Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: Digital Oscilloscopes



Ludwik!

One of the earlier "averagers" you describe is Nuclear Data's
Enhancetron. In the 60's UCSC (advanced lab.) used one to "enhance" the
signal from ruby fluorescence. The enhancetron came with a 256 channel
MCA and the display for both.

bc

Current devices are considerably more sophisticated, for example: http://www.oriel.com/down/pdf/06050.pdf


Ludwik Kowalski wrote:

I would like to thank Michael Edmiston for a brief introduction
to digital oscilloscopes (see below). Following his example, let
me share what I found to be very interesting and probably not
well known.

The signal-to-noise ratio was not very high on my emf displays.
But I was able to improve it by switching from the common
SAMPLE mode of operation to the AVERAGING mode of
operation. The idea was to produce a single picture from many
consecutive records, such as 10 or 100. Each consecutive
display (refreshment on the screen) is the average from the
already examined records. The noise is random and it decreases
when records are averaged. The signal, on the other hand, is
not random and it sticks out averaging. A spike totally immersed
in noise would not be seen in a trace of a single record but it
could thus be recovered by averaging. I was very impressed.
Ludwik Kowalski

I have not used the specific digital scope Ludwik mentions, but I have used
many different models. Terminology is not yet as standardized as I wish it
would be. The triggering modes typically have little resemblance to the
"delayed sweep" of analog scopes. Below is a description of how the
"triggering" works on many digital scopes.

* * * Analog to digital conversions * * *

The scope has an analog-to-digital converter (ADC) that converts the input
voltage to a number in some range. The ADC and associated circuitry can be
set to a particular "conversion rate." This is like setting the time base
on a conventional scope. For the rest of this discussion let's suppose the
ADC is set to make a conversion every millisecond.

The ADC is also set to a particular conversion gain. This simply sets the
range of voltages it can measure, but it also ends up setting the voltage
resolution. For the rest of this discussion this setting does not matter.

* * * Memory * * *

The conversions are stored in memory. The scope will have a certain number
of memory locations for storing these conversions. The size of memory is
often a power of 2 such as 1024, 2048, 4096. For the remainder of this
discussion let's say there is sufficient memory to store 2048 conversions.
We should picture this as a sequential array of numbers. If the ADC is
converting at 1 kHz then the number just stored in memory location 55
probably occurred 4 milliseconds after the conversion already stored in
memory location 51. (The reason for the word "probably" is explained
below.)

* * * Triggering * * *

A clever way to "trigger" the scope is not to trigger it at all. Just let
the ADC convert every millisecond and keep putting the numbers in the
memory. When we arrive at memory location 2047 we wrap back to the
beginning of memory and the next conversion goes to memory location 0, then
1, etc. This means memory gets continually rewritten. Assuming 2048 memory
locations and a 1 kHz conversion rate, there will always be 2.048 seconds of
data stored, and they are "sequential," but the oldest data point is not
typically stored in location zero. The oldest data point is stored in the
location we are just about to overwrite.

Let's suppose we have just written a data point in memory location 500 when
a "trigger event" is detected on the external trigger line of the scope.
What does the scope do? There are several options, each very valuable
under the right circumstances.

(1) Upon receiving the trigger, the scope can just stop taking data. That
means location 500 has the data that represent the voltage 1 ms before the
trigger event occurred. Location 499 has the data for 2 ms before the
trigger. Location zero has the data point that occurred half a second
before the trigger. Location 501 has the data point for 2.047 seconds
before the trigger.

Our "record" becomes data points 501 through 2047 plus zero through 500.
When plotted this way these data represent the voltages measured every
millisecond for the two seconds PRIOR to the trigger. We can say we have
collected data from t = T-2048 milliseconds to t = T. (where t = time and T
= trigger)

(2) Upon receiving the trigger, the scope can continue to take data for
anywhere from 1 to 2048 conversions.

(2a) If it takes data for 2048 conversions, then memory location 501 will
have data for the first millisecond after the trigger, and location 500 will
end up with the last data point which will be 2.048 seconds after the
trigger. If we plot the points 501-2047 followed by 0-500 we will plot the
2.048 seconds of voltages that occurred AFTER the trigger. We can say we
have collected data from t = T to t = T+2048 ms.

(2b) If it takes data for 1300 conversions then stops, then memory locations
501-1801 will have data for the 1.3 seconds after the trigger, and memory
locations 1802-2047 plus 0-500 will have the data for the 0.747 seconds
prior to the trigger. We can say we have collected data from t = T-747ms to
t = T+1300 ms.

(3) Upon receiving the trigger it can take data for a number of conversions
larger than 2048. Lets suppose it takes 2548 conversions. Only the last
2048 are saved, the first 500 conversions after the trigger get written
over. This means the data consist of 2.048 seconds of data starting half a
second after the trigger occurred. We can say we have collected data from t
= T+500 ms to t = T+2548 ms.

Conclusions... The number of data points collected is set by the number of
memory locations present (or utilized). This is the record length expressed
in terms of data points. The time duration of the data is the conversion
period (inverse conversion rate) times the number of memory locations
utilized. This is the record length expressed in time. The starting time
of the acquired data can coincide with the trigger or can precede or post
cede the trigger. The starting time of the acquired data cannot precede the
trigger by more than the record length. The starting time of the data could
conceivably post cede the trigger by any amount.

Nomenclature... Setting a time duration for data we want to keep before the
trigger event is often called the "pre-trigger" setting. Setting the time
duration for "ignoring" data after the trigger is often called the "delayed
trigger" setting.

You don't do both. If it is called a pre-trigger, then a positive
pre-trigger means we save some data prior to the trigger and a negative
pre-trigger means we throw away some data after the trigger.

If it is called a "delayed-trigger" or a "post-trigger" setting then a
positive number means we throw away data points after the trigger, and a
negative number means save data points before the trigger.

Fancier stuff...

a) This is all computer controlled, so the scope could take data at one
conversion gain and conversion rate for a certain number of points, then
switch to another gain and rate for the remainder of the points. This could
simulate the "delayed" sweep feature of an analog scope.

(b) There can be multiple channels (multiple ADCs with their associated
memory) and they do not have to have their pre-triggers, gains, rates set
the same. The different ADCs and their data records can give you data for
different windows of time around a particular trigger event.

(c) The trigger settings might be set via number of memory locations, or by
time, depending on the software in the scope.

(d) The trigger signal takes some time to process, and this might be
automatically adjusted for in the software, or you might need to account for
it in your pre-trigger setting.

(e) Since the ADC is free-running prior to the triggering event, the trigger
can be internal. This simply means it watches for a level-crossing in the
converted signal rather than looking for an external signal to detect the
trigger event.

Conclusions... Digital scopes are really cool, really powerful, and take
some getting used to.

Michael D. Edmiston, Ph.D. Phone/voice-mail: 419-358-3270
Professor of Chemistry & Physics FAX: 419-358-3323
Chairman, Science Department E-Mail edmiston@bluffton.edu
Bluffton College
280 West College Avenue
Bluffton, OH 45817