Note: I have 25 photographs for this thread so this will have to be in six posts in order to have the breaks in reasonable locations. I'll upload the posts over the next few days.
While making an extensive series of measurements to determine how to modify an Amal 1036 to work well on a Gold Star I found the bike "felt" like it ran fine over a very wide range of Air Fuel Ratios (AFR), from ~10:1 to ~15:1. From this work I concluded plug chops and seat-of-the-pants tuning just don't cut it. Read the above thread for the background to my conclusion. If you disagree, now would be a good time to close this thread and read another one.
OK, you're still reading, which means you agree with my conclusion and want to know more about how to make AFR measurements. In this thread I'll describe three levels of instrumentation ('Silver', 'Gold' and 'Platinum') that will provide the information you need, with each a step up in capability and/or ease of use.
I bought my first AFR unit 16 years ago and, although the brand (Innovate) still exists, a decade ago the company it was bought by another company and the level of support they currently provide is sub-par. However, my current AFR instrumentation package evolved from that first one so incorporates components from the same brand, so that is what I'll describe in this thread. If I were starting fresh today I would investigate all the available options before deciding what brand to settle on. Still, irrespective of brand, what follows in this thread should be of general use to anyone thinking about assembling their own AFR system
I. General Background
A. Compatibility Issues of Software and Computers
AFR instrumentation brings jetting the carburetor of an old motorcycle into the 21st Century, but bad things come with the good. Any device that downloads its data to a computer for analysis inevitably involves issues of software and hardware compatibility. One of the devices discussed below was developed in the early '00s at a time of Windows XP and RS-232 serial connectors, so getting it to run on a Windows 10 computer with a USB connector could be as simple as adding an RS-232 to USB adapter, or it could require locating an old version of software to install as well as significant debugging effort. Even if a new device works fine on a given computer running, say, Windows 10 doesn't guarantee there won't be problems getting it to work on a different computer that also runs Windows 10. Anyway, I'll try to answer questions about AFR instrumentation and its physical installation on a motorcycle, but life is too short to try to attempt to debug someone else's software issues remotely.
B. Wide-Band Sensors
Without going into any detail, a Bosch wide-band sensor generates a voltage that can be used to determine how far the composition of the exhaust gas is from that of a perfect air/fuel mixture that burned completely. In the case of perfect "stoichiometry," defined as λ=1, all the oxygen would have combined during combustion leaving no free oxygen in the exhaust gas. If any oxygen is detected, the mixture was lean (λ>1), and if unburned hydrocarbons are detected, the mixture was rich (λ<1).
The Bosch AFR sensor should be at least 24" from the exhaust port to be sure it does not exceed 900 °F/ 500 °C. At that temperature steel is just starting to have a faint glow, which I've never seen even near the head since a motorcycle's pipe hangs in the breeze instead of being insulated under the hood of a car at the end of a header fed by four cylinders. The sensor has an internal heater sensor and should be as near as possible to the top of the pipe so water condensation doesn't contact it when the engine starts up and possibly damage the sensor due to thermal shock.
Additional cautions are that the lifetime of these wide-band O2 sensors is degraded by the use of leaded fuel, gasoline additives, excess oil in the exhaust, and oil with high Zn content. Luckily, none of those are issues with old British motorcycles...
C. AFR vs. λ
All gauges that use wide-band sensors intrinsically measure λ and then at least some of them calculate the AFR assuming the 14.7 stoichiometry of "pure" gasoline (unless they're programmable with different values). Unfortunately, the stoichiometry changes when ethanol is added, dropping from the 14.7 of gasoline to 13.8 for a mixture with 15% ethanol (nominally 'E15'). In what follows I'll use 'gasoline' to mean fuel without any ethanol added, and 'E10' or 'E15' for gasoline with 10% or 15% nominal values of ethanol added.
These days any random refueling stop could leave your tank filled with fuel containing anywhere from 0% to 15% ethanol, i.e. fuels having stoichiometric ratios from 14.7 to 13.8. I'll have more to say about this shortly, but it's important to note here that even if a pump is marked, say, 'E10' or 'E15', U.S. regulations allow the ethanol content to be up to 10% and 10.5–15%, respectively, so the 'E10' you use when jetting your carburetor could very well be different than the 'E10' you fill your tank with on subsequent refueling stops.
