1. A LVDT (Linear Variable Differential Transformer) probe. This is basically a contact probe. This will allow the measurement of the shaft total runout*.
2. If an electrical runout measurement is also required then an additional eddy current probe (also known as a proximity probe) is needed. This is a non contact probe and can be used in conjunction with the LVDT to measure the electrical runout only.
3. A once per revolution tachometer is also required. This could be optical, proximity, magnetic and so on.

With these three sensors it is possible using the DATS Rotor Runout Measurement package to find both electrical and mechanical runout.

* Total runout is a composite tolerance including the effects of cylindricity and concentricity, co-axiality, straightness and parallelism along the axis.

1. Response Analysis

The most straightforward solution is to simply attach an accelerometer (keeping in mind any temperature restrictions) to the part in question. Then run the engine at a steady speed and view an FFT spectrum of the of the vibration.

There are a number of issues with this technique. Most importantly, there is no way to separate which vibrations are resonances of the part of interest and which are caused by resonances of other parts, such as the block or head. Also, there will be a significant content caused by the running of the engine itself. All these sources will be mixed together and impossible to separate. For these reasons, this technique is not recommended. So how could the structural resonances and the vibrations caused by running the engine be separated?

2. Rotational Speed Analysis

First, we attach an accelerometer to the inlet manifold (with the part fitted to the engine) and then measure the engine crankshaft speed using a tachometer sensor. Then we measure both the vibration levels and shaft speed of the crankshaft while we vary the speed of the engine. Using this data we can create a waterfall plot. This plot will show us frequencies associated with the running of the engine. These include the various rotating parts and the ignition phase of the combustion process. More importantly for our test this waterfall will also show the structural resonances.

Although this technique will allow an engineer to find the resonances and vibrations caused by the engine running, it will not allow the separation of the natural resonances of the inlet manifold from the natural frequencies of the other parts of the engine. For that, another technique is required.

3. Hammer Impact Analysis

The most common and generally the most successful test for this sort of project is the Hammer Impact Test. This is often (incorrectly) called Modal Analysis or a Modal Test. See the article What Is Hammer Impact Testing? for more information.

A Hammer Test consists of attaching an accelerometer to a location on the structure or component. In our case this would be the inlet manifold. Importantly the manifold should not be attached to anything. In an idea world it would be in a totally free-free state, floating in space. In practice we suspend the test piece. With the accelerometer fixed in one position we then impact the force instrumented hammer in another position. For example, we might choose to fit the accelerometer at one end of the inlet manifold and impact the hammer at the other.

The hammer has a force sensor fitted to the head. The data acquisition system can then measure the excitation force from the hammer and measure the response acceleration from the accelerometer. Using this information a transfer function is calculated between these two points. We can then see how an excitation will cause certain frequencies to be excited in the structure. The transfer function will show the natural frequencies of the structure. The more complex the structure the more complex the transfer function.

Importantly the engineer will be able to clearly visualise the natural response of only the part in question.

I want to perform some cylinder head and inlet manifold vibration analysis, what should I do?

First we need to consider sensor selection

Amplitude Range

A 100g accelerometer is probably a bit low for a cylinder head, I have often seen values up to 400g, but it depends on the engine in question.

I’d recommend a higher acceleration range, perhaps 500g on any accelerometer selected. It is however not a simple choice. Too high in the available amplitude range and you’ll loose low amplitude resolution. So the best advice is to try some experiments to confirm the best range.

Temperature Range

Temperature will be an issue, especially on the inlet manifold. Most accelerometers will only work up to 150°C (IEPE type). Above that and they will be damaged. Higher temperature options (charge type) are available. It depends on the requirements.

Frequency Range

First, decide the frequency range that you want to examine. This is called the measurement bandwidth. For example, if you want to study up to 1kHz then you need 1kHz of measurement bandwidth.

The Nyquist rate is 2 times that value. I would recommend using 2.5 times the measurement bandwidth.

Using a factor of 2.5, a 2kHz sample rate will give you 800Hz frequency range.

So for a 1kHz bandwidth you’d need 2.5kHz sample rate or 2500 samples per channel per second.

In this case, at the 1kHz cut off point, the anti-alias filters will have already attenuated the signal by 3dB  (that’s called the 3dB point).

So,  therefore, you need to raise the sample rate a little. I’d recommend 4000 or 5000 samples per second per channel for vibration. You can always downsample later in software if you need better low frequency resolution for your frequency analysis.

Data Capture & Measurement - What can you actually get?

When working with engines you’ll need a tachometer input. This will normally be measured on the crankshaft or camshaft. A once per revolution sensor is usually sufficient. A laser sensor is the best but there are many other types available, like infra red or magnetic. You might even be able to use an electronic signal from the coil/electronics.

When you have a vibration signal and a tachometer signal it allows you to measure the vibration levels at different engine speeds. This allows you to see which rotating components are causing the vibration. For example, when the crank shaft rotates once, the cam shafts will rotate twice. Therefore, a vibration that occurs twice per crankshaft revolution is likely to be caused by something to do with cam shafts. The same can be said of various systems in the engine, like pumps or timing gears and so on.

Summary

To sum up we need to

• select an accelerometer with the correct amplitude range
• select a sensor that will work at the required temperature
• choose the frequency range we wish to analyze and thus choose a sample rate to measure at
• configure a reliable tachometer signal to measure the rotational speed of the engine

The value of 10,000 Hz is multiplied by 2.5 to allow for an anti-alias filter during the capture of the data. An anti-alias filter is set to 0.4 of the sample rate, thus the bandwidth or frequency content that can be studied is 0.4 of the sample rate.

For example, when looking to study a frequency up to 10,000 Hz what sample rate should be used?

So we multiply by 2.5…
10,000 Hz x 2.5 = 25,000 Hz

So the sample rate should be 25,000 samples per second to allow frequencies of up to 10,000 Hz to be studied.

Accelerometer calibration screen

Go to Single Channel Calibration screen.

Click on the Tone tab.

With P5000 armed turn on the shaker and monitor the sine wave on the real-time monitor and check Signal Quality as being GOOD.

With a GOOD sine wave click the Calculate button. It is recommended to click Calculate three times.

Check to see the Sensitivity change from what was originally entered when setting up the channel transducer information to the new calculated value. The calculated value should be close to the original.

DC Cal Offset is for DC level accelerometers.

Requested Excitation is for non IEPE (ICP) accelerometers such as capacitive.

After calculating the new sensitivity click on the Use button to make the change in the Transducer Sensitivity Acquisition Setup file.

NOTE: To achieve a GOOD signal reading place the shaker on a flat surface and avoid touching during operation. Make sure the surface the shaker is sitting on does not come in contact with other sources of vibrations or electrical conductance.