Ablaze
0
- Joined
- Oct 19, 2011
- Messages
- 462
- Points
- 0
I'm going to assume that is some sort of analog to digital converter. Have you tried diagnosing your chip? Perhaps replacing it? Have you checked the digital line after the chip?
LPF Donation via Stripe | LPF Donation - Other Methods
Links below open in new window
ArcticMyst Security by Avery
Two questions about the dynamics of a laser diode
- Thread starter Shellback
- Start date
- Joined
- Jun 23, 2010
- Messages
- 7
- Points
- 0
What are the scales on your oscilloscope traces? What does an aerosol particle trace look like? Do you only see the noisy signal when the gas flow is off?
Are there any switching power supplies in the area?
Out of curiosity, what is the size of the tube through which the the particles are flowing when counted?
The 3 L/min flow rate through a 1/8 inch tube would require a flow velocity of ~7 m/s. A micron particle traveling through a micron size beam spot would have a residence time of a (few) hundred ns which probably doesn't cause droplet evaporation. When the gas flow is off, is there a deposit of aerosol particles in the tube? Is it possible that there are particles or a film evaporating under persistent illumination within the tube which is causing an evolving scattering of light?
Cheers
Are there any switching power supplies in the area?
Out of curiosity, what is the size of the tube through which the the particles are flowing when counted?
The 3 L/min flow rate through a 1/8 inch tube would require a flow velocity of ~7 m/s. A micron particle traveling through a micron size beam spot would have a residence time of a (few) hundred ns which probably doesn't cause droplet evaporation. When the gas flow is off, is there a deposit of aerosol particles in the tube? Is it possible that there are particles or a film evaporating under persistent illumination within the tube which is causing an evolving scattering of light?
Cheers
I suspect it's not the chip
NI sbRIO-9631 - Embedded Device with AI, AO, DIO, 1M Gate FPGA - National Instruments
Also, I'm using the 9631 to do my high speed sampling, but I'm using a separate module (USB-9239) to do the data acquisition on my computer simultaneously and comparing the signals and getting the same signal for both methods (just in different resolutions).
1) What are the scales on your oscilloscope traces?
Generally, it's constantly autoscaling. Looking at the pics I posted (these were taken a couple of months ago) we're looking at a voltage range of ~7mV.
2) What does an aerosol particle trace look like?
An Aerosol particle looks like a single spike. Because we're masking off the main laser beam and the refracted beam from a particle scans from one side to another, the actual light beam that is being sensed is only available for an instant before it is being blocked by the mask. And, because of the positioning of the mask that we are using, the refracted beam is only seen once.
3) Do you only see the noisy signal when the gas flow is off?
No, I have been able to pick out the result of this signal when measuring larger particles. Generally, these spikes only affect the counts on the lower sizes that we're looking for, so when we were counting some of the larger particles overnight one night, I came back in the morning and found that an episode had occured during the early morning hours.
4) Are there any switching power supplies in the area?
All of the power supplies in the unit have been monitored for voltage deviations and none were found. I found that even as they cooled overnight, the voltage change was measured in the uV range and did not seem to affect the frequency or duration of the episodes.
5) Out of curiosity, what is the size of the tube through which the the particles are flowing when counted?
The tube is .032" ID. This happens even before any gas is being run through it. I cannot calibrate the unit with gas and particles. Also, we clean the flow cells thoroughly after assembly has completed.
- Joined
- Jun 23, 2010
- Messages
- 7
- Points
- 0
1) The tube is .032" ID. This happens even before any gas is being run through it. I cannot calibrate the unit with gas and particles. Also, we clean the flow cells thoroughly after assembly has completed.
Then your flow rates are even higher... If you are sampling at ~100kHz, then the duration of your signal from a gas particle is probably only one sample long, correct?
2)Generally, it's constantly autoscaling. Looking at the pics I posted (these were taken a couple of months ago) we're looking at a voltage range of ~7mV.
What is the time scale in the traces? What are the width of the noise spikes vs the width of the particle signal?