Despite the AFRs being different for the various ethanol blends, it turns out that, irrespective of the precise fuel (e.g. gasoline, E10 or E15) maximum power occurs when λ=0.85–0.90, i.e. when the mixture is somewhat rich. So, I suggest setting your unit to display λ, and jet your carburetor to result in λ ≈ 0.85–0.90 for whatever fuel you use, whether it be gasoline or E15, or anything in between. In what follows I will use the general term 'AFR' to refer to the air/fuel ratio even though I suggest always working with λ.
Having just written that you should use λ rather than AFR, the following example of data is plotted in terms of throttle position and AFR vs. time where an assumed stoichiometry of 14.7 was used.
This example shows the information an AFR system gives that otherwise wouldn't be known. In orange is the throttle position, with the long plateau at ~1.5 Volts corresponding to applying full throttle in 4th gear starting at a time of ~10 min. 5 sec. Without going into detail, note that from the time full throttle is applied it takes ~18 seconds (until 10 min. 22 sec.) before the AFR finally stabilizes at its final value. Even if a plug chop gave reasonable insight into the mixture, any chop made sooner than that would falsely imply the main jet should be increased in size to give a richer mixture.
D. Effect of Different Ethanol Mixtures on the Jetting
What an AFR of 13.8 vs. 14.7 means is E15 requires less air than does gasoline for a fixed amount of fuel. Or, equivalently, more fuel for a fixed amount of air. Since an engine sucks in a fixed amount of air on each stroke, and if jetted perfectly for gasoline, it will run leaner with E15. But, only 6.5% leaner. I suggest you do your jetting with a gallon of expensive ethanol-free gasoline purchased from a speed shop, and jet it at the 0.85 rich end of the λ range. If you subsequently fill your tank with E15 that happens to have the full 15% ethanol composition, λ would be 14.7/13.8 × 0.85 = 0.905, which means you'll still be fine. If the ethanol content is less than 15% it only will improve a situation that doesn't even need improving. Even if you use gasoline to jet it at 0.9, with E15 that will be 0.96. You'll lose a bit of h.p. but slightly lean 1.06 shouldn't be a problem for an engine used relatively conservatively.
E. Effect of Fresh Air on a γ/AFR Reading
Because of how a wideband sensor works it is extremely sensitive to any amount of fresh air that reaches it. It's not that some amount of fresh air that reaches the sensor will cause some amount of error, any amount of fresh air will cause a huge error that will be impossible not to notice. What this means is the actual sensor has to be located far enough upstream from the outlet of the exhaust pipe so that pulses of fresh air sucked back in on each stroke can't reach it. At the same time it has to be far enough downstream from the head that it doesn't receive excessive heat from the exhaust that would affect the readings even if it didn't damage the sensor. Luckily, there is quite a lengthy sweet spot with any of our engines for installing the sensor. If a sensor is permanently installed in a bung welded on an exhaust pipe, the weld must not have any pinhole leaks in it. If it's welded in a "good" location on the exhaust pipe it will accurately determine the AFR. If in a "bad" location it either will measure a value so far off that there won't be any chance of thinking it is real, or risk damage or error caused by too-hot gas.
II. Instrumentation Packages
As I wrote earlier, the first of my AFR packages is from 2004 so different instruments available today from other companies may offer better features. However, to save money, you can build a system exactly like one of those described below using older components from eBay. You will need to research this yourself, and re-read the note above about possible software and hardware incompatibilities.
In what follows I describe sets of instrumentation capable of providing the necessary data to allow you to accurately jet a carburetor. I've termed these 'Silver', 'Gold', and 'Platinum' in order of increasing level of capability.
A. 'Silver' Instrumentation Package
My simplest instrumentation package consists of an Innovate LM-1 control/display unit (and Belkin RS-232 to USB adapter), Bosch wide-band sensor in a modified Innovate exhaust clamp probe, a small external battery for motorcycles that can't supply the necessary 1 Amp at 12 Volts to power the sensor's heater, some way of mounting the display so that it is easily visible when riding,
Converting from RS-232 to USB requires some electronics in the adapter rather than just a simple rerouting of the wires so a given brand of device could be incompatible with a computer.
For mounting my display I fabricated a simple bracket held to the handlebars by a Manfrotto "Super Clamp" photographic clamp.
Although this system can provide all the information one needs to know to jet a carburetor, it places a heavy burden on the rider. With the throttle marked at the ¼, ½, ¾ and full positions, the rider has to mentally average the fluctuating values of λ at each given throttle setting, and remember those results long enough to write them down along with the throttle positions.
