measuring scope bandwidth with RF signal generator

[ice595]

Hi EEVblog   
How can I properly measure the bandwidth of an oscilloscope(100MHz) with an RF signal generator(2GHz).
I know that I can't simply connect them both together with an coax cable without termination on the scope end.
But how about using the X10 scope probe to probe the N-connector output of the signal gen and wind up the wick until it reaches the -3dB amplitude?

Also, since the scope that I am measuring doesn't have a built-in 50 ohm terminator, and I don't have access to a 50 ohm trough adapter.
Can I just use a BNC T-piece with a 50 ohm metal film resistor on one end? (the scope is only rated for 100MHz, but should I worry about the parasitic inductance/capacitance of the resistor?)

[John_ITIC]

Yes, just put an external 50 ohm resistor across one of the T-adapter ports. Use direct coax without probe. At 100 MHz you don't have to worry about less than perfect signal integrity on the termination.  Nor do you have to worry about parasitics as such low frequency.

Take one amplitude measurement at, say 10 MHz, as a baseline (0dB). Then crank up the signal generator until the measured amplitude goes down by -3dB (0.7 x the original amplitude - see https://en.wikipedia.org/wiki/Decibel). That's your scope bandwidth.

/John.

[tggzzz]

Hi EEVblog   
How can I properly measure the bandwidth of an oscilloscope(100MHz) with an RF signal generator(2GHz).
I know that I can't simply connect them both together with an coax cable without termination on the scope end.

Correct, and your questions below are sensible.

But how about using the X10 scope probe to probe the N-connector output of the signal gen and wind up the wick until it reaches the -3dB amplitude?

Yes, but you will be measuring the bandwidth of the scope and probe combination. See the "Formal" references in https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/

Understand why *10 probes are actually low impedance, whereas "low impedance Z0" probes are actually higher impedance (and higher bandwidth)

Also, since the scope that I am measuring doesn't have a built-in 50 ohm terminator, and I don't have access to a 50 ohm trough adapter.
Can I just use a BNC T-piece with a 50 ohm metal film resistor on one end? (the scope is only rated for 100MHz, but should I worry about the parasitic inductance/capacitance of the resistor?)

At 100MHz, a small surface mount resistor with short leads should be fine - don't use a wirewound resistor . Rule-of-thumb: 1mm of wire is -> 1nH, and you can do the calculations/simulations yourself.

Alternatively use a BNC T piece http://uk.farnell.com/te-connectivity-greenpar/5-1634532-1/adaptor-bnc-plug-bnc-jack-jack/dp/1020971
and cheap terminator http://uk.farnell.com/multicomp/spc15280/rf-coaxial-bnc-terminator-ra-50ohm/dp/2293866 - you can get them anywhere.

If you want to measure subnanosecond risetimes (typical of modern logic) with a 100MHz scope, you are out of luck. But you can use the inexpensive techniques outlined at https://entertaininghacks.wordpress.com/2015/08/11/measuring-digital-signal-edge-rates-without-an-oscilloscope/

[ice595]

Just got my Rigol DS1054z today (after my almost 3 year old DS1102E bit the dust, luckily its still under warranty)
I took a BW measurement BEFORE the 100MHz hack just for reference and I was pleasantly surprised.
It seem like my scope is actually 90MHz out of the box.
Heres a the screen shot of the 10MHz baseline, -3dB before hack and -3dB after hack.

Here how I took the measurement, I used the compensated 150MHz X10 probe (which came with the scope) to probe the N-connector of my RF signal generator.
so the baseline test at 10Mhz shows 330mV, and the -3dB point of that should be 233mV right? (330mV x 0.707 = 233mV)

also, the hardware frequency counter seems funky after 110Mhz

[vk6zgo]

Keep in mind that the signal generator may not be accurate in signal level over the range of frequencies required.

[awallin]

DS2072A before and after upgrade
http://www.anderswallin.net/wp-content/uploads/2015/12/rigol_DS2072A_upgrade.png

at the 300MHz setting the response falls off much faster than a 1st order RC filter would. This is probably a well known thing, but something to keep in mind, i.e. can't use a simple RC-filter formula anywhere close to the specified scope bandwidth on a DSO. Older analog scopes might have smoother response?

[ice595]

I've just found my schmitt trigger inverter circuit like the one that W2AEW built ->
The measured rise time was 2.9ns which is more than 100MHz I suppose 
its silly how the horizontal scale doesn't go down less than 5ns/div

[tggzzz]

Here how I took the measurement, I used the compensated 150MHz X10 probe (which came with the scope) to probe the N-connector of my RF signal generator.

If the output is not properly terminated then a reliable measurement cannot be made.

A *10 probe's frequency response is probably not completely flat, and certainly its input impedance has to be accounted for at these frequencies. If you don't have the maths to calculate it, you should at least do a spice simulation.

The correct way to measure the bandwidth is to use a 50ohm cable to the scope, and to have a 50ohm termination in (or at) the scope.

[tggzzz]

The measured rise time was 2.9ns which is more than 100MHz I suppose 

The standard formula (0.35/tr) is based on assumptions that may or may not be true. It is, at best, a rule of thumb. Chief among the assumptions is that the frequency response of the scope+probe is gaussian.

[Solder_Junkie]

I use a Pico 50 Ohm terminator, they are cheap and are rated to 1 GHz and only cost £10 each (plus post, tax, etc)
https://www.picotech.com/accessories/bnc-terminators-leads/50r-terminator-bnc.

They are very useful when developing/testing radio gear, so at that price it's not worth bothering with T adapters, etc.

[KJDS]

I use a Pico 50 Ohm terminator, they are cheap and are rated to 1 GHz and only cost £10 each (plus post, tax, etc)
https://www.picotech.com/accessories/bnc-terminators-leads/50r-terminator-bnc

They are very useful when developing/testing radio gear, so at that price it's not worth bothering with T adapters, etc.

I've knocked the full stop off the end of your link so it works