If you are using an oscilloscope, make sure you are using the right bandwidth! Choosing the wrong amount could adversely affect your measurement results. Let’s look at what oscilloscope bandwidth is and why you need just the right amount.
What is Bandwidth?
Bandwidth is often regarded as the single most important characteristic of an oscilloscope. Measured in Hertz, the bandwidth of your oscilloscope is the range of frequencies that your oscilloscope can accurately measure. Without enough bandwidth, the amplitude of your signal will be incorrect and details of your waveform might be lost. With too much bandwidth, you will capture excessive noise, providing you with an inaccurate measurement. Here’s why:
You can think of an oscilloscope like a low pass filter, meaning it will only pass frequencies from 0 Hz up to a specified frequency. An oscilloscope’s bandwidth is specified as the 3 dB down point of the filter. What the heck is a 3 dB down point? Read on.
Low pass filters allow signals to pass through them at full amplitude until the signal frequency approaches the high end of the frequencies that the filter can pass. Then a filter attenuates signals passing through them until the signal’s amplitude is dampened to nothing. When the signal is attenuated by three decibels (3 dB), that is the cutoff point for an oscilloscope’s bandwidth specification. If you aren’t familiar with decibels, the 3 dB down point is when the amplitude of a sine wave is 70.7% of its actual height. Look at the diagram below to visualize the frequency response of a low pass filter, depicted in blue.
Figure 1. Frequency response of a low pass filter, depicting the 3 dB down point and cutoff frequency.
So, if you have an oscilloscope that has a bandwidth of 200 MHz, you know that the cutoff frequency of that oscilloscope’s filter is 200 MHz. Why does this matter for your measurements?
Too Little Bandwidth
You can see from Figure 1 that if you are measuring a signal that has a higher frequency than the cutoff frequency, you’ll either see an attenuated and distorted version of your signal or not much of a signal at all. Even measuring a signal as fast as the bandwidth of the scope is not a good idea. Measuring a 200 MHz signal on a 200 MHz oscilloscope will not provide you with the best representation of your signal, as the filter has already begun to roll off and distort your input.
Measuring with too little bandwidth will provide distorted results
Here is the rule of thumb for choosing the right bandwidth:
- Digital signal measurements: five times higher bandwidth than the fastest digital clock rate in your system
- Analog signal measurements: three times higher bandwidth than the maximum signal frequency on an oscilloscope with a flat frequency response
For more detail on these rules, read Evaluating Oscilloscope Bandwidths for Your Application.
So why not just use an oscilloscope with the highest bandwidth possible?
Too Much Bandwidth
Oscilloscopes can capture environmental noise. Oscilloscopes also add noise to your signal from filtering, processing, and digitizing (though a high-quality oscilloscope will do all of this properly and add less noise than a poorly-designed scope). And noise occurs at all frequencies. So if you have a 200 MHz oscilloscope, that scope is only going to show noise up to 200 MHz. But, if you have a 33 GHz oscilloscope, it will add noise to your measurement through its entire measurement range up to 33 GHz, regardless of the frequency of your signal.
Increasing bandwidth increases noise
If you want to measure a 50 MHz signal, a 200 MHz oscilloscope will give you plenty of bandwidth to clearly display your signal without attenuation and filter distortion but not so much that it adds high frequency noise content to your measurement.
Insider tip: If all you have access to is a high bandwidth oscilloscope, but you are measuring low frequencies, turn on hardware filters in the oscilloscope to eliminate that high frequency noise and get a cleaner measurement.
The other reason why you probably don’t want to buy the highest bandwidth oscilloscope out there is price. The higher the bandwidth, the higher the price. If you are worried the bandwidth you need today will not be enough for future measurements, look for an oscilloscope that lets you upgrade the bandwidth with a software license. That way you can buy the bandwidth you need now and upgrade later without having to purchase a new oscilloscope or send it in to the factory for a hardware update. (Most Keysight oscilloscopes can be bandwidth upgraded with a software license for this very reason.)
Don’t be afraid to be the Goldilocks of bandwidth. Did she settle for the porridge that was too hot or too cold? No. She went for the one that was just right. And lucky for us, we won’t be eaten by bears if we set our bandwidth to just the right amount. Here is an example of how even a simple sine wave can be falsely represented on an oscilloscope without the right bandwidth.
In this demonstration, I am measuring a sine wave oscillating with a frequency of 80 MHz and a peak-peak voltage of about 2 volts.
I am using an 8 GHz oscilloscope. This is an excessive amount of bandwidth for an 80 MHz signal. The rule of thumb for analog signals is to use about 3 times the frequency of the signal. While this measurement doesn’t look horrible, let’s see how much better it can get when I apply the rule of thumb.
With only 240 MHz of bandwidth, look at how much cleaner my measurement is.
If I just want a quick check on the basics like voltage and frequency, the difference might not be crucial. But if I’m proving the quality of my design or attempting to pass strict performance or compliance specs, I would want the best (and cleanest) representation of my signal.
Now, I’ll decrease the bandwidth even further. As I mentioned earlier, you shouldn’t measure a signal at the bandwidth of the oscilloscope. The signal will be passing right through the 3 dB down point of the filter.
Here I’m measuring my 80 MHz signal with 80 MHz of bandwidth. You can see that the voltage is decreased from 1.92 V to 1.36 V. This is 70.8% of the voltage we should be seeing. The signal is attenuated by the filter.
To demonstrate the effects of the filter above the cutoff frequency, here is my measurement of the same signal with only 75 MHz of bandwidth. The signal is attenuated even further to 161 mV. The period of my measured signal is displayed as 12.74 ns. This would imply that the frequency of my signal is only 78 MHz, which we know to be false.
And here I’ve measured the same signal again with only 70 MHz of bandwidth. It barely looks like there is a signal at all.
You can see how dramatically the signal is attenuated when you try to measure a signal with frequency beyond the bandwidth of the oscilloscope.
Bandwidth is the most important characteristic of an oscilloscope
While there are many important features of an oscilloscope that you’ll need to evaluate before choosing one for your measurements, clearly bandwidth is the number one spec that you must check before any other. If you don’t have enough bandwidth you’ll see distorted or attenuated signals, giving you inaccurate measurements. If you have too much bandwidth, your measurements will be noisier than necessary. You have to choose a bandwidth that can support a clean and accurate representation of your test signals.
Now that you understand why bandwidth is the most important characteristic of an oscilloscope, check out Basic Oscilloscope Fundamentals to learn the other important oscilloscope characteristics and how to use an oscilloscope.