 |
 |
| • |
Weapons testing |
| • |
Sports training and analysis |
| • |
Locomotion and feeding research |
| • |
Crash analysis |
|
 |
| • |
Reduce jams |
| • |
Speed up line setup and changeovers |
| • |
Lower scrap and rejected material
costs |
| • |
Reduce downtime and maintenance
expenses |
|
 |
| • |
New product design
and development |
| • |
Human biomechanics research |
| • |
Animal behavior analysis |
| • |
Flow visualization studies |
|
|
|
 |
|
How to determine which high-speed digital video camera
is best for your application |
|
How do you determine what
high-speed camera frame rate you need?
When considering the purchase of a high-speed digital camera
like TroubleShooter to examine process anomalies or study motion
mechanics, the first question to ask is, “How high a frame
speed do I need to capture the events that I need to see?” In
the case of production machinery, it is tempting to simply divide
the frame rate in seconds of a camera being considered by the
line speed in units per second to obtain the number of frames
per unit.
Example
The Fastec TroubleShooter LE 250 captures
video at 250 frames per second. |
|
If a system for liquid filling/capping/
 crimping
runs at:
—
400 containers per minute
(about 7 units per second)
—
Divide 250 frames/second
by 7 units/second to get 35.7
frames per unit.
30 frames per unit is usually considered to be enough to capture an
event, so the TroubleShooter LE 250 is the perfect camera by this
measurement.
The table below illustrates this concept by showing frames per piece
at progressive line speeds (in pieces per minute) for different camera
speeds in frames per second. |
 |
The Cream Zone numbers show (greater than 30 frames acquired per
piece)
the “safe” area where
there are reasonable assurances that an event will be effectively
captured.
The Tan Zone (11-30 frames per piece) indicates a “marginal” area
where successive trials may be necessary to capture an event.
The Orange Zone (10 frames or less) is the “inadequate” area
where event capture is not at all certain.
But the situation is often not quite that simple. The liquid filling
system may (and probably does) have multiple heads to fill multiple
containers simultaneously. We usually want to study the action of
only one of these heads. Suppose for simplicity’s sake that
this unit has seven heads, one head does one unit in one second.
We could probably record this event with high-speed digital video,
or just observe with the eye, which can generally distinguish events
down to 1/10 second. The issue now is that this “event”
as we’ve defined it is comprised of several events of interest,
which for this system could be: |
| We know from experience that load/unload can take as much as
half of the second we have to complete this unit. The other five
events must all occur in the half second that is left. If we are
fortunate and they each take about the same amount of time, then
each event can take 0.2 seconds. We’ll need a rate of 150
frames/second to acquire 30 frames for each event. Not too difficult,
but we have not yet considered acceleration. Steppers and servomotors
do not ramp up and ramp down instantly. |
 A
speed vs. time curve
for a motor motion can
look like this:
Ramp-up or ramp-down can be as little as 1/4 of the total interval
of 0.2 seconds, or 0.05 second. If we needed to capture a ramp-up,
we’d need 30 frames in 0.05 second, or 600 frames/second!
Fortunately, most systems are designed such that functional events
do not occur during ramp-up or down. What is important is that it
would be better to get 30 frames in 3/4 of our 0.2 seconds, or 0.15
second, so we need a camera speed of 200 frames/second to do this.
This is still well within reach of the TroubleShooter 250’s
capabilities.
Return to top> |
Sports Training
To briefly consider another application from sports training, suppose
we need to analyze a golf swing. Standard 30 fps video is adequate
to observe the entire swing. But, suppose we want to view the
impact of the club head on the ball in slow motion. We would
want to record the acceleration of the ball off of the head as
it first compresses into the head, and then springs off. This
event would occur in just a few thousandths of a second, and
would probably require frame rates between 500 and 1000 fps.
Under certain circumstances, the anomaly we wish to observe may
be short enough to happen by chance between frames, and will be
missed on a single recording. Does this mean we need a faster camera?
Not necessarily. What we need is the ability to record an event,
observe the frames on the spot, and shoot again if the anomaly was
missed. The probability is that in a few “takes,” we
should have enough information about the problem to make adjustments.
Remember, the goal is not to characterize the problem, but to gain
enough data to make adjustments.
Return to top> |
Determine The Camera Frame
Rate
Let’s now consider modeling an assembly line to determine
the camera frame rate necessary for effective analysis and adjustment.
For each system on the line, we can do the following:
1. If a machine performs multiple simultaneous operations like
our fictional liquid filler, we divide its actual line speed in
units per second by the number of simultaneous operations to get
the speed of each functional unit.
2. Now we can characterize a system in the same way that we did
our liquid filler, identifying those discrete operations of interest
that we would want to observe in slow motion, remembering our acceleration
issues if there are any.
3. By using manufacturer’s data, or interpolating from line
speed performance, we deduce the total interval for each of these
operations of interest. Manufacturer’s data, if available,
can be highly useful in this exercise.
4. Taking the smallest time interval for an operation of interest
on a system, we multiply by 30 frames to obtain the required speed
in frames per second. We now have the lowest effective frame rate
that will help us maintain this one system.
5. After repeating steps 1-4 for each system on our line, we rank
our systems from highest necessary frame rate to lowest. If one
of our systems stands out as requiring much higher frame rates
than the rest of the line, we may have to examine reliability data
for that system, and determine whether or not we need to obtain
a camera that will help us with all of our systems, or purchase
a less expensive camera that will satisfy the bulk of our needs.
Return to top> |
Lens Focal Length
Another factor often considered in the selection of high-speed cameras
is field of view (FOV), or how much of the subject that the camera
actually captures on a single frame. This is really a function of
the lens used on the camera and the distance of the camera from
the subject. The higher the focal length of the lens, the higher
the magnification, and the smaller the FOV, as the following chart
shows for a distance from camera to the sensor of 0.250 meters.
FOV (in meters) |
We appreciate your
questions and comments at Fastec, and we hope you will take the
opportunity to Contact
Us with your application issues. We would like to hear
from you.
Return to top>
Fastec
manufactures portable, point and shoot digital video cameras for
motion analysis in plant maintenance and field service troubleshooting,
research, military test and instrumentation and sports training.
|
|
Contact Us
|