I don't know much about Photomultipliers, what is their bandwidth? MHz? GHz? It seems to me that you expect ultrafast signals on the order of a few nanoseconds. Unless your noise signal is in the upper megahertz range you might be able to high-pass filter it out. The time traces you are showing look like they are from a DAQ card. Do you have a true Oscilliscope available? I suspect that the bandwidth of the aerosol signals ( ~100MHz-GHz) and the noise (~10kHz - ???) are outside the range of the devices you are using to sample them (60kHz). Don't forget the Nyquist theorem!
Of course, I'm guessing all of these numbers based on the info you've given. Please correct me where I'm wrong!
Then your flow rates are even higher... If you are sampling at ~100kHz, then the duration of your signal from a gas particle is probably only one sample long, correct?
2)Generally, it's constantly autoscaling. Looking at the pics I posted (these were taken a couple of months ago) we're looking at a voltage range of ~7mV.
What is the time scale in the traces? What are the width of the noise spikes vs the width of the particle signal?
I don't know much about Photomultipliers, what is their bandwidth? MHz? GHz? It seems to me that you expect ultrafast signals on the order of a few nanoseconds. Unless your noise signal is in the upper megahertz range you might be able to high-pass filter it out. The time traces you are showing look like they are from a DAQ card. Do you have a true Oscilliscope available? I suspect that the bandwidth of the aerosol signals ( ~100MHz-GHz) and the noise (~10kHz - ???) are outside the range of the devices you are using to sample them (60kHz). Don't forget the Nyquist theorem!
Of course, I'm guessing all of these numbers based on the info you've given. Please correct me where I'm wrong!
1) Then your flow rates are even higher... If you are sampling at ~100kHz, then the duration of your signal from a gas particle is probably only one sample long, correct?
The flow rates I mentioned are based on the readout from a flowmeter calibrated for the kind of gas that we're monitoring. The goal is to only have one sample per particle (the precision is not complete because if two small particles are passing through the beam at the same time, it should show up as a particle twice its size) and we are, for the most part, accomplishing that.
2) What is the time scale in the traces?
It should be ~20msec
3) What are the width of the noise spikes vs the width of the particle signal?
As you can see in the pictures I posted (all of which are without particles), the initial onset of the particles are simply spikes which seem to evolve into the square
waves that are shown in the other image. All particles show up as a single spike with an amplitude proportional to the particle size.
4) I don't know much about Photomultipliers, what is their bandwidth? MHz? GHz?
Photomultipliers are essentially vacuum tubes with a photo-reactive anode. When a photon strikes the anode, the anode emits an electron, which is "steered" to a electrode that emits 2 electrons for every one electron that strikes it. The same is true for the following electrodes until it reaches the cathode. As I understand it, the PMT that we're using has an anode, 6 electrodes, and a cathode. Most PMT's are more sensitive to the blue-er end of the spectrum, but the specific PMT that we're using is engineered to be most sensitive at the near IR end of the spectrum (600-700nm).
5) It seems to me that you expect ultrafast signals on the order of a few nanoseconds.
The reaction of the PMT to light is as close to immediate as electrons can be.
6) Unless your noise signal is in the upper megahertz range you might be able to high-pass filter it out.
7) The time traces you are showing look like they are from a DAQ card. Do you have a true Oscilliscope available?
It is a very high speed DAQ device (up to 50k samples/second), but I do not have a true oscilliscope available. With a true O-scope, I would not have the datalogging capability that I have with the sbRIO board. The impulses also happen so fast, I would have a hard time distingushing them in the signal. This way I can gather the data, and analize it in real time in an effort to isolate the problem.
8) I suspect that the bandwidth of the aerosol signals ( ~100MHz-GHz) and the noise (~10kHz - ???) are outside the range of the devices you are using to sample them (60kHz). Don't forget the Nyquist theorem!
Because the particles show as individual spikes in the 1000 sample block, and the threshold value that I'm establishing is above the noise (usually), Nyquist really doesn't come into play. I'm not trying to measure frequencies. Even if it did, I'm not sure how that would relate to what's causing the odd phantom voltage spikes and eventual square-waves that I'm seeing.
The flow rates I mentioned are based on the readout from a flowmeter calibrated for the kind of gas that we're monitoring. The goal is to only have one sample per particle (the precision is not complete because if two small particles are passing through the beam at the same time, it should show up as a particle twice its size) and we are, for the most part, accomplishing that.
