Scope channel Noise difference?

[T3sl4co1l]

What happens if you keep pressing the one-shot button?

Create a table:
Trial ChA ChB
1   5.5mV 6mV
2   ?      ?
etc.

You're doing noise, so you have to do statistics!  If there is a clear trend between these data series, then one may well be noisier.  So far, your two data points ("looks like 5.5, 6" and single shot "1.8, 1.56") aren't suggesting much of a trend.

The scope tells you it's reading some amount of mV, but that's a convenience only; ultimately, it's only measuring what appears at the sampling circuit.  Which is the sum of internal noise (the analog to digital converter isn't perfect, and makes mistakes), buffering and amplification (random thermal noise in the circuits, amplified by a lot because the converter itself probably senses ~1V signals), and what's actually at the input connector itself (which isn't much compared to the rest, even if terminated in 50 ohms and nothing else).  And when it performs a sample operation, it simply pulls that number out of the random cloud: the expected value of a given sample is simply the amplitude of the noise, period.  So from any given series of samples, it's simply going to be... random, no matter how fast or slow you sample at.

The noise amplitude depends on bandwidth, however.  If you have the scope configured for 100MHz (or more?) bandwidth, and the noise spectrum is flat, then reducing it to 20MHz BW will yield sqrt(100/20) ~= 2.2 times less noise.  The reduction is even more significant if you set an even lower bandwidth: there are probably settings for digital filtering (as a percentage of displayed rate), or "high res mode", or averaging mode (with averaging, the expected reduction for un-triggered components is sqrt(1/N), so for N=100, the noise goes down by 10x).

Tim

[dom0]

Interesting that Rigol tries to cheat with GND coupling. How dumb do they think their customers are?

I thought it was a assumption? Did you look look at the schematics?

You'd have to go a faaaaaaaar way to create a system with 300 MHz bandwidth and that sensitivity that has less than 1/2 LSB noise, which is the only way you'd see only zero codes on the scope. Or they zero them digitally.

[Wmacky]

Standby  - Statistical measurements coming up in 5 minutes!

[Wmacky]

OK here's a 2 minute statical measurement.  ( I'm learning alot tonight about using this scope!!!! )

Look at those averaged numbers.  (bottom left)  This newbie thinks that not good. 

I tried shorting the inputs, but I need terminators, the wire isn't working so well. I've moved the unit all around the room, and still channel 1 is higher.   I'm pretty new to this stuff, but my gut tells me it's internal noise.........  What to do................

[rs20]

All this fuss over a 300uV difference? Geez, I thought you were talking about a significant difference. <loses all interest in this thread>

[cyr]

Lots of variability between these single shot numbers. The pics I'm posting are with DC coupling. The very first numbers were with AC coupling.  Bigger channel difference with DC. Notice the central spike on channel 1 that's always there. It moves with the  horizontal position control, and only exists on DC coupling?

Another

That spike is probably there because you are triggering on it. That may explain part of the difference you are seeing as well, check what happens if you trigger on channel 2 instead or move the trigger level way above the noise so the scope runs untriggered.

In any case, nothing worth worrying about.

[Fungus]

Try moving it around the room.

Electrical noise depends on all sorts of things, including position/orientation.

But really ... unless the difference between the channels is huge then you're most likely worrying over nothing - the channel with more 'noise' might actually be the best as it has more bandwidth!   

Some noise is normal - see Dave's video:

Edit: Then again, Rigols have shielding so the noise is probably coming from inside. One channel's input circuitry is probably closer to some oscillator than the other one.

[tautech]

All this fuss over a 300uV difference? Geez, I thought you were talking about a significant difference. <loses all interest in this thread>

+1
This level if measurement will rarely be used in practice and the noise you see could also be from your enviroment.
Try another location.

Have you watched these?
http://www.eevblog.com/2014/04/10/eevblog-601-why-digital-oscilloscopes-appear-noisy/

http://www.eevblog.com/2014/04/27/eevblog-610-why-digital-scopes-appear-noisy-part-2/


I might add if you had a $20K scope you might have something to grizzle about.

DSO's show all sorts of noise unterminated, especially when set to Peak detect.

If it is within Manufacturers specs....it is what it is.
That is all that you should be concerned with.

[ebastler]

Interesting that Rigol tries to cheat with GND coupling. How dumb do they think their customers are?

On my DS1054Z, there clearly is some remaining "real world" noise when the input mode is set to GND. No cheating there. Maybe someone could verify the statement that GND produces a flat line on the DS2000 scopes?

[JohnnyBerg]

On my DS1054Z, there clearly is some remaining "real world" noise when the input mode is set to GND. No cheating there.

Here to. Depending on the timebase setting: 80µV pp @ 2µS/div and  440µV pp @ 200nS/div

[tautech]

Interesting that Rigol tries to cheat with GND coupling. How dumb do they think their customers are?

On my DS1054Z, there clearly is some remaining "real world" noise when the input mode is set to GND. No cheating there. Maybe someone could verify the statement that GND produces a flat line on the DS2000 scopes?

Normal.
They are powered with a SMPS, they have a SMPS for the display LCD backlight.

All the shielding in the world can't prevent some of this EMI from entering the scope circuitry in some small way.
How you use the scope and manage known EMI is important.
So noise originating from SMPS within could be in the 30-50 KHz range could be from the PSU.
On the main PCB there will also be a number of oscillators ranging from 5 to maybe 100 MHz.

It's noise and only noise, understand that, live with it and manage it. 

[ebastler]

Interesting that Rigol tries to cheat with GND coupling. How dumb do they think their customers are?

On my DS1054Z, there clearly is some remaining "real world" noise when the input mode is set to GND. No cheating there. Maybe someone could verify the statement that GND produces a flat line on the DS2000 scopes?

Normal. [...]
It's noise and only noise, understand that, live with it and manage it. 

Sure, we are on the same page here. I do not expect a perfect "flat line" Quite to the contrary, what I said was that I find it re-assuring to see some noise. It indicates that the scope actually GNDs its analog input and keeps the AD conversion active, rather than "simulating" a grounded input by pushing out digital zeros.

[dom0]

Btw. I didn't find any specification from Rigol regarding noise of these scopes. They saw "low noise bla bla", but no numbers. (Batronix has the data sheets, rigolna.com wants personal data, nayyy)

[Fungus]

Btw. I didn't find any specification from Rigol regarding noise of these scopes. They saw "low noise bla bla", but no numbers.

That's because it depends on an awful lot of factors which aren't under Rigol's control.

OTOH Rigol did give you an 'analog display' function which can do a decent job of showing the true signal (see Dave's "why digital 'scopes appear noisy" video for explanation).

[dom0]

Btw. I didn't find any specification from Rigol regarding noise of these scopes. They saw "low noise bla bla", but no numbers.

That's because it depends on an awful lot of factors which aren't under Rigol's control.

That's the case with many specifications. It's not like you can't specify it, they just didn't.

OTOH Rigol did give you an 'analog display' function which can do a decent job of showing the true signal (see Dave's "why digital 'scopes appear noisy" video for explanation).

Well... I have my issues with the nomenclature of a "true" signal, but I get your point. And indeed, averaging is a useful feature to extract a repetitive waveform with high uncorrelated (noise) contents. But then again I mostly care for how the waveform looks on the wire, with all the nice noise on top.

[dom0]

w/o saying anything about bandwidth all noise figures are meaningless. Also it's not only about the converter and it's q. and excess noise but also about the input PGA.

[Rupunzell]

Depends on what one is working on. For most of the common low speed digital stuff and many electronic things in general, that difference of 300 uV noise between channels might not make any difference at all. Still, that difference in noise between channels is indicative of the design, production and parts used. If the designers were very careful and highly skilled with high quality parts used and equally high standards of production, both channels would have identical performance say within 5% or better. If moving the instrument around alter's it response external fields, there are internal shielding and related sensitivity problems that should have been addressed during it's design. If the instrument or other electronic device is dumping HF energy into it's environment, it can affect the performance of other electronic devices near it.

For the Analog folks doing low noise high performance design (less than one nanovolt/root-Hz at the input), that 300uV difference between channels would make that instrument mostly useless. The Tek 1A7A, 7A22 plug-in have 10uV per/division sensitivity and the noise is much less than one division. Granted these have a 1Mhz BW, yet if one puts two 7A22s into a Tek 7000 series mainframe, their noise displayed would be nearly equal, certainly within 5-10% of each other if they meet spec.

Another example would be the Tek 7A13, 100 Mhz differential input-voltage comparator plug-in, the displayed noise spec is 400uV max.


