Author Topic: Rigol MSO5000 HiRes mode  (Read 10724 times)

0 Members and 1 Guest are viewing this topic.

Offline gf

  • Super Contributor
  • ***
  • Posts: 1425
  • Country: de
Re: Rigol MSO5000 HiRes mode
« Reply #50 on: October 30, 2022, 12:25:03 pm »
Interestingly, in the last few screenshots the reported standard deviation is even lower in normal mode than with hires! :-//

These two are very interesting. There should be a sharp dropoff in noise at the high end with Hires enabled.
All I'm seeing is a slight sag in the middle.

The 1GSa/s screenshot is the only one which shows a full 0...fs/2 spectrum, and the difference between normal and hires definitely does not look like the sinc frequency response of a simple moving average filter. Looks like hires were possibly implemented with a different kind of filter.
 

Offline Anding

  • Contributor
  • Posts: 46
  • Country: hk
Re: Rigol MSO5000 HiRes mode
« Reply #51 on: October 30, 2022, 12:29:36 pm »
20ns / division

" alt="" class="bbc_img" />
" alt="" class="bbc_img" />
 
The following users thanked this post: Fungus

Offline FungusTopic starter

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Rigol MSO5000 HiRes mode
« Reply #52 on: October 30, 2022, 12:32:18 pm »
if the noise bandwidth is way below this rate, say few KHz, then hope to destroy this noise in averaging (hi-res mode) is almost nill, ymmv.

If the noise is in the ADC then it's more likely to have extremely high bandwidth than be low frequency (in my unprofessional opinion).

In that case all the stuff we're seeing down low in the FFT would be aliasing of that VHF noise and hires mode would work perfectly.
« Last Edit: October 30, 2022, 12:35:29 pm by Fungus »
 

Offline FungusTopic starter

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Rigol MSO5000 HiRes mode
« Reply #53 on: October 30, 2022, 12:36:01 pm »
Looks like hires were possibly implemented with a different kind of filter.

A filter that does almost nothing.
 

Offline switchabl

  • Frequent Contributor
  • **
  • Posts: 445
  • Country: de
Re: Rigol MSO5000 HiRes mode
« Reply #54 on: October 30, 2022, 12:58:11 pm »
Initially I'd only consider a timebase with the maximum sample rate.
In subsequent steps, slower timebases can be checked too.

If it works in a similar way to lower-end Keysight models then it will only average "extraneous" samples, so at full sample rate, high resolution should not do anything (new EXR/MXR models are different). I don't have a Rigol anymore unfortunately but as there is some common heritage and no explicit setting it would not surprise me.
 
The following users thanked this post: egonotto

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11714
  • Country: my
  • reassessing directives...
Re: Rigol MSO5000 HiRes mode
« Reply #55 on: October 30, 2022, 01:15:47 pm »
if the noise bandwidth is way below this rate, say few KHz, then hope to destroy this noise in averaging (hi-res mode) is almost nill, ymmv.

If the noise is in the ADC then it's more likely to have extremely high bandwidth than be low frequency (in my unprofessional opinion).

In that case all the stuff we're seeing down low in the FFT would be aliasing of that VHF noise and hires mode would work perfectly.

aliasing is subjective (controversial) issue as its non-provable from DSO display (any noise/signal components greater than sample rate / 2) and since we believe they exist, so somehow they will affect measurement on the display. but then on FFT matter, from Anding post we can see reduction in magnitudes at higher frequency components (>300MHz?) means hi-res is effective at reducing those noise components, but then lower components (frequencies) noises are not affected and manifest themselves when larger time/div is selected (1us/div in Anding's earlier post), which lead to balnazzar's confusion, i believe. ymmv.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 
The following users thanked this post: balnazzar

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7462
  • Country: hr
Re: Rigol MSO5000 HiRes mode
« Reply #56 on: October 30, 2022, 01:25:10 pm »
Some nice work here and kudos to Fungus for effort to sort out this question.
I don't have MSO5000 so cannot contribute in that way, but would like to point out this excellent document by Tektronix on the topic.

I deliberately didn't want to bring up any materials by LeCroy because they use different implementation (ERES) that is not what Rigol uses.
That would only bring confusion.

Tektronix Hires explanations explain well how it should work. Page 11.

Just a note: bits stated in document are theoretical. 8 bit scopes don't have ENOB (effective number of bits) of 8 bits but less. But improvements (12 bit is 4 bits more than 8 bits, so 4 bits of improvement) should be reasonably followed.
Also Tektronix says they use fixed point 16bit math in acquistion engine. We don't know anything about how Rigol did it. Claim by Rigol of max 12bits, means max 4 bits of improvement algorithm was implemented.



« Last Edit: October 30, 2022, 03:25:44 pm by 2N3055 »
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: egonotto, mawyatt

Offline FungusTopic starter

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Rigol MSO5000 HiRes mode
« Reply #57 on: October 30, 2022, 01:32:14 pm »
its non-provable from DSO display

Yep. Only Rigol engineers know for sure.

Conspiracy theory:

Maybe this is why the MSO5000 didn't have Hires mode when it was launched - they couldn't get it to work because of the noise spectrum  characteristics.

Maybe they only added it after a manager demanded it be put in the menu, and that this was the best they could do.
 

Offline FungusTopic starter

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Rigol MSO5000 HiRes mode
« Reply #58 on: October 30, 2022, 01:39:50 pm »
Conspiracy theory #2:

Even if conspiracy #1 is true, Hires mode still should work with certain combinations of sample rate and memory size, right...?

Maybe the filter the engineers came up with was carefully designed not to give visibly different results with different settings. People would notice that and complain.   :-X
 

Offline skander36

  • Frequent Contributor
  • **
  • Posts: 897
  • Country: ro
Re: Rigol MSO5000 HiRes mode
« Reply #59 on: October 30, 2022, 03:12:35 pm »
... Hires mode still should work with certain combinations of sample rate and memory size, right...?


On Rigol 5k High Res effect is a little better visible on MSO's LCD display than on pics saved from screen. If you zoom in you can see better but not as clear as on other scopes.
 

Offline FungusTopic starter

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Rigol MSO5000 HiRes mode
« Reply #60 on: October 30, 2022, 03:15:30 pm »
It seems to gain about 1 more bit of ADC resolution.
 

Offline gf

  • Super Contributor
  • ***
  • Posts: 1425
  • Country: de
Re: Rigol MSO5000 HiRes mode
« Reply #61 on: October 30, 2022, 03:20:51 pm »
Just a note: bits stated in document are theoretical. 8 bit scopes don't have ENOB (effective number of bits) of 8 bits but less. But improvements (12 bit is 4 bits more than 8 bits, so 4 bits of improvement) should be reasonably followed.

Good point. It is of course important to distinguish enhancement of numerical precision, and ENOB enhancement. A 16-tap moving average enables 4 additional bits of numerical precision, but ENOB enhanecment is at most 2 bits.

... Hires mode still should work with certain combinations of sample rate and memory size, right...?


On Rigol 5k High Res effect is a little better visible on MSO's LCD display than on pics saved from screen. If you zoom in you can see better but not as clear as on other scopes.

in order not to rely on visual assessment, I had asked for standard deviation measurements.
But these numbers aren't plausible either, when normal mode shows a lower number than hires. :-//
I wonder, whether they are possibly taken from the screen, too, or from the the "real" captured data?
« Last Edit: October 30, 2022, 03:24:31 pm by gf »
 

Offline balnazzar

  • Frequent Contributor
  • **
  • Posts: 417
  • Country: it
Re: Rigol MSO5000 HiRes mode
« Reply #62 on: October 30, 2022, 03:29:37 pm »
I'm not understanding why hires seems to do nothing at mitigating the scope's self noise...
its a bell and whistle after the signal path, like averaging... so take it as what it is... https://www.electronicdesign.com/technologies/test-measurement/article/21801317/why-you-should-care-about-oscilloscope-acquisition-modes if point to point averaging takes, say 8 points averaging to do 1 hi-res point (hires is someone hiring someone), at 8GSa/s, i read it as equivalent 1GSa/s of sampling rate, if the noise bandwidth is way below this rate, say few KHz, then hope to destroy this noise in averaging (hi-res mode) is almost nill, ymmv.
That is a good observation. Well spotted! Basically 4Gs/s or even 8Gs/s is complete, total and utter overkill for a 70MHz oscilloscope. Heck, it would be better to implement a decimation  algorithm that turns 16 samples into 1 sample. That would give 2 extra bits of ADC resolution. There is enough noise to make that work like a charm.

Not so overkill with 350 MHz and 4 channels simultaneously active. But I get your point.
 

Offline balnazzar

  • Frequent Contributor
  • **
  • Posts: 417
  • Country: it
Re: Rigol MSO5000 HiRes mode
« Reply #63 on: October 30, 2022, 03:34:36 pm »
Mh, the ED's article and Tek's doc are good stuff, especially for a beginner like me. Thanks...

Would you tell me again where to find the difference about Lecroy ERES vs common High Resolution?
« Last Edit: October 30, 2022, 03:39:22 pm by balnazzar »
 

Offline skander36

  • Frequent Contributor
  • **
  • Posts: 897
  • Country: ro
Re: Rigol MSO5000 HiRes mode
« Reply #64 on: October 30, 2022, 03:52:24 pm »

... in order not to rely on visual assessment, I had asked for standard deviation measurements.

I wonder, whether they are possibly taken from the screen, too, or from the the "real" captured data?
Some data from excel chart + csv along with stdev.
 
The following users thanked this post: egonotto

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7462
  • Country: hr
Re: Rigol MSO5000 HiRes mode
« Reply #65 on: October 30, 2022, 04:11:59 pm »
Mh, the ED's article and Tek's doc are good stuff, especially for a beginner like me. Thanks...

Would you tell me again where to find the difference about Lecroy ERES vs common High Resolution?

Google it.. It is all public access.. there a few documents on LeCroy
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: balnazzar

Online mawyatt

  • Super Contributor
  • ***
  • Posts: 4117
  • Country: us
Re: Rigol MSO5000 HiRes mode
« Reply #66 on: October 30, 2022, 04:16:08 pm »
Think the important metric regarding HiRes/ERES and other modes of enhancement, however they are implemented, is the ENOB after the mode is invoked which takes into account all the degrading/limiting factors, wether in the analog channel, ADC, or post processing numerical limits.

Things like input referred noise level and complete end to end channel linearity become issues that differentiate various systems within a given bandwidth. Various techniques are available to study such, one we often use/prefer is the classic Two Tone IMD method.

Another important factor, often overlooked, is the channel overload recovery. Where one is "looking" at a tiny signal just after a much larger part of signal that can saturate the channel. This can cause lots of issues with resolution enhancement techniques, since these may employ some form of waveform averaging. Here is where a true higher resolution ADC shines as you can "see" the tiny signal detail with the higher resolution ADC and remain still within the channel limits without resorting to post ADC resolution enhancement techniques. Of course assuming the channel noise level is low enough to allow one to "see" the tiny waveform details, which implies a basic good fundamental ENOB for the channel and ADC without resorting to enhancement techniques.

Anyway, lots of interesting metrics for these new DSOs , which in reality are really Impressive Data Acquisition Systems disguised as simply a Scope  ;D

Best, 
« Last Edit: October 30, 2022, 05:41:43 pm by mawyatt »
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: balnazzar

Offline Martin72

  • Super Contributor
  • ***
  • Posts: 7019
  • Country: de
  • Testfield Technician
Re: Rigol MSO5000 HiRes mode
« Reply #67 on: October 30, 2022, 05:22:52 pm »
Would you tell me again where to find the difference about Lecroy ERES vs common High Resolution?

https://teledynelecroy.com/doc/differences-between-eres-and-hires

 
The following users thanked this post: balnazzar

Offline FungusTopic starter

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Rigol MSO5000 HiRes mode
« Reply #68 on: October 30, 2022, 06:01:21 pm »
Would you tell me again where to find the difference about Lecroy ERES vs common High Resolution?

https://teledynelecroy.com/doc/differences-between-eres-and-hires

So: Gaussian filter (eres) vs. Boxcar filter (hires)
 
The following users thanked this post: balnazzar

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7462
  • Country: hr
Re: Rigol MSO5000 HiRes mode
« Reply #69 on: October 30, 2022, 06:59:40 pm »
Would you tell me again where to find the difference about Lecroy ERES vs common High Resolution?

https://teledynelecroy.com/doc/differences-between-eres-and-hires

So: Gaussian filter (eres) vs. Boxcar filter (hires)

Yes, Something like that.
Technically ERES is linear phase FIR filter with Gaussian shaped convolution kernel, vs. Hires plain boxcar filter.
Difference is that ERES will have better pulse response and less BW reduction and keep number of points i.e. there is no downconversion. Also if ERES is used as a math function after the fact (on scopes that support that) you have benefit you can tweak it your hearths will because you have original data all the time.
Hires usually also yields less data points on output than input, I.E. it downcoverts to resulting lower sample rate by filtering. Note that it doesn't have to downconvert (it could just lose some data at end and begging) but it is not usually implemented that way. It is usually acquisition mode that means you don't have original data. That might have a meaning to user or not, depending of what they are doing with it.
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: balnazzar

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11714
  • Country: my
  • reassessing directives...
Re: Rigol MSO5000 HiRes mode
« Reply #70 on: October 30, 2022, 09:21:12 pm »
Hires usually also yields less data points on output than input, I.E. it downcoverts to resulting lower sample rate by filtering. Note that it doesn't have to downconvert (it could just lose some data at end and begging) but it is not usually implemented that way. It is usually acquisition mode that means you don't have original data. That might have a meaning to user or not, depending of what they are doing with it.
out of curiosity, i checked hi-res mode of my ds1054z (i dont usually use this mode), set display type to dots, sin(x)/x off... at smallest 5ns/div, i can still see dots are in accordance to 1GSa/s, no BW reduction when i check with one of my fastest pulser (Sync output of UTG962 AWG) reading about ~4ns risetime. textbook averaging should show some BW degradation and reduced sample points. on larger time/div (µs-ms/div), hi-res does improve displayed noise on all range setting. so i'm guessing there is trade-secret trick going on in FW (if not in FPGA) that doing hi-res at different sample rate depending on the selected time/div. and it maybe disabled when smallest time/div is selected. i guess what Anding posted earlier looks like MSO5000 is doing "strictly from textbook" boxcar or running average... ymmv.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline gf

  • Super Contributor
  • ***
  • Posts: 1425
  • Country: de
Re: Rigol MSO5000 HiRes mode
« Reply #71 on: October 30, 2022, 10:15:12 pm »
i guess what Anding posted earlier looks like MSO5000 is doing "strictly from textbook" boxcar or running average... ymmv.

This was my first guess as well, but here's evidence that the filter is a different one:
https://www.eevblog.com/forum/testgear/rigol-mso5000-hires-mode/msg4492600/#msg4492600

If it were a boxcar moving average, then the main and side lobes of the corresponding sinc frequency response would be clearly visible in the hires spectrum. But there are no lobes at all in the spectrum.

[ Exponential moving average (IIR) might be a suitable candidate as it has a roughly similar frequency response shape with a fairly flat stop-band and only modest stop-band rejection - but that's just speculation. ]
 

Offline Martin72

  • Super Contributor
  • ***
  • Posts: 7019
  • Country: de
  • Testfield Technician
Re: Rigol MSO5000 HiRes mode
« Reply #72 on: October 30, 2022, 11:18:30 pm »
Quote from: Rigol Manual
Note:
 The "Average" and "High Res" modes use different averaging methods. The
former uses "Multi-sample Average" and the latter uses "Single-sample
Average".
 In "High Res" mode, the oscilloscope improves the measurement accuracy at
the cost of bandwidth. Each time the sampling rate changes, a window
displaying the current bandwidth appears at the lower-left corner of the screen.

 In "High Res" mode, the highest waveform refresh rate mode is not supported.

At all "hi res" pics posted here, I can´t discover such a window...

Offline Anding

  • Contributor
  • Posts: 46
  • Country: hk
Re: Rigol MSO5000 HiRes mode
« Reply #73 on: October 30, 2022, 11:27:51 pm »
It flashes up in a notification window, usually “-3db at 100MHz”, but is on screen only a second or two
 
The following users thanked this post: Martin72

Offline Martin72

  • Super Contributor
  • ***
  • Posts: 7019
  • Country: de
  • Testfield Technician
Re: Rigol MSO5000 HiRes mode
« Reply #74 on: October 30, 2022, 11:29:42 pm »
Quote
usually “-3db at 100MHz”,

No other values ?


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf