Author Topic: 12 bit Keysight... when?  (Read 22444 times)

0 Members and 1 Guest are viewing this topic.

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: 12 bit Keysight... when?
« Reply #50 on: January 15, 2024, 10:08:15 pm »
4MPts memory depth?  C'mon Keysight.  That's like making us use a 4 function calculator instead of Matlab or other.

Memory depth is usually more important for marketing than engineering.  What matters more is the ratio between memory depth and bandwidth which determines how quickly the memory can be processed.  Previous HP DSOs had limited acquisition memory because they had a custom ASIC for processing, and users seem to have loved the fast processing that allowed.
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5560
  • Country: de
Re: 12 bit Keysight... when?
« Reply #51 on: January 16, 2024, 09:33:52 am »
Not sure if there's anything new here but there is a 50 page datasheet :

https://www.distrelec.de/en/oscilloscope-infiniivision-4000g-dso-4x-5ghz-5gsps-spi-rs232-rs422-rs485-usb-keysight-dsox4154g/p/30407321

That datasheet is actually well made and gives a good overview of the 4000x series.
But I do not see any reason to upgrade to the "G" model.

There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 13157
  • Country: ch
Re: 12 bit Keysight... when?
« Reply #52 on: January 16, 2024, 11:13:29 am »
Not sure if there's anything new here but there is a 50 page datasheet :

https://www.distrelec.de/en/oscilloscope-infiniivision-4000g-dso-4x-5ghz-5gsps-spi-rs232-rs422-rs485-usb-keysight-dsox4154g/p/30407321

That datasheet is actually well made and gives a good overview of the 4000x series.
But I do not see any reason to upgrade to the "G" model.
Of course not. The G models are simply a) color facelift, and b) include most of the options for free. There’s no difference in the hardware itself vs. the immediately preceding models.
 

Offline Pinkus

  • Frequent Contributor
  • **
  • Posts: 784
Re: 12 bit Keysight... when?
« Reply #53 on: January 16, 2024, 11:32:08 am »
4MPts memory depth?  C'mon Keysight.  That's like making us use a 4 function calculator instead of Matlab or other.

Memory depth is usually more important for marketing than engineering.  What matters more is the ratio between memory depth and bandwidth which determines how quickly the memory can be processed.  Previous HP DSOs had limited acquisition memory because they had a custom ASIC for processing, and users seem to have loved the fast processing that allowed.
I second this. Very very rarely have I wished for more memory when working with my Keysight MSOX4024a. But (especially when I look at other oscilloscopes in comparison) I am grateful every day for the speed, responsiveness and usefulness of my scope.
 

Offline mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Re: 12 bit Keysight... when?
« Reply #54 on: January 16, 2024, 12:15:58 pm »
4MPts memory depth?  C'mon Keysight.  That's like making us use a 4 function calculator instead of Matlab or other.

Memory depth is usually more important for marketing than engineering.  What matters more is the ratio between memory depth and bandwidth which determines how quickly the memory can be processed.  Previous HP DSOs had limited acquisition memory because they had a custom ASIC for processing, and users seem to have loved the fast processing that allowed.
I second this. Very very rarely have I wished for more memory when working with my Keysight MSOX4024a. But (especially when I look at other oscilloscopes in comparison) I am grateful every day for the speed, responsiveness and usefulness of my scope.
Definitey agree - even when you do hit memory limits, these can often be memory limits can often be mitigated by choice of acquisition mode, limiting channels etc. as well as smarter trigger/segmented setup.
 I have a MSOX3104T and R&S RTM3004 on the bench, and the Keysight gets 95% of the use despite the smaller screen and smaller memory - IMO the UI speed on the R&S is borderline unacceptable for a scope in its price range.   
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 
The following users thanked this post: tooki

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: 12 bit Keysight... when?
« Reply #55 on: January 16, 2024, 01:22:44 pm »
4MPts memory depth?  C'mon Keysight.  That's like making us use a 4 function calculator instead of Matlab or other.

