In the last few years, Ducati has recommended a new method of determining belt tension. The old method relied upon a spring-based tool to measure belt deflection while the new method expresses tension as the frequency (in Hz) at which a section of belt will vibrate at when plucked like a guitar string. The more tension a belt has, the higher the frequency. This method is more accurate and has excellent repeatability. As long as the measuring device has a reasonably accurate clock and is able to count every oscillation, it will remain accurate and periodic calibration is not required.

I'm not the most cautious person when it comes to belt tension as is evidenced by the fact that I've been setting the tension using the guestimate method since I bought my first Ducati. And I'm not shy about running my belts 18,000 miles and well over the two years Ducati recommends. I haven't exactly been losing sleep over this and the old guesstimate method worked fine for me but I am pleased to have discovered a new method to verify proper tension without spending a lot of money for the expensive factory tool that may only see occasional use. This method is just as accurate as the factory tester with optical probe and more accurate and repeatable than the old spring tension gauge that requires periodic calibration. When I used the guesstimate method I never knew how close or far I was from potential trouble.

The graph below is a visual representation of a .wav file I recorded using the built-in microphone of my camera when I plucked the cam belt like a guitar string. I have found I get the best recording with the microphone facing one of the flat sides of the belt near the midpoint where it is plucked. I do not position the microphone facing the edge of the belt, as this may not provide a clean recording. I have also had good results with the microphone in contact with the metal cam pulley. Having done it both ways, I find the non-contact method a bit easier to use.

I downloaded a shareware audio editing program called "Goldwave" to analyze the recording. After opening the audio file, I selected the portion of the recording I was interested in by left-clicking to delineate the left border and right-clicking to delineate the right border. The result is shown above. As you can see at the bottom of the screenshot, Goldwave displays the length of the selection as 0.171 of a second. Multiplying the reciprocal of this number by the total number of cycles (waves) in the selection will provide the frequency in Hz. We have 25 cycles displayed in the example above. This belt measures 146.2 Hz (1/0.171*25=146.2) that is too tight and should be slackened before the engine is run.

Ducati had a bewildering array of belt tension specs that varied depending upon the engine; some engines even had different specs for the horizontal and vertical cylinders. In March of 2009, DNA released a service bulletin containing new belt tension specs. According to this bulletin, "all models" can be adjusted to 110 Hz (+/- 5Hz) on both cylinders when replacing with new belts.

They list the following models:

Streetfighter, 1198, 1098, 1098R 999, 998, 996, 916, 848, 749, M1100, MS4, MS4R, MS4RS, ST4, SC, ST3, ST2, SS1000
HYM 1100, MTS 1100,MTS 1000, MS2R 1000, M696, MS2R 800, M620, M400, M695, SS800, MTS 620

The bulletin also specifies that used belts should be adjusted to 90-100 Hz (+/- 5Hz). It also implies that used belts can be left alone if they measure above 70Hz.

Ducati does not specify a standard belt measuring temperature. I've measured the same belt when the engine was 54F and also after the engine had warmed to 70F and noted the belt had tightened by 7-8% Hz. This is due to the engine expanding slightly which increases the distance between the drive and cam pulleys, which tightens the belt. Presumably, the Ducati specs are for a cold engine at normal room temperature. If my engine is extra cold, I make sure I am not adjusting to the tight end of the specs.

While the ambient temperature can have a small effect, the position of the valves can have a very large effect on the belt tension measurement. Ducati specifies that belt tension should be measured when the valves of that cylinder are at rest. This occurs at TDC of the compression stroke as well as a good amount on either side of TDC. The reason for this is the closer springs have enough energy to drastically alter the tension readings. The following audio recording illustrates this effect. I locked down the belt adjuster and recorded the belt being plucked twice at eight different engine rotations, each recording at 20-degree intervals of crank rotation from the previous.

Click here to listen to .wav file

The following chart is a graphical representation of the audio file above. The chart is more complete because it represents the full 720-degree cycle at 10-degree intervals while the audio file only represents 8 different crankshaft positions at 20-degree intervals. The chart begins and ends at 0 degrees TDC on the compression stroke of the vertical cylinder which is the proper place to actually measure belt tension. Because a four-stroke engine takes two revolutions of the crank to complete one complete rotation of the valve train, the chart requires 720 degrees of crank rotation to illustrate one complete rotation of the valve train.

To illustrate how not to do a belt adjustment, I adjusted the belt to the loose end of spec (105Hz) at TDC of the exhaust stroke for these readings (point 36 on the chart). The proper place to measure belt tension is at (or near) TDC compression which is the far left of the chart.

The horizontal black line and lighter gray stripe across the center of the chart depict the Ducati recommended tension for new belts and the acceptable range of deviation (110 Hz +/-5Hz). The peaks and valleys of the line connecting the data points are the result of the force of the closer springs trying acting upon the cam pulleys. As you can see, the closer springs have a very significant effect on belt tension if the engine is not properly positioned for tensioning the belt of each cylinder. If you look to the far left and right sides of the chart you will see the flatter area around 145Hz-150Hz where the valves are at rest. That is way out of spec because the belt was adjusted to the loose end of spec at TDC of the EXHAUST stroke. If anyone tries to tell you the closer springs are too weak to significantly affect the proper setting of belt tension, do not fall for it! There is a very good reason why Ducati specifies this operation should be done with the valves at rest and the chart above proves it!

Nobody wants to hear the sound of an engine munching itself to pieces but this is what happens all too often when a belt shreds to pieces from being too tight or jumps a few teeth due to being too loose. And since belt tension affects cam timing to some degree, one tight belt and one loose belt can result in an engine that does not run smoothly at all RPM's.

To get consistent results it is necessary to measure the belt tension when the closer springs are not adding (or subtracting) tension to the belt. Modern belts may be extremely durable but they still fail from time to time. And I would be willing to bet the vast majority of cam belt failures are due, in whole or in part, to improper tension by a mechanic who didn't take the time to do the job right.

Which run of belt should the measurement be taken from? I have seen images from Ducati (one for a three valve engine and one for a four valve engine) and both images point to the two long runs of belt on the outside of the "V" formed by the cylinders. On the ST4s this would be between the layshaft that drives the valvetrain and the fixed pulley for the vertical cylinder and the layshaft and the adjustable pulley on the horizontal cylinder.

Accesses: