Why do oscilloscopes have bandwidth limits?

[bdunham7]

yes but why should it be the case that an oscilloscope is a low pass filter?

Because everything is a ultimately a low-pass filter?  The analog front end (input, amplifier, scaling) will have limitations that depend on what you are willing to spend on them.

As BW goes up, it becomes increasingly expensive to make these analog components linear and coherent, especially in the case of DC-to-BW limit designs.  Not only do you get loss of amplitude, you get group delay and and other distortions that make it pointless to continue beyond a certain point.  Your inexpensive 8GSa/s 350MHz scope (gee, I wonder which one that is?    )  doesn't even have 50R inputs, so it is going to be very difficult to get a 1GHz signal cleanly to its inputs, let alone scaling and amplifying it within the scope.  For the digital signals you want to look at, just one probe that could connect those signals to an appropriate input will cost more than your entire scope.

[TimFox]

Even if we have a good noise figure in the amplifier, the fundamental thermal noise from a 50 ohm resistor is 0.9 nV/Hz^1/2.
If you had an infinite (or much too high for the application) analog bandwidth, the amplified thermal noise added to your MHz-bandwidth signal would be huge.
This is why many analog oscilloscopes have a front-panel switch to limit the bandwidth to much less than the full bandwidth:  you can see easily the reduction in trace width (at sensitive V/div settings) with the lower bandwidth.

[Circlotron]

Why can't my car go faster than sound?

Tractive effort goes up with the square of the speed.
Power required goes up with the cube of the speed.
So if your boxy, mid sized 1970s car needs 15 horsepower to travel at 100kph then it will need 28,255 horsepower to travel at 1235kph.
Oscilloscope cost vs performance goes up similarly.

[SL4P]

Read about  dV/dT

[AG6QR]

Series inductance, parallel capacitance, and resistance turns everything into a low pass filter.  Any wires will have at least a bit of series inductance, parallel capacitance, and resistance.

[glarsson]

If you had an infinite (or much too high for the application) analog bandwidth, the amplified thermal noise added to your MHz-bandwidth signal would be huge.

I don't have anything infinite at hand to test with, but wouldn't infinite bandwidth result in infinite thermal noise?

[TimFox]

If you had an infinite (or much too high for the application) analog bandwidth, the amplified thermal noise added to your MHz-bandwidth signal would be huge.

I don't have anything infinite at hand to test with, but wouldn't infinite bandwidth result in infinite thermal noise?

Yes.  That was my point.
In practice, where there is no such thing as infinite bandwidth, increasing the bandwidth increases the thermal noise voltage proportional to the square root of the bandwidth (all other things being equal).
This is clearly visible on a good analog CRO, looking at the trace width at high sensitivity, when changing the bandwidth.

[glarsson]

If you had an infinite (or much too high for the application) analog bandwidth, the amplified thermal noise added to your MHz-bandwidth signal would be huge.

I don't have anything infinite at hand to test with, but wouldn't infinite bandwidth result in infinite thermal noise?

Yes.  That was my point.
In practice, where there is no such thing as infinite bandwidth, increasing the bandwidth increases the thermal noise voltage proportional to the square root of the bandwidth (all other things being equal).
This is clearly visible on a good analog CRO, looking at the trace width at high sensitivity, when changing the bandwidth.

Yes, but huge is not infinite. You can't use an analog cro to prove what happens at infinite bw.

[TimFox]

If you had an infinite (or much too high for the application) analog bandwidth, the amplified thermal noise added to your MHz-bandwidth signal would be huge.

I don't have anything infinite at hand to test with, but wouldn't infinite bandwidth result in infinite thermal noise?

Yes.  That was my point.
In practice, where there is no such thing as infinite bandwidth, increasing the bandwidth increases the thermal noise voltage proportional to the square root of the bandwidth (all other things being equal).
This is clearly visible on a good analog CRO, looking at the trace width at high sensitivity, when changing the bandwidth.

Yes, but huge is not infinite. You can't use an analog cro to prove what happens at infinite bw.