Memory depth is usually more important for marketing than engineering.  What matters more is the ratio between memory depth and bandwidth which determines how quickly the memory can be processed.  Previous HP DSOs had limited acquisition memory because they had a custom ASIC for processing, and users seem to have loved the fast processing that allowed.
That depends on what you are doing. For circuits with microcontrollers / processors you'll need all the memory you can get. I used to have an Agilent model with the 8MPts Megazoom ASIC and that was not enough in many cases. Keep in mind that with the Keysight scopes you'll have 1/4th or even 1/8th of the memory available because it is shared amongst all channels (analog and digital).
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 12 bit Keysight... when?
« Reply #56 on: January 16, 2024, 01:56:24 pm »
I have both MSXO3104T and several long memory scopes (2 x Siglent and Picoscopes).

While Keysight is really fast to react to user controls, it really is very limited in some ways.
As long as you reach certain time-bases sampling rate drops quickly.
If you use it as you would a smarter CRT scope then it is fine.

But very small memory is very limiting for some use cases.


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

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: 12 bit Keysight... when?
« Reply #57 on: January 16, 2024, 07:05:56 pm »
That depends on what you are doing. For circuits with microcontrollers / processors you'll need all the memory you can get.

I would agree that in some cases record length is everything, but more often it is just for marketing where more is better.  Record length can be objectively advertised, but how can fast response of the user interface and ease of use advertised?

It is like how every new DSO is now advertised as "low noise", whether it really is or not.
 

Online EE-digger

  • Frequent Contributor
  • **
  • Posts: 391
  • Country: us
Re: 12 bit Keysight... when?
« Reply #58 on: January 17, 2024, 03:30:12 am »
I had several occasions where tens of megabytes of memory depth were needed for SPI to prove existence of processing glitches in a commercial chip.  For this we switched to a Saleae 16 analyzer which filled the bill nicely.  We performed rapid searches on the PC and proved the problem to ADI at the time.  So this created a dividing line for which I no longer blame the Keysight scopes.  Having this ability in the scope may have given completeness but not the best place for it to reside given PC speeds and space.
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 12 bit Keysight... when?
« Reply #59 on: January 17, 2024, 09:01:48 am »
I had several occasions where tens of megabytes of memory depth were needed for SPI to prove existence of processing glitches in a commercial chip.  For this we switched to a Saleae 16 analyzer which filled the bill nicely.  We performed rapid searches on the PC and proved the problem to ADI at the time.  So this created a dividing line for which I no longer blame the Keysight scopes.  Having this ability in the scope may have given completeness but not the best place for it to reside given PC speeds and space.

There are many occasions where you have glitches because of EMC, or some chip acting up. Some chips will react to very fast pulses and get confused.
Using protocol/logic analyser is fine to find the erroneous message but will not explain why message is wrong. On CAN you can have bus collisions that will show as bad message. But you don't know why. That is why deep memory scope that will keep sampling rate in long captures, and be able to decode is useful..

It really depends on what you do.


"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: Performa01, egonotto, pdenisowski

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5560
  • Country: de
Re: 12 bit Keysight... when?
« Reply #60 on: January 17, 2024, 09:51:03 am »

  For this we switched to a Saleae 16 analyzer which filled the bill nicely. 


This is a perfect combination with a scope and we might never see the abilities of a Saleae Logic Pro 16 in a scope. It always depends a lot on the application.

For me, the Keysight scopes, 3000x, 4000x and 6000x are just perfect as they are.

The only (small) option I am missing is to select what is saved on a screen shot.

There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline tautech

  • Super Contributor
  • ***
  • Posts: 29810
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: 12 bit Keysight... when?
« Reply #61 on: January 17, 2024, 09:54:36 am »
The only (small) option I am missing is to select what is saved on a screen shot.
How so ?

