/public/MSDII/BuildTestDocument/

Build, Test, Document

Table of Contents 1 Design 1.1 PCB Layout 1.2 Revision 2.0 PCB Layout 1.3 DAQ Configurator 1.3.1 GUI Mock-up 1.3.2 Configuration File Format 2 Testing 2.1 MSR145 Tests and Benchmarking 2.1.1 MSR145 Software and GUI 2.1.2 MSR145 Temperature Testing 2.1.3 MSR145 Analog Inputs 2.2 Vibrations Testing 2.2.1 Simulation 2.2.2 Test & Results 2.3 Simulation Results 2.3.1 Attenuation Results 2.3.2 8th Order Filter Simulation Results 3 Build and Integrate 3.1 Attenuation Results 3.2 8th Order Filter Results 3.3 PGA Results 3.3.1 PGA Moving Forward 4 Navigation Links

Design

PCB Layout

The following circuit was built in PCB Artist. This circuit was used to make all of the connections of all of the components that were used in this design. This software will automatically convert this schematic to a PCB layout and maintain all of the connections to ensure the PCB layout will be correct.

Circuit Schematic for the PCB Layout

Top layer of the PCB Layout Bottom Layer of the PCB Layout

The previous two images show the PCB layout after all of the components were placed in their desired location and all of the copper traces were made. The PCB layout was changed to remove all of the 90 degree turns in the layout. The electrons will damage the copper traces turning around 90 corners because they are moving at high speeds. The 90 degree turns were replaced with straight traces and 45 degree turns. Also copper was poured in the blank areas of the board. This copper area will be added to the ground nets and the power nets.

Revision 2.0 PCB Layout

Bottom Layer of the PCB Layout Bottom Layer of the PCB Layout

This version of the PCB features a complete redesign in order to reflect the decision to use the FRDM-KL25Z development board as a component for the data acquisition system. As such, the new design is larger and feature pin headers for connecting to the FRDM-KL25Z, and no longer has a microprocessor on board.

In addition, several changes were necessary after reviewing the Revision 1.0 PCB:

• Lack of power switch
• Addition of JST connector for battery

The PCB that was created was not correct. The top half of the component is mirrored. This happened because of a misinterpretation of the data sheet. The data sheet had three drawings of the SD card socket, one looking from the top, the side and one from the bottom. The PCB layout that is on the datasheet is looking from the backside of the socket. The layout that was drawn was drawn in reverse.

There is also another problem with this board. The pins that will connect to the development board do not line up correctly. The outside part of the board has to sets of pins that would connect to the development board. These pins have an upper and lower section and there is a space between them and this was not accounted for in the board layout.

These problems were fixed when we sat down as a team to review the schematic before sending off the final board to be fabricated.

DAQ Configurator

GUI Mock-up

Configuration File Format

The configuration file for the DAQ, requires the storage of the following information:

• File prefix for results files
• Sampling Rate
• Sampling Accuracy
• Channels Enabled
• For each channel:
  • Channel Name
  • Gain

Example Configuration File:

{
  "prefix" : "Compressor Test 5-15-14",
  "rate" : "5000",
  "accuracy" : 12,
  "channels" : {
    "1" : {
      "name" : "Manifold Temperature",
      "gain" : "30"
    },
    "2" : {
      "name" : "X-axis Vibration",
      "gain" : "0.2"
    }
  }
}

Testing

Testing is an essential part of any design process. As a product comes closer to realization, its functionality needs to be tested in order to show that the product works as intended. This project has a multitude of tests that need to be carried out before it can be considered a working product. Most of these tests are circuit and software based, to see if the device can successfully record and store data. Mechanically, the device needs to be tested to see if it can withstand the harsh conditions within the compressor.

Not only does our device need to be tested for functionality, but it also needs to be tested for accuracy. There is no point in having a device that can store data if the data it collects is wrong. We will test our device against existing DAQs and compare results. The main two systems we will be working with are the MSR145 and the already installed NI DAQ system currently being used on the compressor.

A current, WIP version of the test plan used can be found here

MSR145 Tests and Benchmarking

The MSR145 that was purchased is an unfamiliar device, so its own capabilities need to be tested before we can benchmark it against our device. The MSR145 comes with an internal 3-axis accelerometer, temperature, and pressure sensors. These sensors need to be tested for accuracy. The datasheet for the MSR145 is available here

From a datalogging POV, the supplied software for the device will be installed, and the GUI will be tested for capabilities.

