Bandwidth check of my oscilloscope with simple means ?

[Frank_MV]

Hello,
I would like to check the bandwidth of my oscilloscope (RTB2000- 100MHz), the amplitude level at -3dB.
Unfortunately, I do not have a sine wave generator with constant amplitude.
There is only a simple signal generator available 100KHz - 150MHz - whose amplitude is unfortunately frequency dependent.
Is there a possibility with simple means ?

br,
Frank

[Kibabalu]

you need a rectangular pulse with a slew rate as fast as possible. Then you measure the rise time T_r (time from 10% to 90% of the magnitude). The formula BW = 0.35/T_r gives you the bandwidth your oscilloscope at least has. If the slew rate of your pulse was not fast enough to exceed the bandwidth of your scope, than the real bandwidth could be even higher.

[tggzzz]

Why do you want it measure the bandwidth of a scope? For a scope the important thing is to preserve the fidelity of the wave shape.

The best way to verify that is to examine the rise time of a step function (not an impulse). A clean transition and flat top is important, unless you know the actual waveform and are prepared to do the deconvolution yourself.

A 100MHz scope will probably have a rise time of 3.5ns, so a waveform with a rise time of less that, say, 1ns will give you a good idea of the scope's response.

Of course you can never measure a waveform with a scope, only with a scope plus probes. There are many pitfalls for the unwary associated with poor probing technique.

[2N3055]

Hello,
I would like to check the bandwidth of my oscilloscope (RTB2000- 100MHz), the amplitude level at -3dB.
Unfortunately, I do not have a sine wave generator with constant amplitude.
There is only a simple signal generator available 100KHz - 150MHz - whose amplitude is unfortunately frequency dependent.
Is there a possibility with simple means ?

br,
Frank

As for simple check, no simple answer. You need leveled signal generator (or at least one that you know has decent amplitude accuracy) and sweep.

My best advice is to go to  tggzzz's  web site in his signature and have a read.. He has lots of great info on scope use, probes and such. Quite a lot actually, so it will keep you busy for a time..

[Fungus]

Hello,
I would like to check the bandwidth of my oscilloscope (RTB2000- 100MHz), the amplitude level at -3dB.
Unfortunately, I do not have a sine wave generator with constant amplitude.
There is only a simple signal generator available 100KHz - 150MHz - whose amplitude is unfortunately frequency dependent.
Is there a possibility with simple means ?

Yes. Turn on the "Rise time" measurement on your 'scope, set it to single trigger mode and touch the probe to a 5V supply to get a fast rising edge.

For 100MhZ the rise time displayed should be around 3.5ns.

(or 3V supply, or... battery, or...some low voltage supply)

[tggzzz]

Hello,
I would like to check the bandwidth of my oscilloscope (RTB2000- 100MHz), the amplitude level at -3dB.
Unfortunately, I do not have a sine wave generator with constant amplitude.
There is only a simple signal generator available 100KHz - 150MHz - whose amplitude is unfortunately frequency dependent.
Is there a possibility with simple means ?

Yes. Turn on the "Rise time" measurement on your 'scope, set it to single trigger mode and touch the probe to a 5V supply to get a fast rising edge.

For 100MhZ the rise time displayed should be around 3.5ns.

(or 3V supply, or... battery, or...some low voltage supply)

Don't forget to take account of how the inductance of the lead inductance  (~1nH/mm) and probe tip capacitance will affect the waveform!

[Fungus]

Don't forget to take account of how the inductance of the lead inductance  (~1nH/mm) and probe tip capacitance will affect the waveform!

Not much - it has to work at over 100MHz!

[tggzzz]

Don't forget to take account of how the inductance of the lead inductance  (~1nH/mm) and probe tip capacitance will affect the waveform!

Not much - it has to work at over 100MHz!

A 6"/15cm ground lead (or equivalent) and 15pF tip capacitance will visibly resonate at ~100MHz.

[Fungus]

Don't forget to take account of how the inductance of the lead inductance  (~1nH/mm) and probe tip capacitance will affect the waveform!

Not much - it has to work at over 100MHz!

A 6"/15cm ground lead (or equivalent) and 15pF tip capacitance will visibly resonate at ~100MHz.

a) A properly terminated BNC cable would be better, yes, but OP is asking for "simple means".
b) Resonance won't affect the "rise time" measurement shown on screen - they discard the top/bottom part of the signal when they calculate that.

[BravoV]

The best way to verify that is to examine the rise time of a step function (not an impulse). A clean transition and flat top is important ...

Re-emphasizing with red highlight, as these easily will be ignored or missed by beginners.

Guess where did I knew/learned that. 

[Fungus]

The best way to verify that is to examine the rise time of a step function (not an impulse). A clean transition and flat top is important ...

Re-emphasizing with red highlight, as these easily will be ignored or missed by beginners.

Guess where did I knew/learned that. 

I wonder if there's a reason why oscilloscopes measure "rise time" by discarding the top/bottom 10% of the signal?

[Berni]

No the 10% 90% is because you can't possibly accurately tell where exactly that edge begins and ends. If you want to cheat the rise time then you define it as 20% to 80% (some do use this trick)

Getting bandwidth from rise time is also not always completely accurate because it depends on what shape the frequency response has at the cutoff point. But the shapes that optimize a faster rise time out of the same bandwidth will also generally have more overshoot and ringing. Still the resulting bandwidth number is going to be about in the ballpark of +/- 20% so its good enough.

As people have said square waves are better for testing oscilloscopes because it not only shows the approximate bandwidth but also shows if the cutoff response of the input introduces extra overshoot. A lot of scopes will have a little bit of overshoot and that is fine, but if the overshoot is something like 20% of the signal amplitude there is definitely something wrong.

As a source for generating these fast rise time square waves its easiest to use a fast digital logic chip driving the coax trough a 50 Ohm termination. One popular circuit for doing this uses parallel Hex inverters https://www.epanorama.net/circuits/tdr.html to produce a fast clean square wave (Typically used for TDR measurements of cables, but is also an excellent oscilloscope test signal)

[nctnico]

In reality using risetime is a very poor method because it assumes a certain filter response for the oscilloscope. This isn't standard so using the risetime may give false results. The only way to be sure is to use a levelled sinewave generator (together with a 50 Ohm coax cable and 50 Ohm terminator on the oscilloscope input if the oscilloscope doesn't have a 50 Ohm input mode).

[tggzzz]

Don't forget to take account of how the inductance of the lead inductance  (~1nH/mm) and probe tip capacitance will affect the waveform!

Not much - it has to work at over 100MHz!

A 6"/15cm ground lead (or equivalent) and 15pF tip capacitance will visibly resonate at ~100MHz.

a) A properly terminated BNC cable would be better, yes, but OP is asking for "simple means".
b) Resonance won't affect the "rise time" measurement shown on screen - they discard the top/bottom part of the signal when they calculate that.

On most scopes it won't be correctly terminated, of course: it would be 50ohm//15pF. Whether that has an effect depends on the length of the cable - the standard standing wave stuff that I won't repeat here.

The resonance will affect a risetime measurement, to some extent. The sharper the resonance the greater the overshoot - and correspondingly shorter time taken to reach 90%  of the final amplitude.

Another issue, especially with scopes that digitally process the samples, is the scope's internal filter response. That can be optimised (or pessimised!) in different ways for different purposes.

[tggzzz]

No the 10% 90% is because you can't possibly accurately tell where exactly that edge begins and ends. If you want to cheat the rise time then you define it as 20% to 80% (some do use this trick)

Getting bandwidth from rise time is also not always completely accurate because it depends on what shape the frequency response has at the cutoff point. But the shapes that optimize a faster rise time out of the same bandwidth will also generally have more overshoot and ringing. Still the resulting bandwidth number is going to be about in the ballpark of +/- 20% so its good enough.

As people have said square waves are better for testing oscilloscopes because it not only shows the approximate bandwidth but also shows if the cutoff response of the input introduces extra overshoot. A lot of scopes will have a little bit of overshoot and that is fine, but if the overshoot is something like 20% of the signal amplitude there is definitely something wrong.