Reverse image ?
No visible menu ?
Trace colors user defined ?
Trace labels ?
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: 12 bit Keysight... when?
« Reply #62 on: January 17, 2024, 11:14:16 am »
That depends on what you are doing. For circuits with microcontrollers / processors you'll need all the memory you can get.

I would agree that in some cases record length is everything, but more often it is just for marketing where more is better.  Record length can be objectively advertised, but how can fast response of the user interface and ease of use advertised?
Fast response is not the only thing and ease of use is debatable. Try to setup decoding on a non-touchscreen Keysight scope. It is a horrible experience even when compared to a relatively cheap brand like GW Instek.

2N3055 is right on the money here:
If you use it as you would a smarter CRT scope then it is fine.

In that light it would make a lot of sense for Keysight to come up with a 12 bit version. Unless they can't put one on the market for a competitive price.
« Last Edit: January 17, 2024, 11:35:04 am by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5560
  • Country: de
Re: 12 bit Keysight... when?
« Reply #63 on: January 17, 2024, 11:21:59 am »
The only (small) option I am missing is to select what is saved on a screen shot.
How so ?

Reverse image ?
No visible menu ?
Trace colors user defined ?
Trace labels ?

See enclosed typical screenshot of the 4000X scope

I would like to:
1. Select the text that is shown up on the top of the screen (Custom Text)
2. Definitely be able to disable the serial number and Firmware version of my scope on the top
3. Enable or disable the time and date at the bottom right of the screen shot
(And why do I have the date and time at the top and bottom?)
4. Be able to select 12h or 24h system to show.

Best scenario would be a custom text with option of an incremental number in the text.
 
Same is true for the 3000x and 6000x
Besides these little things I am very happy with these Keysight scopes.
There are 3 kinds of people in this world, those who can count and those who can not.
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 12 bit Keysight... when?
« Reply #64 on: January 17, 2024, 12:34:57 pm »
The only (small) option I am missing is to select what is saved on a screen shot.
How so ?

Reverse image ?
No visible menu ?
Trace colors user defined ?
Trace labels ?

See enclosed typical screenshot of the 4000X scope

I would like to:
1. Select the text that is shown up on the top of the screen (Custom Text)
2. Definitely be able to disable the serial number and Firmware version of my scope on the top
3. Enable or disable the time and date at the bottom right of the screen shot
(And why do I have the date and time at the top and bottom?)
4. Be able to select 12h or 24h system to show.

Best scenario would be a custom text with option of an incremental number in the text.
 
Same is true for the 3000x and 6000x
Besides these little things I am very happy with these Keysight scopes.


Top line is snapshot data.
Serial and FW version are there for traceability. It is a good thing.

But I see no reason why it cannot be made configurable for those that want to disable it.

Time down left is part of the screen. It is a screenshot.

I also cannot comprehend that after 10+ years of platform, Keysight cannot give us 24h clock....
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1827
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: 12 bit Keysight... when?
« Reply #65 on: January 17, 2024, 02:00:06 pm »
I guess because they have the best 10-bit ADC converter on the market.

The important thing about a scope resolution is the "Effect Number Of Bits" or "ENOB".

Many 12bits scope only has around 6~8bits of ENOB (depending on bandwidth and sample rate), while Keysight can easily reach this number with their 10bit ADC. One of my main guesses they actually design a 12bit ADC, but only say "it is 10bit" when marketing.

Surely one day they will reach 12bit, but I don't think they will announce before they polish enough.

The Keysight 10bit converter is not bad - or it is generally hardly "better"...


Edit: Not in the picture: Siglent SDS7000 H12:  7.5bits (3Ghz), 7.3Bits (4Ghz)
Siglent SDS2000X HD: 8.7bits/8.4bits(100/350Mhz)

Another point is the advertising method about resolution. Many bits on paper look nice, but the only thing that really counts is effective bits (ENOB). These figures are *always* below what the ADC can deliver,
and that means that the ENOBs of 12bit ADCs are around 10 at best.