MSR145 Software and GUI

Since no data recording is completely necessary to test capabilities, the software and GUI were the first things that were experimented with.

There is a lot of functionality in the software, including the ability to change sample rate, select which inputs are recording, and even do calculations with any collected data. The customer has expressed that a lot of these features are nice and useful, but they aren't necessary for our own product. A simple text file to upload to the internal memory that contains all basic set-up options would be perfectly fine.

Main GUI for MSR145

Setup GUI for MSR145

Viewer GUI for MSR145

MSR145 Temperature Testing

The MSR145 internal temperature sensor was pitted against an NI temperature datalogger that was readily available. The idea of the test is to see how well the temperature sensor can accurately read the ambient temperature.

The test plan is shown below to the left, and the results of the test are shown below to the right. The data collected was for both DAQs and can be found here (NI) and here (MSR145).

Temperature Test Plan for MSR145

Temperature Test Results for MSR145

This test shows that the internal temperature sensor of the MSR145 is not as accurate as the NI Datalogger. The initial temperature difference of about 4°C is a pretty signicant difference between readings. It is unknown from this test which reading was more accurate, since these are the only two temperature readings of that room at the time available. However, after the move to outside, you can clearly see where the NI DAQ has an edge. It was able to obtain a much faster drop in readings than the MSR145 was, and even reached a steady state. The MSR145 wasn't ever able to find this state as was still decreasing when brought back inside. The same trend can be seen with the increasing temperature, where the NI DAQ was able to much more quickly able to find a steady temperature. From this test we can conclude that while the MSR145 does collect temperature data, it may not be accurate and is also more resistant to big changes in temperature.

MSR145 Analog Inputs

The inputs for the MSR145 were tested against the NI DAQ system in the compressor room. This test shows the accuracy and limitations of the analog inputs MSR145.

The test will be run by connecting both DAQs to a function generator simultaniously. Three different scenarios will be created: A 1Hz sinusoidal wave input, 5Hz, 15.5Hz, 25Hz, and 25.6Hz. Input range will be initially set to go from 0V to +3V. This range will be used for the second analog input of the MSR145, A2, which has an allowable 0V-5V input range. The last two tests will have an input range of 0V-9V to test the third and fourth inputs (A3,A4) of the MSR145, which have an allowable input range of 0V-10V. Sampling rate for the MSR145 is limited to a maximum approximately 50S/s, so that is the rate chosen. The NI DAQ has much higher capabilities, but for a fair comparison the sample rate will be limited to 50S/s for the first two tests. To show the impact upon accuracy sampling rate can have, the last test will have the sampling rate set to 100S/s. Data will be recorder by each DAQ, and compared against each other.

Tests were run, and the results and analysis can be found in the below link:

MSR145 Analog Testing

Vibrations Testing

Simulation

Vibration Stand Finite Element Analysis

Results Table

In order to gain empirical data to validate our theoretical predictions, a Brüel & Kjær vibration test rig was provided by Bill Nowak from Xerox. Accompanying the test rig is a SigLab DAQ system that will generate a transfer function using the input from accelerometers.

In order to qualify how reliable the data from the test rig will be, a Finite Element Analysis was ran on the test stand, to predict it's resonant frequency(s). From this analysis it was determined that the most significant resonant frequency occurs at 1929.2 Hz. The results are below:

Test & Results

The test plan for the case and PCB system was formulated and shown below to the left. It involved mounting the device Mr. Nowak's shaker table. Since the flange mounts on the case were not the same distance apart as the previous team's device, two options are available: either drill new holes into the compressor or create a mounting adapter bracket. The customer preferred a mounting adapter. The plate was created, and the vibration rig was modified to accommodate the plate. The device was mounted to the shaker and two accelerometers were attached; one to capture the input frequencies and one to capture the output response. A picture of the setup can be seen below to the right.