Yes, huge is not infinite.  I wrote "infinite (or much too high for the application".
Infinity only makes sense mathematically in "the limit as the variable goes to infinity".
If you take an analog CRO with a very wide inherent bandwidth and look at the trace width as you vary the bandwidth of the vertical amplifier, you see the trace width increase in such a fashion that, using mathematical language, the width increases without bound as the bandwidth increases.
Were an infinite bandwidth possible, that would give infinite energy in the output, which is obviously non-physical.
In one of the simple derivations of thermal noise, the calculation is done for a resistor and capacitor connected together, in thermal equilibrium;  the bandwidth for the noise voltage calculation is determined by that time constant.
Your quest to measure something with infinite bandwidth will be fruitless.

A related problem (the "ultraviolet catastrophe", q.v.) led to the introduction of Planck's constant in thermodynamics:  see https://web.mst.edu/~kosbar/ee3430/ff/Wireless/noise2/index.html

[glarsson]

Your quest to measure something with infinite bandwidth will be fruitless.

I have not expressed any intention to measure with infinite bandwidth.

[BillyO]

I can explain in one sentence as follows:

In real life we have things like stray capacitance, inductance and resistance coupled with power limitations.

[TimFox]

Quoting from above: "Yes, but huge is not infinite. You can't use an analog cro to prove what happens at infinite bw."

[David Hess]

Linear circuits have gain-bandwidth product limits.  Bandwidth can be increases by sacrificing gain, but then more stages are required and this quickly reaches diminishing returns.

I say linear circuits because *sampling* oscilloscopes avoid this limitation by sampling before amplification, and the result is massive bandwidth determined only by sampling strobe width.

[kcbrown]

Linear circuits have gain-bandwidth product limits.  Bandwidth can be increases by sacrificing gain, but then more stages are required and this quickly reaches diminishing returns.

I say linear circuits because *sampling* oscilloscopes avoid this limitation by sampling before amplification, and the result is massive bandwidth determined only by sampling strobe width.

How does the A to D converter play into the bandwidth characteristics?  Surely it has some effect, no?

[LaurentR]

From experience how far can you use an oscilloscope beyond it's bandwidth limit?   Specifically, I'd like to compare the phase difference between in and out DDR2 data busses between an FPGA and a DRAM chip.  This is important because the input signals will need to be captured inside the FPGA using a clock with a phase delay to the output clock.  The slowest speed of the DRAM is 125MHz, which is starting to imply a 1GHz 'scope for many thousands of dollars.  Just for one measurement!  Will a 350MHz (soft upgraded) scope be any help to me at all?

If you have an FPGA, you may find that the easiest way to measure what you want to measure (e.g. tDQSCK seen from the FPGA I/Os) is probably to do it from inside the FPGA by using on-board circuitry. I am not sure what you're trying to do, but to a first degree, you should be able to make your circuit work without having to measure anything on the board with an oscilloscope.

[ejeffrey]

If you had an infinite (or much too high for the application) analog bandwidth, the amplified thermal noise added to your MHz-bandwidth signal would be huge.

I don't have anything infinite at hand to test with, but wouldn't infinite bandwidth result in infinite thermal noise?

No the plank law cuts in above a few THz and the thermal noise drops off exponentially and has a finite integral out to infinite frequency.

[CatalinaWOW]

Why can't my car go faster than sound?

No reason why it shouldn't but that would require an engine sufficiently powerful to accelerate the car against the restive forces of friction and air resistance which increase with velocity (reference Newton II: F=ma), and sufficient fuel onboard (mass which itself needs to be accelerated) to complete the acceleration.  In the case of a car, it would require rolling gear that maintains it's mechanical integrity at the high RPM for travel above 330 m/s.

Can you put the equivalent physics for a 'scope in two sentences?

Sort of.  You might add some car details such as suspension capable of maintaining stability with the bump frequency that will be encountered.  So now your horsepower limit is representing the sampling rate, the running gear speed capacity is bandwidth and suspension limitations are rise time.  These are not direct analogs, but perhaps an illustration that the performance of a device requires several attributes which may be loosely or tightly related.  The car example is probably more relatable to a lot of people.  Having a great excess of horsepower or sampling rate does not achieve all of the desired performance.  (The reverse, however is true.  An insufficiency of any one of the elements will prevent the desired performance.)