See here: https://electronicprojectsforfun.wordpress.com/to-enob-or-not-to-enob/

It has become a fashion to brag about resolution (e.g., R&S claims "up to 18bits" for their MXO4 and 5), but not to give ENOB numbers. That is deliberately misleading. Keysight and even Rigol are serious and
have their ENOB values documented. Ironically, R&S even has a paper discussing their pro scopes and why ENOB is important. Only their own marketing department did not get it :) for the MXO4 and 5.

The R&S paper is here:
https://scdn.rohde-schwarz.com/ur/pws/dl_downloads/dl_application/application_notes/1er03/ENOB_Technical_Paper_1ER03_1e.pdf


 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 12 bit Keysight... when?
« Reply #66 on: January 17, 2024, 02:45:01 pm »
I guess because they have the best 10-bit ADC converter on the market.

The important thing about a scope resolution is the "Effect Number Of Bits" or "ENOB".

Many 12bits scope only has around 6~8bits of ENOB (depending on bandwidth and sample rate), while Keysight can easily reach this number with their 10bit ADC. One of my main guesses they actually design a 12bit ADC, but only say "it is 10bit" when marketing.

Surely one day they will reach 12bit, but I don't think they will announce before they polish enough.

The Keysight 10bit converter is not bad - or it is generally hardly "better"...


Edit: Not in the picture: Siglent SDS7000 H12:  7.5bits (3Ghz), 7.3Bits (4Ghz)
Siglent SDS2000X HD: 8.7bits/8.4bits(100/350Mhz)

Another point is the advertising method about resolution. Many bits on paper look nice, but the only thing that really counts is effective bits (ENOB). These figures are *always* below what the ADC can deliver,
and that means that the ENOBs of 12bit ADCs are around 10 at best.

See here: https://electronicprojectsforfun.wordpress.com/to-enob-or-not-to-enob/

It has become a fashion to brag about resolution (e.g., R&S claims "up to 18bits" for their MXO4 and 5), but not to give ENOB numbers. That is deliberately misleading. Keysight and even Rigol are serious and
have their ENOB values documented. Ironically, R&S even has a paper discussing their pro scopes and why ENOB is important. Only their own marketing department did not get it :) for the MXO4 and 5.

The R&S paper is here:
https://scdn.rohde-schwarz.com/ur/pws/dl_downloads/dl_application/application_notes/1er03/ENOB_Technical_Paper_1ER03_1e.pdf

R&S is "selling" Hires mode as real bits..

Siglent for instance not only publishes ENOB figures but unlike some others also specifies how published number was measured.
From datasheet for SDS2000X HD :
ENOB*1 (typical):
(350 MHz) 8.4-bit, (200 MHz) 8.6-bit, (100 MHz) 8.7-bit
* 1:99.99 MHz input(100 MHz model uses 49.99 MHz),-0.5 dBFS,20 mV/div,50 Ω input impedance

Many manufacturers have the all the numbers, marketing don't care. 

Generally speaking marketing always speaks only off things that they can leverage as perceived advantage. They emphasize what they have...
While Keysight is talking it's ass off how important wfms/s is when talking about Megazoom scopes, they don't even mention it on Infinium-S series. Cause it has really low wfms/s because it is long memory analytical scopes. In datasheets for those it is all about how large memory is great..
So which one is true?
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: Performa01, egonotto

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: 12 bit Keysight... when?
« Reply #67 on: January 17, 2024, 07:02:49 pm »
Generally speaking marketing always speaks only off things that they can leverage as perceived advantage. They emphasize what they have...
While Keysight is talking it's ass off how important wfms/s is when talking about Megazoom scopes, they don't even mention it on Infinium-S series. Cause it has really low wfms/s because it is long memory analytical scopes. In datasheets for those it is all about how large memory is great..
So which one is true?

Marketing may also drive engineering.