Vibration Test Plan

Vibration Test Setup

The assumptions used during this test included that the PCB inside the enclosure would display the same vibrational effects as the enclosure. This way the enclosure could be tested without severe modification to access the PCB inside. The shaker was set to input a 'chirp' to the system, then rerun with a 'random' input. Each input was teseted with and without the ruber endcaps to see the difference between the two situations. The resultant transfer functions can be seen below. What this data shows is that the enclosure sees the same vibrations as the input to the system up to about 400 HZ. This means that there should not be any interference of the vibrational modes with the data collection for the low compressor frequency of 12 Hz.

Chirp Input with Rubber Endcaps	Chirp Input without Rubber Endcaps
Random Input with Rubber Endcaps	Random Input without Rubber Endcaps

Simulation Results

Attenuation Results

Circuit Built in LTSpice to simulate the Programmable Gain Amplifier

Simulation Results

In the above diagram, the the green curve represents the input of a 0-10V sine wave and the blue wave shows the output of a 0-3.3V sine wave. The blue curve represents the analog signal that will go into the ADC in the micro-controller. This simulation was done using precision op amps. This test was repeated with different input signals. The attenuation needed to be changed for the different signal and this is done by changing the resistance values of the resistors. When the programmable gain amplifier is used, the gain or attenuation is set in software not by changing resistor values. This allows us for the gain or attenuation amount to be changed without having to make any hardware changes.

8th Order Filter Simulation Results

The diagram above shows the results from the simulation using the MAXIM spice program on their website. The simulation shows that the MAX7403 chip has a sharp cutoff at 10kHz. This simulation was repeated for various input signals and the results were always the same.

Simulation Results of the input and output of the 8th order Filter MAX7403 for a 1khz signal

Simulation Results of the input and output of the 8th order Filter MAX7403 for a 5khz input signal

Build and Integrate

Attenuation Results

The above image shows the input and output of the attenuation circuit. The yellow curve is the input signal. The input signal was a 5V P-P sine wave with a 5V DC offset. The resulting curve is a sine wave going from 0V to 10V, centered around 5V. The blue curve is the output is a sine wave that goes from 0-3.3V. The output does not go exactly to 3.3V because of the resistors had 10% tolerances. Another reason for the output not being exactly 3.3V is because of error in the oscilloscope.

8th Order Filter Results

Simulation Results of the 8th order Filter MAX7403

Input and Output of the Filter with Different Input Frequencies

The above left image shows the input and output of the MAX7403 8th order LPF. The input signal was a sine wave with a 3V P-P input with a DC offset of 1.5V at 10kHz. The above right table shows the input and ouput of the filter for different frequencies on the input. The input voltage was always kept the same and the frequency was changed.

Simulation Results of the 8th order Filter MAX7403

Simulation Results of the 8th order Filter MAX7403 with changing frequencies

The above images shows the output of the filter with different frequencies. The input signal was a sine wave with a 3V P-P input with a DC offset of 1.5V at 10kHz. The cutoff frequency was set to 10kHz. As the input frequency increases less of the signal passes through. The output signals with frequencies higher than 10kHz are centered around 1.5V because of the DC offset. This demonstrates that the filter is an effective low pass filter.

PGA Results

The figure below shows the results of passing a signal through this system. These are all curves that were measured via an oscilloscope. The first curve on the left was the input to this system. The signal was a 2 Vp-p signal at a frequency of 1kHz. The middle curve is the output of the PGA. This signal is still 2Vp-p at 1 kHz but it is offset by 1.5V. This PGA has a built in offset and the offset is dependent on the output signal. The last curve is the output of the low pass filter and the input to the ADC on the microcontroller. This curve is also a 2Vp-p signal at a frequency of 1kHz. This demonstrates that the system can successfully pass through a signal from the input channel to the ADC on the microcontroller. These tests were repeated on both channels and at different voltages and frequencies. From this it was learned that the cutoff frequency on the filter works properly and can be adjusted. Also through testing it was discovered that the maximum input to the PGA is half of the VDD, which in this cause is 2.5V. This was overlooked on the datasheet when picking the part. 7

PGA Moving Forward

P14452:/public\MSDII\BuildTestDocument\PGA replacement options.docx

---

Navigation Links

MSD I	MSD II	Overview
Planning & Execution Problem Definition Systems Design Detailed Design	Planning & Execution Build, Test, Document Final Presentation Technical Paper Poster	Home Photo Gallery Project Review