Author Topic: x1 Oscilloscope Probe Shootout  (Read 6100 times)

0 Members and 1 Guest are viewing this topic.

Offline gf

  • Super Contributor
  • ***
  • Posts: 1425
  • Country: de
Re: x1 Oscilloscope Probe Shootout
« Reply #25 on: September 25, 2023, 09:52:26 am »
Best way would be to go with coax from AWG output to a T that is connected directly on BNC on reference channel set to 50Ω. Then connect a probe to that T. That creates 25Ω feed point directly on reference input channel.

The probe "sees" 25Ω, nevertheless the Vin/Vref quotient still represents the frequency response as if the probe were fed with zero source impedance, because the measured Vref is the voltage at the probe tip and not the open circuit voltage of the virtual 25Ω source. Whenever you measure Vref at the probe tip, it does not matter which source impedance the pobe can "see" during the measurement -- it cancels out.
« Last Edit: September 25, 2023, 09:54:16 am by gf »
 
The following users thanked this post: 2N3055

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 39026
  • Country: au
    • EEVblog
Re: x1 Oscilloscope Probe Shootout
« Reply #26 on: September 25, 2023, 10:02:54 am »
Best way would be to go with coax from AWG output to a T that is connected directly on BNC on reference channel set to 50Ω. Then connect a probe to that T.
That creates 25Ω feed point directly on reference input channel.
That is optimal way for Bode and also for waveform comparisons...
It is not important what is on output of AWG but what the reference scope channel sees..

Yes, this is the optimal way. I'll rerun these tomorrow, and maybe test some more brand probes. I'll also turn off the annoying graph autoranging.
 
The following users thanked this post: 2N3055

Offline tautech

  • Super Contributor
  • ***
  • Posts: 29810
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: x1 Oscilloscope Probe Shootout
« Reply #27 on: September 25, 2023, 10:11:23 am »
Best way
use of two generator channels.
The input capacitance of X1 probes is 50-90 pF.
Without interference and distortion, T-split operates up to 2 MHz.

Bode plot works by comparing against reference chanel. It is differential measurement...
FYI, the Bode plot connection image artur0089 linked is from rf-loop in first page of his Bode Plot II thread.
It is a valid connection type once AWG channels have been copied/linked and aligned and does not require BNC Tees.
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline artur0089

  • Contributor
  • Posts: 28
  • Country: ru
Re: x1 Oscilloscope Probe Shootout
« Reply #28 on: September 25, 2023, 10:26:53 am »
Amplitude +8 dB.
 :-+
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7465
  • Country: hr
Re: x1 Oscilloscope Probe Shootout
« Reply #29 on: September 25, 2023, 10:56:15 am »
Amplitude +8 dB.
 :-+

That is just some 1x/10x 60MHz probe from old Picoscope.

Now that you mentioned it I went back and did another measurement.
But I did something else before.

Despite what people tell you that compensation trimmer should not make difference in 1x mode and is used only in 10x mode, on some probes it does stay in circuit, at least partially, or maybe simply parasitically.

Therefore I went and adjusted compensation in 1X mode for minimum peaking.  That of course completely screwed up 10x compensation but that is not important now.
Point is that I managed to change frequency response very much..
Now peak is only 3.4 dB... And -3dB point is 35 Mhz

That shows how comparison of measurement equipment, even as simple one as probes calls for lot of scrutiny, preparation and attention to detail to ensure really level playing field.

"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7465
  • Country: hr
Re: x1 Oscilloscope Probe Shootout
« Reply #30 on: September 25, 2023, 11:06:35 am »
Best way
use of two generator channels.
The input capacitance of X1 probes is 50-90 pF.
Without interference and distortion, T-split operates up to 2 MHz.


Bode plot works by comparing against reference chanel. It is differential measurement...
FYI, the Bode plot connection image artur0089 linked is from rf-loop in first page of his Bode Plot II thread.
It is a valid connection type once AWG channels have been copied/linked and aligned and does not require BNC Tees.

I remember. He used this setup with DUT that has 50Ω impedance inputs itself and not probe testing. In addition to that he does say to use it only on low frequencies.