[james_s]

Sort of.  You might add some car details such as suspension capable of maintaining stability with the bump frequency that will be encountered.  So now your horsepower limit is representing the sampling rate, the running gear speed capacity is bandwidth and suspension limitations are rise time.  These are not direct analogs, but perhaps an illustration that the performance of a device requires several attributes which may be loosely or tightly related.  The car example is probably more relatable to a lot of people.  Having a great excess of horsepower or sampling rate does not achieve all of the desired performance.  (The reverse, however is true.  An insufficiency of any one of the elements will prevent the desired performance.)

One can look at the various land speed record cars to get an idea of the real world challenges. One of the issues you run into is making wheels that can spin as fast as they need to without the tires and even the rims themselves breaking up from the centrifugal force. Then you have to keep the car on the ground, not easy when it is more like an airplane than a car in the first place. Then as you actually approach the speed of sound you have the shockwave to deal with.

[Fungus]

One of the issues you run into is making wheels that can spin as fast as they need to without the tires and even the rims themselves breaking up from the centrifugal force.

That's the reason CD drives (remember those?) stopped getting faster at 56x.

[David Hess]

Linear circuits have gain-bandwidth product limits.  Bandwidth can be increases by sacrificing gain, but then more stages are required and this quickly reaches diminishing returns.

I say linear circuits because *sampling* oscilloscopes avoid this limitation by sampling before amplification, and the result is massive bandwidth determined only by sampling strobe width.

How does the A to D converter play into the bandwidth characteristics?  Surely it has some effect, no?

With an exception which does not matter here, ADCs have the same limitation and contribute to the bandwidth like any other linear stage.

Starting on page 24, Circuit Concepts - Vertical Amplifier Circuits from Tektronix discusses how several cascaded single-pole responses result in close to a Gaussian frequency response.

A point made a couple pages later is that this response is deliberately chosen because it results in little to no overshoot.

[Black Phoenix]

One of the issues you run into is making wheels that can spin as fast as they need to without the tires and even the rims themselves breaking up from the centrifugal force.

That's the reason CD drives (remember those?) stopped getting faster at 56x.

And even some of the 52x would start to create cracks in the centre of the disc after some time. I have some Microsoft Windows CDs with airline cracks in the centre and fixed (at least one that I remember) drive who a CD just disintegrated inside.

[TimFox]

If you had an infinite (or much too high for the application) analog bandwidth, the amplified thermal noise added to your MHz-bandwidth signal would be huge.

I don't have anything infinite at hand to test with, but wouldn't infinite bandwidth result in infinite thermal noise?

No the plank law cuts in above a few THz and the thermal noise drops off exponentially and has a finite integral out to infinite frequency.

Yes, that was Planck's solution to the "ultraviolet catastrophe", keeping the total radiated power output from a black body finite.  The rest is history.

[james_s]

One of the issues you run into is making wheels that can spin as fast as they need to without the tires and even the rims themselves breaking up from the centrifugal force.

That's the reason CD drives (remember those?) stopped getting faster at 56x.

Of course I remember those, I still use them often.

[Anding]

From experience how far can you use an oscilloscope beyond it's bandwidth limit?   Specifically, I'd like to compare the phase difference between in and out DDR2 data busses between an FPGA and a DRAM chip.  This is important because the input signals will need to be captured inside the FPGA using a clock with a phase delay to the output clock.  The slowest speed of the DRAM is 125MHz, which is starting to imply a 1GHz 'scope for many thousands of dollars.  Just for one measurement!  Will a 350MHz (soft upgraded) scope be any help to me at all?

Not perfect but the MSO5000 with 100MHz bandwidth limit can help me measure the phase difference of 100MHz clock signals.  The generated duty cycle is 50% and the phase difference is 90 degrees whereas the reading is 64% and 100 degrees - but this is close enough to be helpful to problem solving with entry-level equipment

" alt="" class="bbc_img" />
" alt="" class="bbc_img" />