sniffer development and testing

Object 1: See if it is possible to make a sniffer apparatus for the o2 sensor that can be adapted for the wide variety of 2-stroke scooter exhaust pipes.

Object 2: run the same sensor on at least two test bikes using both the sniffer and a welded plug (300-400 mm from the piston face)]

prototype sniffer

My first one followed the general instructions on the sensor instructions. This was made from tube about the same as a stock vespa and lambretta tailpipe. The idea is that I can use it INSIDE a larger expansion chamber - which would be the most common and important use for the device - and also attempt to attach the end to a stock style scooter using rubber tube and clamps. This would force ALL of the exhaust through the sniffer device.

The section on the sniffer where the sensor is attached is much larger in diameter in order to slow the velocity down and prevent cooling of the sensor.

Already I have come across a pipe that fits neither description. I attempted to run a much finer tube (12mmod) into the tailpipe and let that pipe stick into my sniffer device. The first casual test indicates a leaner reading than results using the bung plug on the same scooter. I will now attempt to seal the end of the pipe and force all of the gasses through the sniffer. See what happens then.


- Test scooter 1 - malossi 110 PM tuning pipe - sniffer pipe is about three quarters of the outlet diameter. Sniffer pushed as far into the tailpipe as possible. - reading shows very lean (18.6) on idle, going to a perfect 12oclock on the dial in the midrange then riching off at 10:30 at full throttle. These are not under load, just initial tests for the sniffer.

This result led me to ponder the possibility that low revs did not read well at the tail pipe.

- Test Scooter 2 - malossi 177 with malossi (1980's) pressed 'faux' expansion chamber. This pipe has a narrow exit hole (smaller then the sniffer) and a flat end can, preventing any easy way of attachment or sealing.

So far I took a smaller tube and poked it into the end can. i then allowed the other end to poke into my sniffer.

This result indicates - 1) that the results of test scooter 1 may simply be that the scooter has a very lean idle jet, considering that test scooter 2 showed a reading at idle - 2) the temporary fix for Test Scooter 2 is inefficient and showing a leaner mixture than the more clinical 'screw in sensor'.

next test

test scooter 1 - blocking the air filter pod and winding out (richen) the fuel screw changed the reading to a bit lean rather than very lean at idle. I have bought richer idle jets for this carb and will test further tomorrow (under load)

test scooter 2 - taping the thinner tube into the sniffer thus sealing it and poking the thin end into the end can proved to show the same results as the bung plug method. so far so good.


- the welded bung gives an accurate reading of o2 levels, making it a benchmark for testing a sniffer

NEXT TEST - on the dyno

Test scooter 3 - 186 kitted lambretta with stock style pipe (cloverleaf shape tailpipe)

i attempted to run all the exhaust through the sniffer usinf hose and clamps. This got a reasonable reading showing a bit rich. Under load the readings showed a slight lean spot at about a third throttle. Although the scooter showed a slightly rich setting at the main jet, at full throttle ther is a dramatic drop in power and revs.

When run on the dyno using zero to half throttle the bike accellerated smoothly and well showing ideal 02 levels.

Removing the air hose seemed to make little difference to the readings

I then adapred the sniffer to reduce to a smaller tube (sealed this time) so the majority of the exhaust pressure would bypass the sniffer. This seemed to give similar results, which is good news as it is much easier just to place the sniffer in the end pipe.

to add sniffer prototype image


  • poking a smaller tube in the tailpipe does not seem to change the results compared to channelling ALL of the exhaust gas.
  • Although the bike was clearly far too RICH at the main it didn't really show very quickly at the sniffer. The reading did change to the richer side so I guess that's a clue
  • The readings are all leaner under load.
  • one particular throttle position spiked lean several times in a row.


For now the jury is out until

1) I develop a standard sequence of testing so I can record the numbers at particular throttle positions and compare changes accurately

2) the o2 sensor readings are logged to software

3) I have welded bungs on several scooters and tested them both ways under load to see the accuracy of the sniffer.

/@api/deki/files/1093/=limone_acido_o2_test.xls limone acido o2 test.xls


2) manually done

3) fully tested My 'Lab Gerbil' (see spreadsheet) in three ways using  sensor in Sniffer at the rear, sensor in welded threaded bung 32 cm from piston  WITH SNIFFER STILL IN PLACE, and sensor in welded threaded bung with sniffer removed. Very interesting results. One thing to consider is that the pipe used on this scooter has quite a narrow hole (around stock) so the prescence of the sniffer could have less effect in a larger expansion chamber.


I found with the prototype sniffer, the performance of the expansion chambers was affected by the prescence of the pipe in the chamber. I remade a new sniffer using much narrower, semi-flexible copper tube. (photos to follow).

I am also finding that the tuned scooters at higher revs seem to consistently display rich readings. I have tested this against traditional methods of testing main jet size and have to conclude that for now the o2 sensor must be used only for idle and mid range.

MAY 2012 update

I have finally laid eyes on images of the dynojet o2 sensor. It is similar to my latest sniffer, only uses compressed air to draw the sample throught to the probe. onsidering the dynojet system is 3000 dollars it is tempting to try and adapt mine to this system. The only problem is the datalink module required to run your own sensors into the dynojet stack is 1200 dollars.


someones interesting experiments....

Paul McIntosh,
Mar 25, 2015, 12:46 AM
Paul McIntosh,
Mar 25, 2015, 12:46 AM
Paul McIntosh,
Mar 25, 2015, 12:46 AM
Paul McIntosh,
Mar 25, 2015, 12:46 AM
Paul McIntosh,
Mar 25, 2015, 12:46 AM