Worked on getting rejection ratio with the 4000A-002 accel. I had to get the accel dynamics before the RR. When I did this, the transfer function showed a "blip" at around 95Hz. This is a problem because when I then went to take the rejection ratio (after having replaced the necessary electrical components), the RR was very bad at 95 Hz, where this blip occurs.
At around 2:00pm Leslie gave me the OK to send the Vixi analyzer to White Sands. I spent most of the rest of the day shipping this, then moving all my instrumentation over to the Ethyl analyzer to continue the RR tests.
I also shipped out the Kaman eddy current probe frequency demodulator box to get repaired.
Friday, October 31, 2008
Wednesday, October 29, 2008
Wednesday, October 29, 2008
Called Will Meenan at Kaman and talked about what to do with the broken KD2300. He said the only thing now is to just send it back for repair, which will cost $500.
Discussed the RR results with Brad and Paul J. Now that the 4000A accels are back we will redo the RR with those accels. First the circuit needs to modified to account for their dynamics.
I also conducted the DC temperature test on all 4 measurement specialties accelerometers. This test was done to see the deviation of sensitivity at different temperatures.
While doing the temp test I improved the GUI I made to plot DCS data. It works pretty well now, however I'm sure there are still a few kinks to work out.
Discussed the RR results with Brad and Paul J. Now that the 4000A accels are back we will redo the RR with those accels. First the circuit needs to modified to account for their dynamics.
I also conducted the DC temperature test on all 4 measurement specialties accelerometers. This test was done to see the deviation of sensitivity at different temperatures.
While doing the temp test I improved the GUI I made to plot DCS data. It works pretty well now, however I'm sure there are still a few kinks to work out.
Tuesday, October 28, 2008
Tuesday, Oct 28, 2008
Showed Brad results of rejection ratio test when tuned at 65Hz. He thought the RR was too high at the higher frequencies, and suggested I use an oscilloscope that can compute FFTs to tune the circuit. So, I input the signal in the oscilloscope and set it up to do an FFT, and put a cursor at 65Hz. I then adjusted the circuit until the amplitude of the FFT at 65Hz was at a minimum. I then retook the RR and got a great looking curve. It wasn't as good at low frequencies as the curves I had gotten before, but it was darn good at the higher frequencies.
The 4000A-002 accels arrived back from repairs today. I took a noise floor test with them and got a noise floor of 430uV. The spec says 500uV so these are working great now.
I set up a little for the temperature test on the 4000As and 4002M1s.
Talked to Will Meenan about the Kaman probes. He found out that the Kaman probes with CSAIDs 2001 and 2002 came calibrated from the factory with a +/-5V span, which is not how they typically are made. Usually they only have a +/-2.5V span. This interesting tidbit aside, he told me to try swapping the broken capacitor for the one that is in the demo electronics he sent. I did this and it still does not work.
The 4000A-002 accels arrived back from repairs today. I took a noise floor test with them and got a noise floor of 430uV. The spec says 500uV so these are working great now.
I set up a little for the temperature test on the 4000As and 4002M1s.
Talked to Will Meenan about the Kaman probes. He found out that the Kaman probes with CSAIDs 2001 and 2002 came calibrated from the factory with a +/-5V span, which is not how they typically are made. Usually they only have a +/-2.5V span. This interesting tidbit aside, he told me to try swapping the broken capacitor for the one that is in the demo electronics he sent. I did this and it still does not work.
Monday, October 27, 2008
Monday, October 27, 2008
Learned a little more about the LK-G32. It can be set up for spectral reflection or diffuse reflection. Spectral reflection is the perfect, mirror like reflection of light, where the angle the light hits the surface at is the angle it bounces off at. For Diffuse reflection, the light bounces off in a range of directions.
For diffuse reflection, the laser beam is perpendicular to the surface being measured. For specular reflection, you tilt the laser head so the beam is coming in at an angle. You also have to change the setting in the laser's controller to "mirror reflection".
Picked up some electronics for Brad from Jameco.
Dropped off car at Independence Acura to get fixed.
Worked a little bit on the frame for the Fast GVT. Ordered some pipe clamps that will hopefully work for 1" dia pipe.
Found out that the way we are calculating rejection ratio might not be correct since we scale the differenced output from the circuit.