With Bode test we are not concerned with fact that signal from AWG will get amplitude or phase variations. We only need to ensure reference channel sees exactly the same signal as DUT as much as possible.
 It is relative measurement.
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 39026
  • Country: au
    • EEVblog
Re: x1 Oscilloscope Probe Shootout
« Reply #31 on: September 25, 2023, 11:10:01 am »
Here you go, tests re-run with practically the same results as my original video.
CH1 ref input 50ohm terminated, and CH2 probe to BNC adapter on a CH1 T-piece.
« Last Edit: September 25, 2023, 11:27:00 am by EEVblog »
 
The following users thanked this post: thm_w, 2N3055, pdenisowski

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 39026
  • Country: au
    • EEVblog
Re: x1 Oscilloscope Probe Shootout
« Reply #32 on: September 25, 2023, 11:26:12 am »
4 more probes, part numbers in the annotation on screen.
 
The following users thanked this post: thm_w, 2N3055, pdenisowski

Offline artur0089

  • Contributor
  • Posts: 28
  • Country: ru
Re: x1 Oscilloscope Probe Shootout
« Reply #33 on: September 25, 2023, 11:33:45 am »
That shows how comparison of measurement equipment, even as simple one as probes calls for lot of scrutiny, preparation and attention to detail to ensure really level playing field.
Why, if everything has been tested and proven a long time ago?
And, most importantly, it complies with the schedule from the manufacturer!
« Last Edit: September 25, 2023, 01:38:16 pm by artur0089 »
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7465
  • Country: hr
Re: x1 Oscilloscope Probe Shootout
« Reply #34 on: September 25, 2023, 11:49:16 am »
That shows how comparison of measurement equipment, even as simple one as probes calls for lot of scrutiny, preparation and attention to detail to ensure really level playing field.
Why, if everything has been tested and proven a long time ago?
And, most importantly, it complies with the schedule from the manufacturer!

So why are you measuring probes if you already know what manufacturer said?. What is the point of the topic ?  ^-^

Take one of those 1x10x probes and experiment with compensation trimmer and see how depending of how probe is set will change response.
With 1x or 10x fixed probe it will also differ which scope are using it for too...
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline gf

  • Super Contributor
  • ***
  • Posts: 1425
  • Country: de
Re: x1 Oscilloscope Probe Shootout
« Reply #35 on: September 25, 2023, 12:07:29 pm »
Keysight distinguishes two different frequency responses of a probe
  • The response with respect to the probe tip voltage as the probe loads the circuit (which is independent of source impedance)
  • The response with respect to the voltage at the probe point as if the probe is not there (assuming 25 Ohm source impedance)
Nobody here seems to consider #2 :-//
(Only rf-loop mentioned "according to normal practice Probes are classified by measuring the 25ohm source impedance" which suggests that he did mean to the latter one.)
 

Offline artur0089

  • Contributor
  • Posts: 28
  • Country: ru
Re: x1 Oscilloscope Probe Shootout
« Reply #36 on: September 25, 2023, 01:34:49 pm »
So why are you measuring probes if you already know what manufacturer said?
Trust but check!
What is the point of the topic ?  ^-^
Bode plot on this Rohde & Schwarz - not everything is clear...
« Last Edit: September 25, 2023, 01:51:49 pm by artur0089 »
 
The following users thanked this post: 2N3055

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7465
  • Country: hr
Re: x1 Oscilloscope Probe Shootout
« Reply #37 on: September 25, 2023, 02:00:19 pm »
Keysight distinguishes two different frequency responses of a probe
  • The response with respect to the probe tip voltage as the probe loads the circuit (which is independent of source impedance)
  • The response with respect to the voltage at the probe point as if the probe is not there (assuming 25 Ohm source impedance)
Nobody here seems to consider #2 :-//
(Only rf-loop mentioned "according to normal practice Probes are classified by measuring the 25ohm source impedance" which suggests that he did mean to the latter one.)

That is something different.

Probe will, by virtue of it's own electrical characteristics, when connected into a circuit, become part of the circuit.
This means that connecting the probe will change voltages, phases, resonant frequencies, currents and whatever in that part of circuit you connected a probe to. That is reason why you cannot measure accurate frequency or sometimes measure at all crystal oscillators etc.. etc.

