100mhz scope, what does it really mean

[Rick Law]

I was going to watch that scope video, but my system keep hanging.  Not sure if it is the net or my machine or youtube itself.

I used to think a 100mhz scope means it could display a complete cycle of sine-wave at 100mhz without distortion.  The more marketing specs I read, the less sure I am now.

A 100mhz scope, what does it really mean?  If I want to see a complete sine-wave at 100mhz, what bandwidth do I really need?

100mhz is just a number out of the air, could be 10mhz, could be 1khz, or even 50/60hz.

Thanks
Rick

[Hydrawerk]

100MHz oscilloscope displays a 100MHz sinewave with -3dB less amplitude.

[Rick Law]

HydraWerk and Aurora, thanks both.  I didn't know that "without distortion" is numerically defined as 3db drop.  I just recalled it as "without distortion" which is rather subjective.  3db - I learn something new today, that makes it a good day.  Thanks!

Oh, the other good thing is, it looks like my recollection is essentially correct, good to know.  Last time I used a scope was over 30 years ago - it was a Tektronix...  No digital scopes back then.

[w2aew]

It really has nothing to do with distortion, it's simply about frequency response. As the others have stated, it means that a 100MHz sinewave will be no more than 3dB down in amplitude.

Is this the video you were trying to watch?

[Rick Law]

It really has nothing to do with distortion, it's simply about frequency response. As the others have stated, it means that a 100MHz sinewave will be no more than 3dB down in amplitude.

Is this the video you were trying to watch?

You are right - your post make me think a bit.  I am too analog in my thinking here.  Before storage scopes and digital scopes, all the effects you are going to get is in the visual representation of the wave form.  So, I "file it" in my mind as the visual distortion.  At 1g samples per second, a choke point of 100mhz somewhere in the pipeline has different implications than merely visual representation.  My mind was locked into merely thinking about the visual representations.  Your point is thought provoking.  I have to rethink (enough so that I can explain it to myself to make myself understand).  I have to rethink digitally.

The video I was trying to watch was the "Intro to Oscilloscope".  This one you are point out is one I should watch also.  Once I finish that, I will see if I really understand scope bandwidth.

Thanks for pointing it out.

Man, how much one can actually forget after 30+ years.  Hard to imagine, there was a time when I can grab a bunch of parts and make an 808x CPU work on a breadboard designing it on the fly and program the darn thing with dip-switches.

Rick

[jpb]

For a general signal containing a mixture of frequencies, the frequency response will give rise to distortion - things like a square wave or a ramp. The better scopes have a smooth roll off with frequency whilst some of the cheaper scopes have peaks and troughs before the 3dB point. There is a thread on this forum where people have plotted their scopes frequency responses (though you need access to the right equipment to measure it accurately).

Generally, if you want an accurate representation of an analogue signal I think the rule of thumb is you can go up to around a third of the 3dB bandwidth, so a 100MHz scope allows you to accurately display signals up to 30MHz or so. Of course you'll still see something for signals well above 100MHz.

Here is a useful Agilent app note:

http://cp.literature.agilent.com/litweb/pdf/5989-5733EN.pdf

[alm]

Having some knowledge of frequency domain representations (eg. Fourier transform) of signals is very helpful here. Square wave for examples have many odd harmonics. A 30 MHz square wave will have the third harmonic at 90 MHz, this will be about -3 dB down, and the fifth at 150 MHz, which will be much more attenuated. So you're essentially making a square wave from just the first and third harmonic. This will be a pretty ugly square wave.

One third of the bandwidth sounds very optimistic to me, I'd rather stick to one fifth. This is for a decent visual representation of the signal. Zooming in or using automated measurements in a DSO might require much more bandwidth. Which is why the Agilent document uses rise times, not repetition rates, for digital signals.

[jpb]

Having some knowledge of frequency domain representations (eg. Fourier transform) of signals is very helpful here. Square wave for examples have many odd harmonics. A 30 MHz square wave will have the third harmonic at 90 MHz, this will be about -3 dB down, and the fifth at 150 MHz, which will be much more attenuated. So you're essentially making a square wave from just the first and third harmonic. This will be a pretty ugly square wave.

One third of the bandwidth sounds very optimistic to me, I'd rather stick to one fifth. This is for a decent visual representation of the signal. Zooming in or using automated measurements in a DSO might require much more bandwidth. Which is why the Agilent document uses rise times, not repetition rates, for digital signals.

I did say analogue signals, a square wave is probably a digital signal. But yes, 1/3 is probably a minimum. The app note gives a fairly balanced view (though obviously as it is produced by a scope manufacturer it is going to err on the side of higher bandwidth being needed).

It depends to some extent what you want to look at. With the 30MHz square wave example, yes it would not give you very accurate rise times and even the shape would be wrong but you would be able to get the duty cycle and the general level of the signal and spot if there was a runt pulse. You'd also be able to estimate the frequency quite well (though counters are quite cheap).

The third harmonic would be down by about 26% and the fifth by 44.5% in terms of amplitude but in the harmonic series for a square wave the terms go as 1/n where n is the harmonic so a 26% reduction in the third harmonic corresponds to about 8% overall and a 44.5% in the fifth corresponds to around 9% overall but the two have peaks in different places so the errors wouldn't simply add. The 30MHz square wave wouldn't be that ugly. For a simple Gaussian response (1 pole) the first harmonic would be down 3% which might actually help the shape though it would make the amplitude measurement less accurate.

As an experiment I used graph plotting software to plot a square wave up to the 19th harmonic (it got a bit tedious as I was doing it by hand) and then the same 19 harmonics attenuated as if a 30MHz wave on a 100MHz scope (assuming just a simple Gaussian roll off) - here is the graph:

[w2aew]

If you're interested, I did a video a while back discussing the harmonic content of square waves, including "building" a waveform from the harmonic components individually to see how the process works.

[Rick Law]

Having some knowledge of frequency domain representations (eg. Fourier transform) of signals is very helpful here...

If you're interested, I did a video a while back discussing the harmonic content of square waves, including "building" a waveform from the harmonic components individually to see how the process works.
 

It depends to some extent what you want to look at.

I sure am interested.  I will hit the video over dinner.  Yes I am familiar with Fourier Transform.  But I need to relearn this.  I have post-grad training in Physics (master degree) and I did my share of electronics in my Physics labs.  But since graduate school, I have not worked in the industry.  I forgot too much.

I caught myself explaining something wrong to my daughter.  It took me a moment to remember some basic things.  I don’t have anything specific in mind to look at with a scope per-se.  I just want to relearn some of the stuff I forgot.

Thanks for your inputs, guys.  Most helpful.