# Expanding oscilloscope bandwidth by amplifying higher frequencies

Is it possible to increase the oscilloscope bandwidth by:

• Adding an amplifier circuit to the oscilloscope input.
• This circuit would have a frequency-dependent gain selected in such a way as to correct the attenuation of higher frequencies inside the oscilloscope.
• The voltage gain would be equal to 1 in the case of low frequencies, reach about 1.3 (3dB) at the oscilloscope bandwidth and further increase as a function of frequency to correct the attenuation inside the oscilloscope

In my case, I have an oscilloscope with a sampling frequency of 8GS/s and a bandwidth of only 100MHz, which can be extended to 350MHz for a relatively large fee.

1. Is it possible to "unlock" a 350MHz using such a system?
2. Is it possible to go further and increase the bandwidth even more?

Edit:

Example circuit that works in LTspice:

• Which oscilloscope? Higher end ones are software filtered. Lower end ones have an RC filter for variant control. Some Tektronix ones had their DRM hacked years ago. Commented Feb 28 at 10:41
• How high a freq are you going up to? If it is in the higher freq realms of RF, there might be a lot of noise looking at it on a scope.
– Abel
Commented Feb 28 at 10:47
• Oscilloscope is Rigol MSO5104, I thought about 500MHz Commented Feb 28 at 10:50
• Does this apply to you? instructables.com/… Commented Feb 28 at 10:52
• What you're asking about fits into the general topic of deconvolution, where you take a system that has its bandwidth limited/distorted, measure the transfer function, and then apply the inverse transfer function to correct. This is absolutely possible, but the extent to which it helps is going to depend intimately on what the transfer function of your scope looks like. Have you tried measuring a simple fast rise time signal to see what the step or impulse response looks like? Commented Feb 28 at 19:03

Generally speaking, bandwidth is the -3 dB point of the scope's analog front end. So if the scope's bandwidth is 100 MHz then a 100 MHz signal coming to the channel will be attenuated by -3 dB, or the amplitude will be reduced to its ~70%. Higher frequencies will be attenuated further.

Using a pre-amplifier may work for sine waves only as the main intention is to "gain back" the lost amplitude. However, if the incoming signal is composition of different frequencies (e.g. square wave or triangular wave) then you'll see a distortion and it's inevitable because the bandwidth is still there and limiting.

• I am aware of this, but I assume that the amplifier would amplify different frequency components with different intensity, just as a low-pass filter attenuates different frequencies with different intensity. Commented Feb 28 at 10:52
• @piotr That may work in theory but the problem is, how'd you know the attenuation amounts of the scope you are working with? Your amplifier should be fast to make the frequency analysis quickly and amplify the components accordingly. Also, the BW doesn't have to be set by a first order RC filter. It could be an n-th order one. Or, even worse, what if there's a brickwall digital/software filter? Commented Feb 28 at 11:28
• The second part doesn't make sense to me. A square wave is a sum of sine waves, as we know. If you correct the amplitude of the higher-frequency components this reverses some of the distortion caused by the band-limiting. Commented Feb 28 at 21:59
• @user253751 did you read my comment above? Commented Feb 28 at 22:35
• @RohatKılıç DSP is not the only way to make a frequency-dependent amplifier. Commented Feb 29 at 10:23

Because you can 'unlock' a higher bandwidth, the implication is that, internally, the frequency response is being controlled by DSP filters.

The overall frequency response of DSP filters is extremely complex and very different to a classic analog filter and will vary widely depending on what type of filter (or filters) have been implemented, number of taps and what window (if any) has been applied.

You can see a number of plots at Mathworks.

Trying to reverse engineer the frequency response of your oscilloscope would be a daunting (and nigh on impossible) task.

• I disagree that measuring the impulse response of a scope's filters would be nigh on impossible. It's actually very simple, just feed in an impulse and look at the response. I have done this on many digital systems. The problem is that a digital filter can have very high attenuation, so in practice you're going to be limited by SNR to small improvements in bandwidth. Commented Feb 28 at 14:46
1. Like you said, it is possible to expand the bandwidth by unlocking the system. Legal way to do it is to pay the "relatively large fee". For some scopes, illegal ways have existed.

2. No. Or maybe in theory, but you have to reverse engineer the system and "hack" it yourself, which in practice may be impossible to you. It may be that the components and calibration of your model may not perform well so that is why the model you bought has the specs that the model is guaranteed to have. You don't know the limitation of the analog parts, so 350 MHz is the bandwidth limit for your scope as a whole, both in analog and digital hardware and in the software. If you need more bandwidth, buy a model with the bandwidth you need.