During the Tektronix 22xx and 24xx era of DSOs, engineering came up with a way for the DSO to indicate that aliasing was likely present; just compare the trigger frequency with the sample rate.  So they had a prototype with an indicator to show when aliasing was present.  Leaving aside how useful this might be to inexperienced users, and keep in mind that everybody was an inexperienced user as far as DSOs back then, this feature was obviously dropped because it implied that Tektronix DSOs suffered from aliasing when their competitors did not.

Is there any actual use for the ENOB specification?  I would rather know the SFDR (spurious free dynamic range) which is directly observable with an FFT, and very likely where they are calculating the less useful ENOB from anyway.

Oscilloscopes used to specify aberrations, which is roughly equivalent to settling time.  I imagine these specifications would look especially bad on 12-bit and higher resolution instruments, some of which will never settle.  One reason sampling oscilloscopes are useful at lower frequencies is overload immunity and instant settling time.
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 13157
  • Country: ch
Re: 12 bit Keysight... when?
« Reply #68 on: January 17, 2024, 08:08:42 pm »
It has become a fashion to brag about resolution (e.g., R&S claims "up to 18bits" for their MXO4 and 5), but not to give ENOB numbers. That is deliberately misleading. Keysight and even Rigol are serious and
have their ENOB values documented. Ironically, R&S even has a paper discussing their pro scopes and why ENOB is important. Only their own marketing department did not get it :) for the MXO4 and 5.

The R&S paper is here:
https://scdn.rohde-schwarz.com/ur/pws/dl_downloads/dl_application/application_notes/1er03/ENOB_Technical_Paper_1ER03_1e.pdf

R&S is "selling" Hires mode as real bits..
How do you figure, given that the MXO 4 and 5 have 12-bit resolution all the time (the native resolution of the ADCs) and have a "high def" mode you can enable on top of that?
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1827
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: 12 bit Keysight... when?
« Reply #69 on: January 17, 2024, 08:45:43 pm »
It has become a fashion to brag about resolution (e.g., R&S claims "up to 18bits" for their MXO4 and 5), but not to give ENOB numbers. That is deliberately misleading. Keysight and even Rigol are serious and
have their ENOB values documented. Ironically, R&S even has a paper discussing their pro scopes and why ENOB is important. Only their own marketing department did not get it :) for the MXO4 and 5.

The R&S paper is here:
https://scdn.rohde-schwarz.com/ur/pws/dl_downloads/dl_application/application_notes/1er03/ENOB_Technical_Paper_1ER03_1e.pdf

R&S is "selling" Hires mode as real bits..
How do you figure, given that the MXO 4 and 5 have 12-bit resolution all the time (the native resolution of the ADCs) and have a "high def" mode you can enable on top of that?

Lets stick to physics:

- There are never more bits than an ADC can provide. Everything else is done by boxcar averaging and oversampling. While this can reduce noise, it will only make the inherent nonlinearities of the ADC more visible  >:D
These nonlinearities create spikes in the spectrum, automatically reducing dynamic range and also the resulting ENOB.

Most serious ADC chips have ENOB specs, depending on sampling rate, signal frequency, amplitude, ... The R&S MXO uses a Texas Instrument ADC (not even a bad one), and this one has ENOB values around 10
according to its datasheet. So - soory - no magic anywhere, except for the marketing guys who may plead incompetent.

An extreme (low frequency) example of a huge discrepancy between ADC resolution bits and useful bits are 24-32bit ADC used for digitizing. *Several* of the lower bits just drown in noise, still these ADCs are in the worlds best multimeters.
Its the linearity that makes an ADC premium.
 
The following users thanked this post: Performa01, 2N3055

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7466
  • Country: hr
Re: 12 bit Keysight... when?
« Reply #70 on: January 17, 2024, 09:05:02 pm »
Generally speaking marketing always speaks only off things that they can leverage as perceived advantage. They emphasize what they have...
While Keysight is talking it's ass off how important wfms/s is when talking about Megazoom scopes, they don't even mention it on Infinium-S series. Cause it has really low wfms/s because it is long memory analytical scopes. In datasheets for those it is all about how large memory is great..
So which one is true?