For diffuse reflection, the laser beam is perpendicular to the surface being measured. For specular reflection, you tilt the laser head so the beam is coming in at an angle. You also have to change the setting in the laser's controller to "mirror reflection".
Picked up some electronics for Brad from Jameco.
Dropped off car at Independence Acura to get fixed.
Worked a little bit on the frame for the Fast GVT. Ordered some pipe clamps that will hopefully work for 1" dia pipe.
Found out that the way we are calculating rejection ratio might not be correct since we scale the differenced output from the circuit.
Thursday, October 23, 2008
Made slides for GVT PDR.
Replaced broken capacitor in Kaman box. Unit still does not work. Learned that the capacitor that broke was as Tantalum capacitor rather than a ceramic capacitor. Tantalum is a highly corrosion resistant element. The Tantalum capacitor that broke is also polarized, meaning it does matter which way you put it in.
Tried gain tuning inertial referencing circuit at 65Hz. The unit is very very finicky. It is hard to get it right. Also, there is some problems overlaying different frequency ranges to get a smooth rejection ratio plot. I don't know why the rejection ratio should be so different for different frequency ranges.
Replaced broken capacitor in Kaman box. Unit still does not work. Learned that the capacitor that broke was as Tantalum capacitor rather than a ceramic capacitor. Tantalum is a highly corrosion resistant element. The Tantalum capacitor that broke is also polarized, meaning it does matter which way you put it in.
Tried gain tuning inertial referencing circuit at 65Hz. The unit is very very finicky. It is hard to get it right. Also, there is some problems overlaying different frequency ranges to get a smooth rejection ratio plot. I don't know why the rejection ratio should be so different for different frequency ranges.
Wednesday, October 22, 2008
Wednesday, October 22, 2008
Found out today some more about the inertial referencing circuit. Paul told me that when he designed it, he did not design each part of the circuit (the disp part and the accel part) to have a gain of 2 in/s/V. He uses the gain of the last part of the circuit(which differences the signals) to do a final gain adjustment and bring it to 2in/s/V.
Also, I was having some pains doing the gain tuning. I couldn't match the gain well in order to give me a good RR. What I ended up doing was using a very small sample frequency. This cleaned up the differenced signal well enough so that I could easily tell when it was 90 degrees out of phase, which signifies that it is perfectly gain matched.
Also, I was having some pains doing the gain tuning. I couldn't match the gain well in order to give me a good RR. What I ended up doing was using a very small sample frequency. This cleaned up the differenced signal well enough so that I could easily tell when it was 90 degrees out of phase, which signifies that it is perfectly gain matched.
Tuesday, October 21, 2008
Tuesday, Oct 21, 2008
Learned a lot about the inertial referencing circuit yesterday and today. The 4th order filter used on the displacement part is to compensate for the phase change seen by the accelerometer.
Phase change of the accelerometer explained:
Even though the natural frequency of the accel is much higher than our analysis band, the phase change associated with the natural frequency is substantial enough at our analysis band that it must be accounted for. To do this, we referenced the accel against one with a much much higher natural frequency. This gave us a transfer function for the accel(FRF between circuit accel and reference accel), which you can fit a model to in Matlab using invfreqs. Paul then does some magic and creates a filter that will mimic this behavior for the displacement sensor.
Also, one of the parts of the circuit for the accel is to account for the laser sensors measurement delay. We had to switch one of the resistors in this section because the laser sensor we are using now is different then before.
Once all the circuitry was changed to account for the new sensors, I ran another RR test and got a good RR for the entire analysis band. Brad was pleased but wanted me to gain tune the circuit at 65Hz to see if we could get the high frequency part better. Gain tuning involves changing the variable resister to try and match the gains of the two velocities from the sensors. If done correctly, the gains will be identical and the only thing that will cause problems is the phase difference between the two signals.
The rest of the day I spent preparing the instrumentation to get shipped to White Sands for Leslie's test.
Phase change of the accelerometer explained:
Even though the natural frequency of the accel is much higher than our analysis band, the phase change associated with the natural frequency is substantial enough at our analysis band that it must be accounted for. To do this, we referenced the accel against one with a much much higher natural frequency. This gave us a transfer function for the accel(FRF between circuit accel and reference accel), which you can fit a model to in Matlab using invfreqs. Paul then does some magic and creates a filter that will mimic this behavior for the displacement sensor.
