the box to drive two coils. With the adjustable fuel pressure regulator, I hope to be able to rein that in. If not, I have the other throttle body and the intake fab *that* will require. I still want to put a wideband O2 sensor in there.

Of course, none of that addresses the cosmetics I have in mind for Buzz.

In any case, it’s lunchtime, time for another ride….