With all the concern about the short life of the sensors, I attempted to make a simple sensor test jig. I was hoping that I could come up with an easy way to repeatably test my sensor against one that would not be used on an engine (my gold standard).

I was thinking about how complex the math must be (I am not a chemical engineer) to back out the AFR from the combustion process. I tried to find a graph that would compare % or PPM oxygen in nitrogen to AFR. I really had no idea how sensitive these things are.

To start out I just used the old Dynatek sensor. I wanted to see if it really was linear.

To get the oxygen I just use air and dilute it with nitrogen. I did not want to have the gas shop custom blend mixes for me. Basically using a couple of regulators, two adjustable rotometers going to a T and a long line to mix the two. I then just welded a bung to the end of a pipe, capped off the other end and mounted a couple of fittings for the in and out. The out is just vented to atmosphere with another long line. I strapped 4 Watlow firerod heaters to the outside of the pipe.

First attempt was way too sensitive. Using pure nitrogen versus 150ml nitrogen mixed with 10ml of air was enough to drive the sensor from one rail to the other. I guess that would be about 1.4%.

I then diluted the air with more nitrogen then mixed this in with more nitrogen. This gives me some control.

A few things I notice is that even with just the nitrogen gas, I can't get 5 volts out of the sensor. After purging all the lines with nitrogen then letting it settle for about a half hour the voltage made it to 4.621. It does not appear linear at all. I did not want to cut the unit up but would venture to guess that the black box is nothing more than an amplifier. Wonder now if anyone ever used this setup with any success. Seems like you would have had to pick your sweet spot using a dyno and a real gas analyzer for a reference.

Now that I can at least control the ratio somewhat, next step would be to make up a new test tube with two bungs. The LM2 can read two sensors so I can now just sweep the gas and compare the two.

Seems like there would be an easy way to know how the sensor is performing.

Test setup with the Dynatek O2 sensor

Not very good heat transfer...

Notice the sensor will not go to its full range with just its heater. The external heaters must be used.

Notice how sensitive it is to supply voltage.

Without a fast way to purge all of the parts and lines used for the air, the small amount of O2 left really slows the system down.

The new jig...

setup with the Bosch reference and Dynatek sensors


I took data from 30 to 14.7, with more points taken as O2 levels were reduced.

After running the test with these sensors, I had to dive into how they work and what is really go on. I was way off in my thinking. My concern with this test setup is if these sensors can fail in a way that the lean side would be fine while the rich side is not. While the Bosch data sheet does call out two test gas blends to get two different calibrated lambda points, with the reference sensor I don't think I need to go that far to test it. I plan to call the local welding/gas supplier and see what they can mix.

good stuff geek.keep the good work up mate.cheers.mark

The local welding place wants $180 to blend up an non certified 2-gas mix. The fully certified mix with papers is much more. For that cost I could just buy more sensors.

I spoke with a friend of mine who tunes bikes on a dyno and he uses these Bosch sensors. He say's that they always seem to fail the same way where they go flat at 14.

To test on the rich side, I may try some draino (drain cleaner) or get my soap suds out. I am thinking I could just feed this into the low side of the air pump to get the pressure up.

I did manage to finally get the Innovate to the track and got some pretty interesting data that I will show later on.

Made a few easy passes to try the Innovate out with my logger. I was not doing a burnout, leaving slow and manually shifting. Once it got moving I would open the throttle. Tracks plenty long enough, so I would stay into it long past the finish line until it peaked out.

I started out with my ultra fat jetting and sure enough, the LM-2 shows it running lean. The next jetting allows the motor to run without missing (at least from what I can hear). The engine does not make much power and it ran about 120 mph before I closed the throttle.


It took about 250mS for the wideband to react to the closed throttle.

As I continued to lean it out, the sensor does not appear to track. I had thought that it would be very accurate down to 8 but I am not so sure.

For the last two passes, the jetting was good enough to make some power so I did a burnout and ran the bike normal. Speed was now in the mid 150s. One of the setscrews on the sprocket ring backed out (next time red) and hit the shaft sensor. So I can only go by the tracks ticket at 154.15. Still far from close but what is interesting is that even with the jetting now much leaner, the Innovate actually shows it being richer by about 0.2.



Looking at the plug's heat lines and how slow the bike is, I know the jetting is still a long way off and it could be around 9. This picture of the sensor was taken after my last pass.



I wanted to compare it with the virgin sensor in my test jig, then discovered that the LM-2 can only handle 1 sensor. Fuck. Sure enough they have a different part number for the two channel unit. I tried to call them to see if it could be upgraded but after sitting on hold for several minutes, I gave up. Still no forum and they have never responded to any of my emails, so support is what you figure out on your own. I took the LM2 apart and even though they have the second connector, there is a small 1" sq. mezzanine board that is not populated.

So, I ran some tests where I used the virgin sensor to setup the gas mix, then would run the second sensor. A pain in the ass. The sensor appears to be fine. So what ever is happening, it must be normal. Maybe they are no good below 10.

You may have to be careful in that alot of todays race fuels are oxygenated and if it is in a rich condition the unburnt fuel has unburnt oxygen hence the I'm fat but think I'm lean. I think oxygen sensors are only a guide and you always need too read your plugs. I run 5 sensors on my bike and look at plugs after every run. The best system I've seen is the peak cylinder pressure sensor that is mounted in a special plug corrilated with deg of timming but at over $20 000 for the set up I saw I think I'll pass lol. I have a friend of my brother who is an electrical engineer he sub contracts for defence and is looking at making me this system fingers crossed he will pull it off. He does all the digital data logging on the fighter jets and knows of the sensors that can do it the hard part will be mounting we may have to drill a dedicated hole in the head time will tell. Keep up the good work

I use C16 not Q16. It doesn't mean that this isn't what I am seeing. We know when it is on any limiter, it will go very lean. This makes sense with all that unburned fuel and air getting into the pipe. You can see it do this as well when it shifts and the kill is active. What's a little strange is that once its reached full boost and everything is stable, the numbers are not ultra lean like a miss fire. 0.2 points is not what I would have expected if there was a miss.

The nitrous seemed to effect the heat line. I am hoping that the wideband may shed some light on what is happening. For now, conservative tunes and no nitrous.....

What are the 5 sensors on your bike measuring?

I have wondered about the pressure sensors after we talked about it on dragbike.com. I still wonder about the Motorola ignition as well. Cheap way to look at combustion events. We need to invite our Nology friend to this site. He should have made some progress with his setup by now.

When I removed the diaper to drain the oil, there was the set screw from the shaft sensor. I made a new bracket for the speed sensor. Lucky the sensor was not damaged. I put new setscrews in with red, then covered the holes in RTV. While I was turning the wheel to check the sensor I noticed the chain broke. I got lucky this time.



They must not plan on heavy riders on heavy bikes. Chain was not near the stretch limit. Link that broke was on the out side. The setscrew that fell out is on the inside, so doubt the two are related. It's a tiny ass setscrew anyways compared to the SHB chain.

5 oxygen sensors one in each pipe one in the collector

Nice setup. How do you test to make sure all 5 track or are you just looking at relative numbers?