2) What is the time scale in the traces?
It should be ~20msec
3) What are the width of the noise spikes vs the width of the particle signal?
As you can see in the pictures I posted (all of which are without particles), the initial onset of the particles are simply spikes which seem to evolve into the square
waves that are shown in the other image. All particles show up as a single spike with an amplitude proportional to the particle size.
4) I don't know much about Photomultipliers, what is their bandwidth? MHz? GHz?
Photomultipliers are essentially vacuum tubes with a photo-reactive anode. When a photon strikes the anode, the anode emits an electron, which is "steered" to a electrode that emits 2 electrons for every one electron that strikes it. The same is true for the following electrodes until it reaches the cathode. As I understand it, the PMT that we're using has an anode, 6 electrodes, and a cathode. Most PMT's are more sensitive to the blue-er end of the spectrum, but the specific PMT that we're using is engineered to be most sensitive at the near IR end of the spectrum (600-700nm).
5) It seems to me that you expect ultrafast signals on the order of a few nanoseconds.
The reaction of the PMT to light is as close to immediate as electrons can be.
6) Unless your noise signal is in the upper megahertz range you might be able to high-pass filter it out.
7) The time traces you are showing look like they are from a DAQ card. Do you have a true Oscilliscope available?
It is a very high speed DAQ device (up to 50k samples/second), but I do not have a true oscilliscope available. With a true O-scope, I would not have the datalogging capability that I have with the sbRIO board. The impulses also happen so fast, I would have a hard time distingushing them in the signal. This way I can gather the data, and analize it in real time in an effort to isolate the problem.
8) I suspect that the bandwidth of the aerosol signals ( ~100MHz-GHz) and the noise (~10kHz - ???) are outside the range of the devices you are using to sample them (60kHz). Don't forget the Nyquist theorem!
Because the particles show as individual spikes in the 1000 sample block, and the threshold value that I'm establishing is above the noise (usually), Nyquist really doesn't come into play. I'm not trying to measure frequencies. Even if it did, I'm not sure how that would relate to what's causing the odd phantom voltage spikes and eventual square-waves that I'm seeing.
- Joined
- Jun 23, 2010
- Messages
- 7
- Points
- 0
1)The flow rates I mentioned are based on the readout from a flowmeter calibrated for the kind of gas that we're monitoring.
I trust your flow rate measurement I was simply commenting that my guesstimate of a 7m/s flow speed was too small!
2) The goal is to only have one sample per particle (the precision is not complete because if two small particles are passing through the beam at the same time, it should show up as a particle twice its size) and we are, for the most part, accomplishing that.
I suspect that you are easily accomplishing that! It appears your sample period is 8 microseconds and the events you are counting happen in nanoseconds.
3) The reaction of the PMT to light is as close to immediate as electrons can be.
The Photomultiplier recieves a photon and produces a current. There is a finite amount of time between receiving a photon and the photocurrent reaching its peak value --- this is the devices rise time. A quick google search suggests that rise times of PMTs may on the order of a few nanoseconds. Because it takes something on the order of 1 nanosecond to produce the current signal, the bandwidth of the device is something on the order of 1 GHz. The actual bandwidth of your device is probably not too far from this.
4) As you can see in the pictures I posted (all of which are without particles), the initial onset of the particles are simply spikes which seem to evolve into the square
waves that are shown in the other image. All particles show up as a single spike with an amplitude proportional to the particle size.
You must have an integrating ADC then.
5) It is a very high speed DAQ device (up to 50k samples/second), but I do not have a true oscilliscope available. With a true O-scope, I would not have the datalogging capability that I have with the sbRIO board. The impulses also happen so fast, I would have a hard time distingushing them in the signal. This way I can gather the data, and analize it in real time in an effort to isolate the problem.
The DAQ you are you using is quite nice, as far as DAQs are concerned. However, DAQs are relatively poor in temporal resolution as compared to the average oscilloscope (DAQs usually emphasize multi-channel and high resolution, whereas oscilloscopes usually emphasize high bandwidth). I understand that you want to capture aerosol particle events as a single sample spike; however, that means you are not resolving the full bandwidth of analog signal from the PMT... for the purposes of your counting device, you don't have to resolve the full bandwidth. This is not the issue.