Bernice

It's noise and only noise, understand that, live with it and manage it. 

[Fungus]

w/o saying anything about bandwidth all noise figures are meaningless. Also it's not only about the converter and it's q. and excess noise but also about the input PGA.

It's also a lot about the probes plus whatever it is you're connecting them to at the time. I'd say the ringing and other distortions caused by the probes is far more important than a bit of noise in the uV range.

[Wmacky]

I have to say, I'm not feeling that I yet have a definitive answer as to if I have a problem or not.

The issue is that I have 1 day in my easy return window to determine if this scope is working normally or not.  The answers seem to have ranged from, the channels should match to it's doesn't matter, and  it doesn't matter but the overall noise level is too high.  As mention there are no noise stats for the scope The only reference I have is a video by Dave testing a 2202, and the noise appears well matched between  channels, and he commented that the unit had low noise.  This with no terminators and which I will nopt be able to obtain until after my window closes.

Some mentioned bandwidth.  I did mention earlier that the bandwidth is 300 MHZ, and the pics, stats are from the max BW.

Below are 2 additional pics limited to 100 MHZ.  And 20 Mhz respectively.

As you can see, the noise level drops as expected, but the 1st channel consistently reads high.

[Wmacky]

So I guess it comes down to this.

Do I just enjoy the new scope being assured it's working just as it should, or do I call the vendor Monday morning and tell them there is a problem. So far the responses to to lean both ways?  I'm not sure what other data I can offer? Ideas?

EDIT:   I was able to properly short the input to ground using a small wire.    I also moved the scope to different locations. No difference seen.   What I'm seeing appears internal?


Anyone out there with a 2072, 2202, or a 2302 seeing the same noise discrepancy between channels, or noise levels?   Perhaps using  band limiting to level the field between the different BW models?

[wraper]

I have to say, I'm not feeling that I yet have a definitive answer as to if I have a problem or not.

The issue is that I have 1 day in my easy return window to determine if this scope is working normally or not.  The answers seem to have ranged from, the channels should match to it's doesn't matter, and  it doesn't matter but the overall noise level is too high.  As mention there are no noise stats for the scope The only reference I have is a video by Dave testing a 2202, and the noise appears well matched between  channels, and he commented that the unit had low noise.  This with no terminators and which I will nopt be able to obtain until after my window closes.

Some mentioned bandwidth.  I did mention earlier that the bandwidth is 300 MHZ, and the pics, stats are from the max MW.

Below are 2 additional pics limited to 100 MHZ.  And 20 Mhz respectively.

As you can see, the noise level drops as expected, but the 1st channel consistently reads high.

So did you try to set channel 2 as trigger source as others suggested? I don't see that you mentioned doing this.
BTW on my 200 MHz unlocked DS2072 which have almost the same front end noise is higher than yours.

[Wmacky]

I did change the trigger.  The spike went away, but all  channel noise level difference remained.

Also mine is actually a 2072A unlocked to 300 Mhz.  Interesting that your noise is higher. How matched are the channels regarding noise?   The channel difference was of more concern than overall noise, but I glad to see the total level is not out of the norm for the scope. That's good!

Guy's I know 300uV isn't much, My concern was just that one channel consistently reading different could indicate a issue causing other concerns?  If all is well then great. I just don't have the experience to make the call, which is why I'm bending the ear of my well experienced forum mates! 

[dom0]

For LF you can easily test if the channels match well with a Y-like cable.
(For RF you'd want a power splitter to do that.)

[Fungus]

Have you done a calibration? Maybe it will go away...   

[OldNeurons]

So I guess it comes down to this.

Do I just enjoy the new scope being assured it's working just as it should, or do I call the vendor Monday morning and tell them there is a problem. So far the responses to to lean both ways?  I'm not sure what other data I can offer? Ideas?

EDIT:   I was able to properly short the input to ground using a small wire.    I also moved the scope to different locations. No difference seen.   What I'm seeing appears internal?


Anyone out there with a 2072, 2202, or a 2302 seeing the same noise discrepancy between channels, or noise levels?   Perhaps using  band limiting to level the field between the different BW models?

Here are my results at 500µV and 1mV, 300 / 100 / 20 MHz bandwith.
Rigol DS2102A Unlocked (all options, 300MHz)
Hope this will help you.
My scope also shows a bit more noise on CH 1 but I really don't care.