Also, one of the parts of the circuit for the accel is to account for the laser sensors measurement delay. We had to switch one of the resistors in this section because the laser sensor we are using now is different then before.
Once all the circuitry was changed to account for the new sensors, I ran another RR test and got a good RR for the entire analysis band. Brad was pleased but wanted me to gain tune the circuit at 65Hz to see if we could get the high frequency part better. Gain tuning involves changing the variable resister to try and match the gains of the two velocities from the sensors. If done correctly, the gains will be identical and the only thing that will cause problems is the phase difference between the two signals.
The rest of the day I spent preparing the instrumentation to get shipped to White Sands for Leslie's test.
Thursday, October 16, 2008
Thursday, Oct 16, 2008
Still sick but no sore throat. Came in to work.
Finished up repairing and cleaning long BNC cables. Had John Harmon help me put them away.
Called Jim Gooding about sending Leslie Chrezan accelerometers for her test in White Sands, NM. He said he should be able to send them to her.
Performed a cal check on one of the VTI1436 cards. You have tell IDeas to do a calibration or it wont be in cal. I ran a precision 5V signal to 8 of the channels. Before I performed the cal, channels were off by as much as 1.5%. Afterwords, they were off only by .16%.
Talked to Brad about next step for the inertial referencing system for the Fast GVT. He gave me a list of things to do. He pointed me to a very good reference book for electrics: "The Art of Electronics", by Horowitz and Hill.
I put the thermocouple cables from the Yuma M777 shock test in 2 plastic bins in storage.
Calculated what I need the resistance op amp in the displacement arm of the circuit in order for us to be able to use the full sensitivity range of the Keyence sensor.
Finished up repairing and cleaning long BNC cables. Had John Harmon help me put them away.
Called Jim Gooding about sending Leslie Chrezan accelerometers for her test in White Sands, NM. He said he should be able to send them to her.
Performed a cal check on one of the VTI1436 cards. You have tell IDeas to do a calibration or it wont be in cal. I ran a precision 5V signal to 8 of the channels. Before I performed the cal, channels were off by as much as 1.5%. Afterwords, they were off only by .16%.
Talked to Brad about next step for the inertial referencing system for the Fast GVT. He gave me a list of things to do. He pointed me to a very good reference book for electrics: "The Art of Electronics", by Horowitz and Hill.
I put the thermocouple cables from the Yuma M777 shock test in 2 plastic bins in storage.
Calculated what I need the resistance op amp in the displacement arm of the circuit in order for us to be able to use the full sensitivity range of the Keyence sensor.
Wednesday, October 15, 2008
Wednesday, Pctober 15, 2008
Woke up with a sore throat and feeling a little off. Came into work anyway.
Worked on removing tape and putting new BNC connectors onto 250' cables from Yuma. Will try to finish them tomorrow.
Started gathering test equipment for Leslie for her test in White Sands, NM. We don't have all the accels she needs so I will have to call the Albuquerque office tomorrow to see if they can help.
Did a tap test for Raman for his WISE job. Had Brad OK my work.
Wrapped up the Thermocouple cables from the Yuma test. Brad said to put them in a tub, put our initials, the job number, and date on it, and store it in the other building.
Worked on removing tape and putting new BNC connectors onto 250' cables from Yuma. Will try to finish them tomorrow.
Started gathering test equipment for Leslie for her test in White Sands, NM. We don't have all the accels she needs so I will have to call the Albuquerque office tomorrow to see if they can help.
Did a tap test for Raman for his WISE job. Had Brad OK my work.
Wrapped up the Thermocouple cables from the Yuma test. Brad said to put them in a tub, put our initials, the job number, and date on it, and store it in the other building.
Tuesday, October 14, 2008
Tuesday, Oct14, 2008
Retook noise floor measurements for the inertial referencing circuit. I did this because before I was not powering the circuit properly. I was only supplying 12V when it actually takes +/- 15V. The new results give lower noise floors which is good. No harm done.
I also took another rejection ratio measurement. Still only getting about -16dB attenuation which is pretty bad.
The rest of the day Josh helped me put the 250' cables onto spools (which finally arrived after 4 weeks). This took a lot of time but we were able to get all ~33 cables done. Now they just have to be tested to make sure they are good.
