Products > Test Equipment
Rigol DHO800 memory depth / bandwidth upgrades and noise level!
ebastler:
--- Quote from: JDW on January 17, 2024, 04:25:57 am ---When I try to repeat the same test using the same settings, even when testing only a single channel (leaving the probe disconnected as Fungus apparently did), I get 60Hz noise that makes it impossible to detect the true noise floor. Noise remained unchanged with or without the ground cable attached to wall socket Earth Ground.
--- End quote ---
But your probe is not disconnected! Your photo shows that it is connected to the scope, and its tip is left "floating", not connected to any defined potential. It is normal that the probe essentially acts as an antenna then, picking up all kinds of noise from the environment.
I don't know how Fungus had configured the inputs for his earlier measurements, but it am pretty sure it was not the configuration you show:
* When people refer to "open inputs", they mean that nothing at all is connected to the scope's BNC jack.
* To really measure the internal noise of the scope, the inputs should be shorted directly at the BNC jack, or a 50 Ohm termination installed there.
* If you want to measure the background you can expext in a "real world" scenario, you can connect the probe to the scope, and then connect the probe's little ground strap to the tip. Even that little loop can pick up magnetic fields from the environment; we had reports from one user who could receive nearby FM radio stations that way.
JDW:
--- Quote from: ebastler on January 17, 2024, 08:20:27 am ---
--- Quote from: JDW on January 17, 2024, 04:25:57 am ---When I try to repeat the same test using the same settings, even when testing only a single channel (leaving the probe disconnected as Fungus apparently did), I get 60Hz noise that makes it impossible to detect the true noise floor. Noise remained unchanged with or without the ground cable attached to wall socket Earth Ground.
--- End quote ---
But your probe is not disconnected! Your photo shows that it is connected to the scope, and its tip is left "floating", not connected to any defined potential. It is normal that the probe essentially acts as an antenna then, picking up all kinds of noise from the environment.
I don't know how Fungus had configured the inputs for his earlier measurements, but it am pretty sure it was not the configuration you show:
* When people refer to "open inputs", they mean that nothing at all is connected to the scope's BNC jack.
* To really measure the internal noise of the scope, the inputs should be shorted directly at the BNC jack, or a 50 Ohm termination installed there.
* If you want to measure the background you can expext in a "real world" scenario, you can connect the probe to the scope, and then connect the probe's little ground strap to the tip. Even that little loop can pick up magnetic fields from the environment; we had reports from one user who could receive nearby FM radio stations that way.
--- End quote ---
Thank you for explaining. I choose to be more clear in my own personal wording in that if I mean for somebody not to connect any probe at all, I will tell them "no probe attached." That's 100% clear, with zero guesswork for veterans or newbies alike. Words like "disconnected" can mean multiple different things. Even "open input" isn't necessarily clear if you've not learned the meaning. "No probe attached" is just that.
I don't have a 50Ω terminator, and I am basically trying to replicate what Fungus did because that is established data we can go by. No, his work is not a gold standard, but it is a nice frame of reference that other people can try to see if we get the same thing.
Currently, my scope's "Probe" settings for each channel are set to 10x, and that's because I normally use my Probes mechanically switched to the 10x setting. But we're doing this particular test with "no probe attached" so...
* When displaying only CH1 and with CH1 set to Probe: 10X: The "Current" column shows about 600uV (Fungus was about 200uV)
* When displaying only CH1 and with CH1 set to Probe: 1X: The "Current" column shows about 60uV (Fungus was about 200uV)
I assume Fungus must have used 1x or 10x, so in either of those cases, my numbers are not close to his.
When I attach my probe to CH1 and mechanically set its switch to 10x, then set the scope's CH1 setting to Probe: 10x too, and then connect the probe's GND to the probe's tip (as per your bullet point 3): The "Current" column shows about 476uV (Fungus was about 200uV) But you will likely explain that as being a matter of the ground strap picking up noise and radio stations.
I then tried your bullet point #2, shorting the inner gold pin of the CH1 BNC to its ground using the shortest wire possible. I get a "Current" column value of 256uV, which is higher than the values Fungus showed, but we don't know exactly how he measured either. Even so, it least this test was in the ballpark of his. So I am largely satisfied with that result.
Thank you for your kind assistance.
2N3055:
It depends on what are you measuring.
Putting a 50Ω on 1MΩ input "simulates" what noise would be if input were 50 Ω.
Putting a shielding cover over BNC better "simulates" noise at 1MΩ input impedance.
Setting probe ratio to 1:1 will measure "native" noise.
Setting probe to 10x (which is really 1/10 but that is another story) will measure probe tip referred noise, i.e. what is your baseline noise floor when using 10x probe.
A probe "simulator" could be made, with low noise 9 MΩ resistor to ground, in a nice shielded box, connected to input BNC directly. That would simulate connected probe with grounded tip, with no interference. WIth scope set to 10x it would best simulate what a baseline nose would be at 1 MΩ with the ideal 10x probe.
Like I said before, just measuring something without establishing clear protocols and rules ends up with bunch of measurements that means very little because they cannot be directly compared and have no correlation any real world scenario.
And unlike BW measurements that need expensive and calibrated signal generators, some things require little investment, but some rigor and attention and some set rules.
ebastler:
--- Quote from: JDW on January 17, 2024, 08:50:37 am ---
* When displaying only CH1 and with CH1 set to Probe: 10X: The "Current" column shows about 600uV (Fungus was about 200uV)
* When displaying only CH1 and with CH1 set to Probe: 1X: The "Current" column shows about 60uV (Fungus was about 200uV)I assume Fungus must have used 1x or 10x, so in either of those cases, my numbers are not close to his.
--- End quote ---
When you are interested in the noise generated internally in the scope, you should always use 1x as the probe setting. In the 10x setting the scope will "assume" that the signal it sees has been divided by 10 by an external probe. So e.g. 60 µV of noise from its internal input amplifiers will be reported as 600 µV, which is misleading.
If you don't have a termination resistor handy, leaving the BNC input unconnected but shielded is probably the most meaningful test you can do. As 2N3055 explained, that will give you the noise generated by the scope's input section when set to 1 MOhm (i.e. ready for an external probe), but without the "antenna" contributions from an actual probe. You can try and make some improvised shielding from copper or aluminim foil around the BNC jack.
When comparing noise figures, you will also notice that they depend on the scope's time base setting and sampling rate. Going to slower time bases lets the scope capture the low-frequency components of the noise -- which are significant, since there are noise cmponents where the amplitude follows a 1/f curve. Increasing the sampling rate (by increasing the memory depth used, in the Acquisition menu) lets the scope capture higher-frequency contributions, which also add more noise. So when comparing to measurements made by others, be sure to use the same acquisition settings.
Aldo22:
Interesting topic (noise measurement).
I only have a Hantek at the moment, but the problem will probably remain the same.
Without a checklist, you always forget a setting.
- Acquisition Mode: Normal - Average - Peak - HR give different results.
- 20MHz bandwidth limitation
- Probe attenuation settings
- V/div and s/div
- Input coupling GND
- With the Hantek, the sampling rate also makes a difference.
Have I forgotten something?
All these settings give me different noise values. Is it the same with the Rigol?
The "calibration kit" (Open, Short, Load) of the NanoVNA is quite handy for testing with a BNC-SMA adapter (Attachment) ...and cheap.. ;)
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version