In some Keysight scopes, they have a software correction mode (Source Estimate) that by presuming 25Ω source impedance it artificially "removes" loading, i.e. it approximates what loading and signal distortion probe caused when connected and tries to approximate how signal looked before. It does that by presuming clean 25Ω resistive source impedance and by knowing what that particular probe loading looks like.

The fancy sounding "True View" is just like normal probe view on every oscilloscope.

This is remotely related to topic here but not relevant. Good read though.

Problem with probes is this:

1. Probes load circuit. By doing that they change real signal in different degrees and levels.
2. Amount of signal "degradation" will depend on signal source impedance and source impedance characteristics.
3. Then we have probe's own transfer function, the measure of how signal on input relates to signal on output, as consequence of it's electrical schematics.
4. Scope input also presents complex schematic but lets pretend that probe output and scope input are already married in probe's electrical schematic to simplify things because probe is supposed to be used that way and this works for us in this case.
5. Combine all of that.

To make possible to have at least some kind of way to compare probes, standardization bodies made standard to test probe BW with 25Ω source impedance.  A terminated 50Ω generator. That allows for some kind of comparison but BW measured that way will not correspond with frequency response of probe when connected to any kind of other circuit.
If you connect probe to high impedance source, your BW will be dominated by RC filter (or RLC if  source is inductive) created from source impedance and shunting capacitance of probe's tip.

One other way to measure probe is with a Bode plot. But that introduces another complication. Namely, Bode plot will plot phase/gain frequency response, but will REMOVE loading to the signal source by the probe. How? Because it will measure delta, difference, between input and output signal as a ratiometric measurement, compensating for changes in input signal.. In effect, it will provide result of measurements as if source impedance of generator was 0Ω. Of course it is not ideal, but as long loading is within dynamic range of inputs it won't care.

Huh? So no more answers but more questions...

So what do we conclude form this? 

One example: audio power amplifiers or power supplies have very low impedance on their outputs for instance. When measuring those, your probe will behave similar to Bode plot (with added benefit that you will also know phase response of the probe).

If you wanted to measure on 600Ω system (audio) you would be prudent to insert impedance converter between your AWG and probe and run a frequency sweep with 600Ω impedance, characterizing it for that use...

I just tried now and with paralleling terminators I dropped source impedance to 12.5Ω. -3dB pint raised from 24MHz to 28 Mhz.
So when measuring some PSU output for ripple or noise, in usual 20MHz BW for that purpose, my simple 60MHz(10x) rated at 6Mhz when 1x will comfortably and accurately perform that task.

Everything else will be anybody's guess...
« Last Edit: September 25, 2023, 06:08:58 pm by 2N3055 »
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: blackdog, Performa01, mawyatt, ozkarah

Online mawyatt

  • Super Contributor
  • ***
  • Posts: 4117
  • Country: us
Re: x1 Oscilloscope Probe Shootout
« Reply #38 on: September 25, 2023, 03:52:01 pm »

Probe will, by virtue of it's own electrical characteristics, when connected into a circuit, become part of the circuit.
This means that connecting the probe will change voltages, phases, resonant frequencies, currents and whatever in that part of circuit you connected a probe to. That is reason why you cannot measure accurate frequency or sometimes measure at all crystal oscillators etc.. etc.

Sorta the Heisenberg Uncertainty Principle in our field, anything and everything you measure is fundamentally "Wrong", how Wrong depends on many factors!! Thus, we must attempt to minimize the things that influence the measurements towards more Wrong!!