As a source for generating these fast rise time square waves its easiest to use a fast digital logic chip driving the coax trough a 50 Ohm termination. One popular circuit for doing this uses parallel Hex inverters https://www.epanorama.net/circuits/tdr.html to produce a fast clean square wave (Typically used for TDR measurements of cables, but is also an excellent oscilloscope test signal)

Circuit with imperfect but useful measurements (<300ps risetime): https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg1902941/#msg1902941

[Fungus]

Blah, blah.

I just did it with a capacitor shoved into an Arduino and I got very consistent times around the 3.5ns mark. 

(on my 100MHz Micsig)

[tggzzz]

Blah, blah.

I just did it with a capacitor shoved into an Arduino and I got very consistent times around the 3.5ns mark. 

What did you do with the capacitor?

I'm not surprised about the risetime; modern jellybean CMOS is faster than most people imagine. Especially those that toggle the output at 1Hz and think they have a 5Hz bandwidth waveform (5*fundamental frequency, cough).

[2N3055]

This is second topic with exactly same question from OP.

This one also devolved into discussion about creating fast pulses, like most of them..

Again, answer to OP question, no, there is no simple improvised way to measure -3dB point of scope front end without leveled signal generator.
If you had 200-500 signal generator with decent amplitude accuracy across the frequency (like SDG6000X or similar) you could fairly well estimate what the curve is.

OP never explained what he wants to accomplish or why he asked the question.  Without that it is hard to help him.

[Berni]

Circuit with imperfect but useful measurements (<300ps risetime): https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg1902941/#msg1902941

Yep it works great for testing oscilloscopes up to about 500MHz, for higher up it becomes too slow.

For going faster i would recommend just buying Leo Bodnars fast pulse generator:
http://www.leobodnar.com/shop/index.php?main_page=product_info&products_id=295


I have two of them (SMA and BNC version) and they are great little generators for making ridiculously fast pulses. Well worth the pretty low price of them. Being small is also a nice thing since you can bring the generator to where it is needed rather than using a cable (the regular garden verity coax cable really degrades the rise time of pulses that are this quick)

EDIT:
Also one way to level a wobbly freqency response signal generator is to use one of these:
https://www.keysight.com/en/pd-1000001913%3Aepsg%3Apro-pn-8472B/low-barrier-schottky-diode-detector-10-mhz-to-18-ghz?cc=SI&lc=eng
This takes in a high frequency RF signal and AM demodulates it into a DC voltage while having a pretty flat and stable frequency response. So you can use this DC voltage as feedback to adjust the signal generator to the same level at various frequencies.
These RF detectors can sometimes be picked up cheep on the used market.

[Fungus]

Here's the results from a coin cell battery held like this:

The capacitor was 100% necessary to get a good time. Without it I was getting results in the uS range.



Results are much less consistent for some reason (maybe because I'm touching the wires and both sides of the battery) but it can be done.

Here's my best one:

[Fungus]

What did you do with the capacitor?

I stuck it into the +5V and GND pins as a place to clip the probe onto. 

(no software needed!)

The Arduino was the closest thing I had to a power supply on the table.

[Fungus]

Maybe the simplest method of all is simply to believe what a reputable manufacturer prints on the front of the device.

(nb. it'll be a bit more than the labelled value - because they want to be sure...)

[Fungus]

I'd be interested to see what people can get using this cowboy technique on higher bandwidth scopes. 

[Elasia]

Someone needs to stay out of the special candies...

Case in point to not always go by what the mfg says... Siglent SDS2100x+.. can ONLY be legitimately raised to 350MHz but actually has a front end on all units that goes to 500MHz+, typically -3db of around 580-620MHz.

The best way to measure just the scope itself is to get one of leo's pulse gens you can directly attach and once you have a reference you can then attach your probes and see what they are rated for.  You can also check things like bnc cables... half the time people dont know just how horrible they are.  On the other hand most of this is moot if you are dealing with 100MHz and lower

[Fungus]

The best way to measure just the scope itself is to get one of leo's pulse gens ...

Sure, we all know what the best way is but OP wants "simple means".

(and his 'scope is 100Mhz)

Someone needs to stay out of the special candies...

It worked, didn't it?