I also took another rejection ratio measurement. Still only getting about -16dB attenuation which is pretty bad.
The rest of the day Josh helped me put the 250' cables onto spools (which finally arrived after 4 weeks). This took a lot of time but we were able to get all ~33 cables done. Now they just have to be tested to make sure they are good.
Monday, October 13, 2008
Monday, October 13, 2008
Jeff found out that the VXI has not been doing its self calibration during start up. Even when he changed the vxi setup file (hpvxi.setup, calibrate_option), it still didnt seem to do it. Therefore,to do it manually, bring up the menus in IDeas (ctrl+m) and find the perform calibration option.
I called Kaman and left a message about the capacitor in the frequency demodulator. I checked the broken capacitor, it is a 1.0uF, 35V capacitor.
I called Jeff Chang at measurement specialties about the status of our 4000A accels.
I worked on the rejection ratio test for Fast GVT. The gains for the circuit (2 in/sec/Volt) are set based on the calibration factors of the sensors. Thus if the sensor's cal is different, the circuit will not perform the same. In my case, both sensors are different than the ones used to tune the gains on the circuit. The cal on the disp sensor can be changed to account for this, however the cal on the accelerometer cannot.
Brad asked me to get John Harmon to organize the BNC cables from Yuma.
I called Kaman and left a message about the capacitor in the frequency demodulator. I checked the broken capacitor, it is a 1.0uF, 35V capacitor.
I called Jeff Chang at measurement specialties about the status of our 4000A accels.
I worked on the rejection ratio test for Fast GVT. The gains for the circuit (2 in/sec/Volt) are set based on the calibration factors of the sensors. Thus if the sensor's cal is different, the circuit will not perform the same. In my case, both sensors are different than the ones used to tune the gains on the circuit. The cal on the disp sensor can be changed to account for this, however the cal on the accelerometer cannot.
Brad asked me to get John Harmon to organize the BNC cables from Yuma.
Thursday, October 9, 2008
Wednesday, October 08, 2008
Performed noise floor tests on inertial referencing circuit. To do this I shorted the displacement input to ground, then took the PSD of the displacement-velocity. I did the same to the accelerometer input and acceleration-velocity.
Once I had the noisefloors of the circuit and the sensors, I overlayed them on a plot. I had a difficult time understanding what calibration factor to apply to the circuit noise floors to get them into in/s. In the end the confusion came from two things. One, I was overcomplicating things…the output of each part of the circuit is in Volts. It doesn’t matter what happened to that voltage along the way, volts are volts. This new voltage was simply proportional to velocity instead of disp/acceleration now. The second thing that was confusing was the gain to apply. I thought it would have more to do with the cal factors of the sensors, but that is not so. I talked this over with Brad and the actual circuitry itself has a gain of 2 (in/s)/V for both accel velocity and disp velocity. I had no way of knowing this beforehand, which was slightly frustrating.
During this day I also edited the read_ati_file.m and read_afu_file.m. Before now, they did not have a field which contained a description of each channel, or have a field which specified the units. I used V=get(*ideas_fn file) to get a list of all the possible fields which can be written to the .mat file. I could not find one for units, so I just had the read files record what units you told it to export in.
Once I had the noisefloors of the circuit and the sensors, I overlayed them on a plot. I had a difficult time understanding what calibration factor to apply to the circuit noise floors to get them into in/s. In the end the confusion came from two things. One, I was overcomplicating things…the output of each part of the circuit is in Volts. It doesn’t matter what happened to that voltage along the way, volts are volts. This new voltage was simply proportional to velocity instead of disp/acceleration now. The second thing that was confusing was the gain to apply. I thought it would have more to do with the cal factors of the sensors, but that is not so. I talked this over with Brad and the actual circuitry itself has a gain of 2 (in/s)/V for both accel velocity and disp velocity. I had no way of knowing this beforehand, which was slightly frustrating.
During this day I also edited the read_ati_file.m and read_afu_file.m. Before now, they did not have a field which contained a description of each channel, or have a field which specified the units. I used V=get(*ideas_fn file) to get a list of all the possible fields which can be written to the .mat file. I could not find one for units, so I just had the read files record what units you told it to export in.
Tuesday, Oct 7, 2008
Moved circuitry of inertial referencing for Fast GVT to a wood plank to get it off of the power unit.
Converted .ati files for Scott P. for World View
Reviewed circuit diagram for inertial referencing circuit and compared against the actual circuit to make sure it was wired up correctly. The only differences I could see between the actual circuit and the diagram were 1) for the accelerometer signal conditioning, pins 1 and 8 are supposed to be connected for three of the stages by a variable resistor, and are not, and 2) there is no SW2 as is shown in upper right of diagram.
Performed noise floor on new Measurement Specialties accels (4002M1-002-060 and 4002M1-005-060). Tried using another, lower noise floor accelerometer to glue back to back with the test accel. Found out that some of the ICP sensors are very temperature sensitive. Just temperature deviations in the lab will cause the signal to wander a little. Also, just holding the accel for a little bit will cause serious drift to the signal. Lesson learned!
I am still unclear about what to do about the two resulting random signals. If you add two random signals, what is the resulting signal? Is it’s RMS value twice that of the other two? If not, why?
I also helped scott plot some old WV tests, and convert some .afu files from a local modes test I did awhile ago.
Converted .ati files for Scott P. for World View
Reviewed circuit diagram for inertial referencing circuit and compared against the actual circuit to make sure it was wired up correctly. The only differences I could see between the actual circuit and the diagram were 1) for the accelerometer signal conditioning, pins 1 and 8 are supposed to be connected for three of the stages by a variable resistor, and are not, and 2) there is no SW2 as is shown in upper right of diagram.
Performed noise floor on new Measurement Specialties accels (4002M1-002-060 and 4002M1-005-060). Tried using another, lower noise floor accelerometer to glue back to back with the test accel. Found out that some of the ICP sensors are very temperature sensitive. Just temperature deviations in the lab will cause the signal to wander a little. Also, just holding the accel for a little bit will cause serious drift to the signal. Lesson learned!
I am still unclear about what to do about the two resulting random signals. If you add two random signals, what is the resulting signal? Is it’s RMS value twice that of the other two? If not, why?
I also helped scott plot some old WV tests, and convert some .afu files from a local modes test I did awhile ago.
Monday Oct 6, 2008
Spent the morning trying to shape a PSD so that we could use it to get calculate the rejection ratio of the inertial referencing systemfor the GVT SBIR over the entire band of 1-100Hz. The set up consists of gluing the test accel onto the shaker, and mounting the Keyence laser sensor so that it points at the accel. That way you are measuring the EXACT SAME motion with both sensors. The signals are then differenced in order to calculate the rejection ratio.
Tried using the hydraulic shaker at first, however this shaker causes too much vibration to be reacted back into the work plate, which would cause the laser to shake. This being the case, I switched to an electrodynamics shaker. This proved more suitable and easier to work with.
I tested several different source voltages and bandwidths. As it turned out, I needed to source a relatively high voltage from the VXI source card in order to get enough energy in at low frequencies, but then had to use a lowpass filter to filter out the higher frequencies that were also inevitably high in energy due to the high source voltage.
After this I ordered some LN for Bryce, and talked with Scott P. about synchronizing the last Kaman probe box that was used for World View. He said he would discuss it tomorrow.
Brad talked to me more about the rejection ratio test. The noise floors I had been using as a basis for my previous PSD shapes were not the right ones. You need to use the noise floor of the signals AFTER they have gone through the circuit. He outlined a new approach going forward in an email.
Tried using the hydraulic shaker at first, however this shaker causes too much vibration to be reacted back into the work plate, which would cause the laser to shake. This being the case, I switched to an electrodynamics shaker. This proved more suitable and easier to work with.
I tested several different source voltages and bandwidths. As it turned out, I needed to source a relatively high voltage from the VXI source card in order to get enough energy in at low frequencies, but then had to use a lowpass filter to filter out the higher frequencies that were also inevitably high in energy due to the high source voltage.
After this I ordered some LN for Bryce, and talked with Scott P. about synchronizing the last Kaman probe box that was used for World View. He said he would discuss it tomorrow.
Brad talked to me more about the rejection ratio test. The noise floors I had been using as a basis for my previous PSD shapes were not the right ones. You need to use the noise floor of the signals AFTER they have gone through the circuit. He outlined a new approach going forward in an email.
Subscribe to:
Posts (Atom)