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: Performa01, tautech, 2N3055, KungFuJosh

Offline artur0089

  • Contributor
  • Posts: 28
  • Country: ru
Re: x1 Oscilloscope Probe Shootout
« Reply #39 on: September 25, 2023, 07:50:36 pm »
The fewer add-ons, the more truthful the result.
Sweep generator - probe - oscilloscope.
Newfangled things are harmful when they are in large quantities.
Take one of those 1x10x probes and experiment with compensation trimmer and see how depending of how probe is set will change response.
The changes were ±2 MHz.
« Last Edit: September 26, 2023, 02:03:21 am by artur0089 »
 

Offline artur0089

  • Contributor
  • Posts: 28
  • Country: ru
Re: x1 Oscilloscope Probe Shootout
« Reply #40 on: September 26, 2023, 03:43:55 am »
Why didn't Rigol implement Bode in his previous generation instruments, following the example of Siglent?
The question is rhetorical.
« Last Edit: October 28, 2023, 10:26:37 pm by artur0089 »
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7465
  • Country: hr
Re: x1 Oscilloscope Probe Shootout
« Reply #41 on: September 26, 2023, 09:00:43 am »
Why didn't Rigol implement Bode in his previous generation instruments, following the example of Siglent?
The question is rhetorical.


You are wrong, non rhetorically...

They did, and it was a complete joke.

And answer is that Siglent and Rigol seem to have different product strategy.

Respectfully, how is this relevant to the discussion about probes?
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline artur0089

  • Contributor
  • Posts: 28
  • Country: ru
Re: x1 Oscilloscope Probe Shootout
« Reply #42 on: September 26, 2023, 10:52:04 am »
The topic was created based on a video, in which Dave does something that no manufacturer has advertised in their devices.
And what? A complete failure, both in amplitude and phase! Results at frequencies above 15 MHz are unstable and cannot be cloned.
Siglent has a great idea, to use both channels of the generator. I don't have such a kit and can't check it.
All that remains is to regret the “support” of the software from Rigol.
« Last Edit: September 28, 2023, 12:05:56 am by artur0089 »
 

Offline artur0089

  • Contributor
  • Posts: 28
  • Country: ru
Re: x1 Oscilloscope Probe Shootout
« Reply #43 on: September 27, 2023, 05:11:48 am »
By the way, if anyone wants to make a Bode plot using the Siglent kit, using two generator channels.
The Rigol DG1000Z comes with only one 50 Ohm coaxial cable.
But, “an old warrior is a wise warrior”...
Immediately, I bought a pair of identical cables on AliExpress. Above, a comparison at 40 MHz, with them. The phases coincided at 0 degrees.
Below is a comparison of the original one (yellow) and the purchased one (blue). The difference in length is 10 mm, the original one is shorter. Length 1120 mm.
Which one is more correct?
Up to 15 MHz the difference is unnoticeable.
« Last Edit: September 28, 2023, 12:06:14 am by artur0089 »
 

Offline artur0089

  • Contributor
  • Posts: 28
  • Country: ru
Re: x1 Oscilloscope Probe Shootout
« Reply #44 on: October 25, 2023, 10:53:43 pm »
Manually, from two generator channels.
At each frequency, I calibrated the phase and signal level.
« Last Edit: October 28, 2023, 10:31:09 pm by artur0089 »
 

Offline ballsystemlord

  • Regular Contributor
  • *
  • Posts: 248
  • Country: us
  • Student
Re: x1 Oscilloscope Probe Shootout
« Reply #45 on: October 26, 2023, 02:41:05 am »
Nice video...but...but... absolutely essential informations are missing, so evaluating whether the received data is bullshit or whether it even has any value other than entertainment.

Or is this some kind of "engineer joke".

How you have connected signals (exactly) and how you have setup BodePlot. Is it displaying/measuring Vin/Vout or Vout.. (I do not know how these settings are in R&S

Think example possible VSWR effect. If Bode measure Probe in and then compare it to Ref In (after this coaxial (mismatch to 1M input) and after this other mismatch done with  T split to probe input.
It has also been completely forgotten that according to normal practice Probes are classified by measuring the 25ohm source impedance.

So nearly all possible mistakes have been made carefully.

Look your generator output where is connected Probe input using T  for split it to Probe and then to coaxial what goes to Bode Reference.
Now if Bode is on Vin/Vout mode signal in probe input is totally other than signal in reference input point... except when signal is very low freq or DC.

Also  @artur0089 test may give some reason for thinking...

This is bit alarming in this video. Is this really engineer made test setup what is partially visible here:

Is there a tutorial on how to do this test properly?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf