Products > Test Equipment
Measuring Oscope Probe Bandwidth with an RF Sig Gen
<< < (3/4) > >>
noisyee:

--- Quote from: killingtime on November 08, 2023, 12:09:38 am ---Why the 50 ohm termination resistor? Is that to reduce reflections back into the rf generator?
You would normally only have to terminate into 50 ohms if you've got a long transmission line (coax say), to stop the line acting as a voltage transformer when terminated with a miss-match (SWR>1), but in my case the probe it stuck right on the end of the sig gen N socket, so there is almost no transmission line - less than 1/20th of a wavelength at the freq I'm using, so we can ignore it.

--- End quote ---

It's true when the amp of the sig gen is right after the connector. But some rf gen have a quite long pipe between amp and connector, which in some case can't be ignored.
Verifying it or adding termination resistor at your preference.


--- Quote from: killingtime on November 08, 2023, 12:09:38 am ---I tried that and got some weird results at LF and VHF using a nanoVNA. Capacitance crept up. Might need to build a dedicated test rig. Too many projects on right now.

--- End quote ---

In my test result, the slope of the impedance curve slowly decrease, which indicate an increasing of equivalent capacitance. So I think it's quite normal, like T3sl4co1l said:

--- Quote from: T3sl4co1l on November 07, 2023, 02:57:20 am ---if there is a lie, it's that an equivalent circuit only works over some frequency range; they might simply be leaving off further, extended details of that equivalent circuit -- whether because they don't care, or they don't know.

--- End quote ---

In some case where probe loading should be treated seriously, manufacturer will provide measured impedance curve and equivalent circuit, or even snp files for simulation.
David Hess:

--- Quote from: killingtime on November 08, 2023, 12:09:38 am ---Why the 50 ohm termination resistor? Is that to reduce reflections back into the rf generator?
You would normally only have to terminate into 50 ohms if you've got a long transmission line (coax say), to stop the line acting as a voltage transformer when terminated with a miss-match (SWR>1), but in my case the probe it stuck right on the end of the sig gen N socket, so there is almost no transmission line - less than 1/20th of a wavelength at the freq I'm using, so we can ignore it.
--- End quote ---

My Tektronix PG506 is designed that way also.  It uses coaxial parallel source termination built into the BNC socket which is directly attached to the fast diode which disconnects the source, so it does not require a load termination to produce a clean output.  However the load termination results in 25 ohms instead of 50 ohms driving the probe tip, so it doubles the bandwidth and duplicates how probes are specified.
Berni:

--- Quote from: killingtime on November 08, 2023, 12:09:38 am ---Why the 50 ohm termination resistor? Is that to reduce reflections back into the rf generator?
You would normally only have to terminate into 50 ohms if you've got a long transmission line (coax say), to stop the line acting as a voltage transformer when terminated with a miss-match (SWR>1), but in my case the probe it stuck right on the end of the sig gen N socket, so there is almost no transmission line - less than 1/20th of a wavelength at the freq I'm using, so we can ignore it.

--- End quote ---

The internal guts of the signal generator are not 0mm away from the output connector. So not having a terminated output can also cause reflections inside the coax/traces that connect components inside the signal generator itself. A lot of reflection could also possibly upset a automatic signal level control loop inside the signal generator.

A lot of things start to matter when you go into low single digit dB measurements.

But in general you have independently discovered why passive probes become fairly useless at >100MHz. When you put a 100MHz sine wave into the 12pF of input capacitance the total probe impedance is 130 Ohm to GND. As you might imagine placing a 130 Ohm resistor to ground on your signal of interest is definitely going to affect the signals amplitude. So the probe is not wrong, it is measuring what is there, the problem is that the signal is loaded down once the probe is connected, so it is no longer the same shape/size.

This is the reason why active probes are a very useful at high frequency. Yes they are expensive, bulky, fragile..etc but they also have only about 0.5pF to 1pF of loading. This brings our 100MHz sinewave example loading down a lot, to just 3000 Ohm.
David Hess:

--- Quote from: joeqsmith on November 08, 2023, 12:30:16 am ---
--- Quote from: David Hess on November 08, 2023, 12:22:54 am ---
--- Quote from: TimFox on November 07, 2023, 12:15:17 am ---The usual rule-of-thumb that RiseTime = 0.35/Bandwidth is not true in general for oscilloscope amplifiers, but is commonly used.
--- End quote ---

I have found it to be the rule rather than the exception, at least up to 300 MHz, but all of my oscilloscopes are old.

The oldest oscilloscopes that I know of where it is not the rule are some of the Tektronix 11k series plug-ins, and the 250 MHz option version of the 7704A.
--- End quote ---

Really old...
https://www.eevblog.com/forum/testgear/scope-speed-for-digital/msg3728101/#msg3728101
--- End quote ---

Mind are old, but the even older exceptions I mentioned above did not have Gaussian or first order responses despite being analog oscilloscopes, (1) and the DSOs that I have do have Gaussian responses so do obey the 0.35 rule, as do many modern DSOs which eschew the processing needed after digitization to produce a maximally flat response.

And then there are modern DSOs like the Rigol DS1000Z series which do not obey the 0.35 rule because their full power bandwidth is less than their small signal bandwidth, so their bandwidth varies with signal level.

(1) These old exceptions were tuned for maximally flat response like the DSOs discussed in the article, and suffered more aberrations in their pulse response, just like described in the article.
David Hess:

--- Quote from: Berni on November 08, 2023, 09:51:32 am ---
The internal guts of the signal generator are not 0mm away from the output connector. So not having a terminated output can also cause reflections inside the coax/traces that connect components inside the signal generator itself. A lot of reflection could also possibly upset a automatic signal level control loop inside the signal generator.
--- End quote ---

They are not, but as my example shows, it is practical to have an unterminated line length of only millimeters, which when combined with the limited bandwidth of the signal source results in no measurable reflections.

General purpose signal generators will not be constructed this way, so either a load termination is required or bandwidths must be kept lower than in my 500 MHz example.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod