The motivation behind this tool is to be able to identify when during the rotational cycle (2 rotations of the crankshaft) an event occurs. Working with a customer concerned with the identification of subjectively objectionable disturbances (e.g. piston slap, valve tick,… etc) in internal combustion engines, a requirement arose for a tool which allows the engineer to pin point where during the engine cycle the disturbance occurs. The customer had neither the time or budget to buy the more specialised tools available from Prosig for what was a short term requirement. So an application was developed using only the basic processing tools available. The following documents how this application is implemented.
Typically vibration signals are measured on the engine during a steady state operation. However, if the problem exhibits itself as significantly loud (greater than 10 dB above background noise of running engine) acoustic measurements can also be used. It is quite important that the measurements are made while the engine is running at a very controlled speed (constant rpm). Typically, these measurements are made while the engine is running under load to minimize rpm fluctuations.
Along with the vibration and/or acoustic signals, there must be a way to establish the position in the engine rotational cycle when the disturbance occurs. Measuring the rotational position on the engine crankshaft is generally not specific enough as the crankshaft goes through 2 rotations for each cycle of the engine. Said another way, if the disturbance is due to the piston or valve train at a specific engine crankshaft position, this might be due to an anomaly on either of 2 cylinders of the engine (assuming multi-cylinder 4 stroke engine). Therefore, it is important to use a signal from the valve train and/or be able to identify the actual firing top dead center (TDC) of a specific cylinder so you know where in the rotational cycle the engine actually is when the disturbance occurs. Engine firing on a specific cylinder might be utilized, however this varies as the engine ignition timing changes. Taking a rotational position off the valve cam lends itself to identifying where in the engine cycle a disturbance occurs. However, with the advances in engine technology to improve engine efficiency, variable valve timing is becoming more common and this is becoming a less than optimal solution. This deficiency may be overcome by disabling the variable valve timing on the engine for the test.
II. Measuring the position in the engine cycle
It is becoming more common for engine manufactures to use unequally spaced pulses/revolution on the engine shaft sensors. These may be ‘multiple pulses – n’ (n missing pulses) during the rotation, or something more elaborate such as 4 equally spaced pulses with 3 pulses interspersed giving a total of 7 pulses/revolution of the cam or crankshafts. Two examples of these are shown in figures 1 & 2 below.
Inspection of the cam signal reveals there is noise on the rising edges of these pulses and it becomes obvious negative going zero crossings are the better choice. The idea is to know where in the engine cycle one of the pulses has a negative going zero crossing.
III. Processing of the tachometer signal data
There is an analysis selection in the DATS software called TACHONEW which uses the analysis module TACHOCROSSING. This very effectively allows the user to take a signal with any number of pulses per revolution and create a new tachometer signal with one pulse per revolution. The setup parameters for this module allows the user to choose which pulse to place the single pulse. The only issue is to indicate to the module where during its rotational cycle to begin counting. This is readily addressed by using the workzone facility in the software. This allows a user to easily select a section of a signal to work on. So we set the starting point of the workzone just prior to one of the 7 pulse repetitive sequence. If the desired position is on the negative going zero crossing, this can be accommodated by simply multiplying the signal by -1 to invert the signal. Now the negatively going pulse becomes a positive going pulse and the TACHONEW analysis module easily handles it. The noisiness now occurs on the negative going slope of the curve and the module triggers on the positive going zero crossing of the modified signal eliminating the noise issue.
One of the parameters in this module is to specify which pulse to start counting on. This allows the user to specify which pulse in the repetitive pulse train to place the single pulse.
IV. Time domain to rotational/cyclical domain
We now have a single pulse/revolution of the engine camshaft (ignoring variable valve timing) which identifies relatively when in the engine cycle the disturbance occurs. Remembering this data is still in the time domain, it makes sense to be able to observe what is happening in the rotational domain. This is to say we now know when something happens in time, but where in the engine cycle does it happen?
The key to this is to be able to resample the data synchronously specifying a number of equally spaced samples for each engine cycle. (NOTE: I have stopped talking about rotational position and now use the term of where in the cycle). If, in the time domain, all the signals are sampled at a very high frequency (suggested 32768 Hz or higher), for a constant engine rpm of 600 RPM (10 revolutions/second = 5 engine cycles/second), there are 5 engine cycles per second or 6553 actual data samples per every engine cycle. This is the maximum sampling resolution for the originally selected sampling frequency. Setting it any higher does not provide any additional information. Resampling all the data channels (including the tachometer signals) at 6553 samples/cycle now gives us the ability to pinpoint where during the engine cycle a disturbance occurs.
Now it is an easy matter to post process this re-sampled data using the trigger of the 1 pulse/engine cycle and capture 1 engine cycle. This is done with the TRIGEXT module to extract events and each event is 1 engine cycle long. The sample data captured actually was just shy of 40 engine cycles. The cycles are extracted for the desired data channel and plotted in a 3 dimensional plot reminiscent of a waterfall display (Figure 3 below).
From this data it is seen there is a predominant event which occurs during each engine cycle at approximately 260 degrees into the engine cycle from the selected tachometer mark point determined in section III above.
Application of this tool has proven to be a valuable for 4 stroke internal combustion engine diagnostics. Other applications might be diagnostics of electric motors or any other machinery which has a repetitive/cyclic nature associated with its operation. Other potential applications might include multi-pole electrical motors, generators, and pumps of various designs.
A DATS worksheet has been created to perform this analysis allowing the user to enter the pertinent analysis parameters when running the worksheet. This worksheet will be made available on the Prosig Support Blog in the near future.