The problem is understanding what the noise signal is. It is unclear from the time traces you have supplied whether your DAQ device is capturing the full bandwidth of this rouge signal. If this noise signal is composed only of frequency content at 25kHz and lower, then your DAQ measurement is sufficient to characterize the noise signal; however, it does not appear that the time trace of the noise signal is temporally resolved.
Resolving the frequencies is helpful for determining the source of the noise. Many switching-type circuits have switching frequencies in the 100s of kHz. Many op-amp circuits may "ring" at MHz. Because it's intermittent, I suspect that there is some device which is turning on and causing these bouts of noise in your signal. That device may be located inside or outside your device. The character of the noise may help you determine the culprit
What environment is this device being used in? What other machines are in the lab? You mentioned a compressor. Are there process chillers, computers, monitors, fans, power supplies, lasers? Mechanical vibrations?
Is there a possibility of retro-reflection of laser energy back into the diode?
You say the problem gets better after switching out diodes.. why kind of alignment process do you do when switching diodes?
I trust your flow rate measurement
I was simply commenting that my guesstimate of a 7m/s flow speed was too small!2) The goal is to only have one sample per particle (the precision is not complete because if two small particles are passing through the beam at the same time, it should show up as a particle twice its size) and we are, for the most part, accomplishing that.
I suspect that you are easily accomplishing that! It appears your sample period is 8 microseconds and the events you are counting happen in nanoseconds.
3) The reaction of the PMT to light is as close to immediate as electrons can be.
The Photomultiplier recieves a photon and produces a current. There is a finite amount of time between receiving a photon and the photocurrent reaching its peak value --- this is the devices rise time. A quick google search suggests that rise times of PMTs may on the order of a few nanoseconds. Because it takes something on the order of 1 nanosecond to produce the current signal, the bandwidth of the device is something on the order of 1 GHz. The actual bandwidth of your device is probably not too far from this.
4) As you can see in the pictures I posted (all of which are without particles), the initial onset of the particles are simply spikes which seem to evolve into the square
waves that are shown in the other image. All particles show up as a single spike with an amplitude proportional to the particle size.
You must have an integrating ADC then.
5) It is a very high speed DAQ device (up to 50k samples/second), but I do not have a true oscilliscope available. With a true O-scope, I would not have the datalogging capability that I have with the sbRIO board. The impulses also happen so fast, I would have a hard time distingushing them in the signal. This way I can gather the data, and analize it in real time in an effort to isolate the problem.
The DAQ you are you using is quite nice, as far as DAQs are concerned. However, DAQs are relatively poor in temporal resolution as compared to the average oscilloscope (DAQs usually emphasize multi-channel and high resolution, whereas oscilloscopes usually emphasize high bandwidth). I understand that you want to capture aerosol particle events as a single sample spike; however, that means you are not resolving the full bandwidth of analog signal from the PMT... for the purposes of your counting device, you don't have to resolve the full bandwidth. This is not the issue.
The problem is understanding what the noise signal is. It is unclear from the time traces you have supplied whether your DAQ device is capturing the full bandwidth of this rouge signal. If this noise signal is composed only of frequency content at 25kHz and lower, then your DAQ measurement is sufficient to characterize the noise signal; however, it does not appear that the time trace of the noise signal is temporally resolved.
Resolving the frequencies is helpful for determining the source of the noise. Many switching-type circuits have switching frequencies in the 100s of kHz. Many op-amp circuits may "ring" at MHz. Because it's intermittent, I suspect that there is some device which is turning on and causing these bouts of noise in your signal. That device may be located inside or outside your device. The character of the noise may help you determine the culprit
What environment is this device being used in? What other machines are in the lab? You mentioned a compressor. Are there process chillers, computers, monitors, fans, power supplies, lasers? Mechanical vibrations?
Is there a possibility of retro-reflection of laser energy back into the diode?
You say the problem gets better after switching out diodes.. why kind of alignment process do you do when switching diodes?