The Innovate LM-1 controller logs data 12 times/sec. for 44 minutes which, although it can be later transferred to a computer for more careful analysis, still requires know when during the run the throttle was at a given position. Innovate's current LM-2 model records at the same 12 Hz, although it can record 32 channels vs. only 5 for the LM-1. Also, the LM-2 data is recorded on a 2 GB card so has a much longer recording time than the 44 min. of the LM-1. That said, I found the 44 min. to be adequate for all but an extended road test after the final jetting has been determined.
The LM-1 requires Innovate's 'LogWorks2' software to download and display its data for analysis. The version of this software the company currently has in the "Legacy" section of their webpage does not work on my Thinkpad X250 computer running Windows 10 computer, but an old version I have on a CD from 2004 does work so that is what I installed. I have a newer laptop but have not tried installing the software on it since I continue to use the older Thinkpad for other tasks as well. Anyway, this is an example of the software issues you might encounter that I mentioned at the start of this thread.
2. Removable Probe
As shown in the next photograph, this 'Silver' system uses an earlier version of the Innovate Exhaust Clamp probe that I modified to make it suitable for use on motorcycles.
The design is like that of a Pitot tube, with the exhaust flow past the holes in the side of the shorter tube reducing the pressure to draw the mixture in the end of the longer tube, past the sensor, then back through the shorter tube and out the holes after a long round trip. I added a spacer on the inlet tube to make sure it can't sit flush against the exhaust pipe where the flow is stagnant.
Unfortunately, an Innovate probe with the stock length has two limitations, one more serious than the other. The more serious limitation, in the sense that there is no solution for it, is that baffles in a silencer can keep the probe from being inserted very far into some exhausts. The less serious limitation, because it can be solved, is that fresh air is sucked some distance up the exhaust pipe every other revolution of a 4-stroke engine where it can reach the inlet to the fairly short probe and invalidate the readings.
Since the stock probe is basically useless as-is on most old motorcycles, I extended it by 10" so it now is in the form shown in the first photograph. This elongated sampling unit lets me determine the AFR on the road without having to modify the exhaust pipe or use a "Bunsen valve," as described next. However, whether or not this lengthened probe can be used at all depends on the design of the silencer, so it won't work on all bikes. For example, it fits in the pipes of my Matchless G80 and Triumph 500, but not the Trident. Still, for bikes where it will work it is an alternative to either a Bunsen valve or permanent modification of an exhaust pipe.
3. "Bunsen Valve"
It's possible to overcome the problem of air being sucked up the exhaust pipe with a "Bunsen valve" arrangement, although it's not an ideal solution. The next photograph shows the Bunsen valve I made from high temperature silicone material and Velcro that wraps around the silencer. The pressure pulses from opens the cloth to let exhaust out, but closes it when the pressure drops, which keeps air from being sucked in.
Unfortunately, since the silicone material flaps in the breeze it isn't suitable for really high speeds or for long distances.
MM, I have yet to find any of your projects uninteresting, although I'm not a class room guy so I confess I do sometimes struggle to keep up. Please continue to post even if it is for the minority.
As for this threads particular subject matter, I would not do this to my old bikes. Partly because with a young family I would not be able to justify the cost. Also because I work all day on modern vehicles, spending hours trying to interpret data from a scan tool to diagnose faults has taken the fun out of my job and my bikes are a welcome respite. The most fun I have had of late was a 52 Morris Minor that the owner could not get to run right. Seized mechanical advance, broken points wire, no points gap, incorrect rotor arm fouling the cap and an incredibly week mixture. Gave me chance to drag out the old analogue dwell meter, timing light and colour tune. Great way to confuse the younger guys in the shop.
Regarding bungs in headers, in the light aviation world, Exhaust Gas Temperature sensors are now commonplace. These work by inserting a probe or probes into the exhaust headers, and routing them back to the panel where the information is presented on a simple guage. My Piper PA18 had an EGT setup, which was installed before I bought it.
This is not all that relevant to MM's work, except in the sense that attaching EGT probes into exhaust headers is something that is routinely done on small airfields all over the world by knowledgeable airplane mechanics (known as A&Ps in the US.) A knowledge base that is well within reach of those of us with old motorcycles to set up.
This info is so immoderate. hats off to ye MM, I would be happy with basic instrumentation, Bronze age package.
I have previous with Resistance temperature detectors, RTDs, no matter what type they are very prone to spikes and over reading if all the connections are not A1 , IMO they are not to be trusted on vibrating equipment, but, better than nothing. The only temp read outs I have faith in come from Mercury in steel. The Lambda sensor stuff is new to me , taking it in sponge style.