Record length: One More Question When Purchasing Oscilloscope!
A. Two misunderstanding on oscilloscope field:
1. Focus on sample rate, miss the record length
When selecting oscilloscope, people used to focus mainly on sample rate. Many of them tangle if the sample rate is 4G or 5G. However, the sample rate depends on record length from time to time. If your oscilloscope only gets 2.5K record length, the sample rate would never reach 4G even it supports.
2. The nominal sample rate VS the real-time sample rate
Almost every oscilloscope manufacturer chooses to print the sample rate (like “4GSa/s”) on PVC label and stick on oscilloscope. However, the nominal sample rate is the highest sample rate one machine can reach, not the real-time sample rate. The oscilloscope only keeps on the highest sample rate for a little while. Some oscilloscopes even eliminate the real-time sample rate display, which misleads the engineers to believe that the sample rate holds still all the time.
B. Parameter Explanation
Oscilloscope First Formula: Record Length= Sample Rate × Waveform Time
Similar with Battery Formula: Battery Volume= Consumption Rate × Working Time
The battery volume is fixed, to extend the working time, consumption rate should lower down.
The same truth that the sample rate must lower down while extending the waveform time.
Conclusion: for waveform observation, if the bandwidth is fit, the core parameter is the real-time sample rate. Real-time sample rate depends on record length. Let’s see the real-time sample rate and time scale under different record length.
Sample Rate under 512M record length
Sample Rate under 10M record length
Sample Rate under 2M record length
We can tell from the chart, the sample rate drops down when waveform time rise up.
C. The Core Value of Large Record Length
1. Malfunction Detect
Every oscilloscopes can easily trigger strong character error signals. But what frequency does it appear? Does it regularly show? It couldn’t tell without the help of large record length.
2. Data gathering
Large record length means more original data captured. The measurement sample points would be more. In measurement, waveform based on 512M points and 1M points are highly different.
D. Common Oscilloscope Record Length in Recent Market
In recent market, record length is small for oscilloscope under 1G bandwidth. Especially for well-known brand oscilloscope, stuck in older structure frame, instrument cannot deal with large quantity data itself. There is a brief compare chart:
T Company 3000 series oscilloscope owns 10M/channel record length, 4000 series oscilloscope owns 20M/channel. Even if you select the relevant scale, the instrument would also crash or stuck for loading too much information.
K Company holds great ability on instrument fluent running. But it is based on 4M record Length at maximum.
OWON XDS3102A holds a 40M deep record length. Ensure the great sample rate and keeps fluent running.
E. Hard Points to Implement Large Record Length
Is it so easy to change the Record Length Memory as it is on PC hard disk changing?
No. There are much more data burst out if the record length increases. The waveform needs to be rebuilt, trigger, decode, measure as well as fluent running guarantee. Cooling issue cause by heavy loading should also be concerned. If not considering these issues, the operating experience will largely lower down. Like you should wait 3s to implement autoset and 5s to wait for parameter measurement, is it worth large record length?
The point other oscilloscope manufacturer won’t tell you is:
1. The real-time sample rate will drop down as the waveform time increases. Nominal sample rate is the highest sample rate, the actual sample rate is in accordance with record length and time scale.
2. For most oscilloscopes under 1G bandwidth, the highest sample rate can only stay at waveform time within 2ms even 1ms.
Advantage for large record length oscilloscope:
1. Keep the highest sample rate for more time. Not only a flash.
2. Instead of capturing typical signals to study, analyze the waveform based on mass data.