Marketing may also drive engineering.

During the Tektronix 22xx and 24xx era of DSOs, engineering came up with a way for the DSO to indicate that aliasing was likely present; just compare the trigger frequency with the sample rate.  So they had a prototype with an indicator to show when aliasing was present.  Leaving aside how useful this might be to inexperienced users, and keep in mind that everybody was an inexperienced user as far as DSOs back then, this feature was obviously dropped because it implied that Tektronix DSOs suffered from aliasing when their competitors did not.

Is there any actual use for the ENOB specification?  I would rather know the SFDR (spurious free dynamic range) which is directly observable with an FFT, and very likely where they are calculating the less useful ENOB from anyway.

Oscilloscopes used to specify aberrations, which is roughly equivalent to settling time.  I imagine these specifications would look especially bad on 12-bit and higher resolution instruments, some of which will never settle.  One reason sampling oscilloscopes are useful at lower frequencies is overload immunity and instant settling time.

Thank you for interesting story!

I believe it was dropped because it was not very useful, except in case of sinusoidal signals. And sounded bad like you said.

As for specifications, all the specifications are useful for somebody... ENOB was mentioned because of bits mentioned, because not all bits are same.

I agree with you that it would be nice if more of critical data would be specified. SFDR, SINAD, results ot two tone test (IMD) etc..
But I guess that gets complicated fast. With attenuators, VGA etc. you get many combinations of signal path to ADC, so that would be complicated tables of all the specs in all combinations...

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

Offline electr_peter

  • Supporter
  • ****
  • Posts: 1443
  • Country: lt
Re: 12 bit Keysight... when?
« Reply #71 on: January 17, 2024, 09:53:47 pm »
During the Tektronix 22xx and 24xx era of DSOs, engineering came up with a way for the DSO to indicate that aliasing was likely present; just compare the trigger frequency with the sample rate.  So they had a prototype with an indicator to show when aliasing was present.  Leaving aside how useful this might be to inexperienced users, and keep in mind that everybody was an inexperienced user as far as DSOs back then, this feature was obviously dropped because it implied that Tektronix DSOs suffered from aliasing when their competitors did not.
Interesting story. Need to think how useful (or not) this feature would be.

I think that current DSOs could show something related on the screen. A popup of some sort can light up whenever sampling rate is <2x of scope nominal BW (or with selected 20MHz BW filter 40MS/s or less). This is left to the user AFAIK - actual sampling rate is shown, but it is not immediately obvious if it is sufficient.

Such popup would at least indicate that some signals may alias and sinc interpolation may not be appropriate. DSOX3k scopes solve these problems behind the curtains so to speak (via random decimation and by disabling sinc interpolation with lower sample rate).
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: 12 bit Keysight... when?
« Reply #72 on: January 18, 2024, 01:22:23 am »
- There are never more bits than an ADC can provide. Everything else is done by boxcar averaging and oversampling. While this can reduce noise, it will only make the inherent nonlinearities of the ADC more visible  >:D
These nonlinearities create spikes in the spectrum, automatically reducing dynamic range and also the resulting ENOB.

Nothing prevents having a converter with greater linearity than its resolution will support, but they are rare and except as describes below, I am not aware of any fast ones.

Modern fast ADCs all seem to be pipelined subranging types with 3 or 4 stages in series, with the earlier stages having a linearity much better than their low 3 to 5 bit resolution will support because it directly affects the later stages.  Error correction then discards some of the generated bits to produce the specified resolution at the interface.  Multiples of these subranging converters are interleaved for even higher sample rates and their self calibration encompasses this as well.

I think that current DSOs could show something related on the screen. A popup of some sort can light up whenever sampling rate is <2x of scope nominal BW (or with selected 20MHz BW filter 40MS/s or less). This is left to the user AFAIK - actual sampling rate is shown, but it is not immediately obvious if it is sufficient.

Some of the appeal of what the Tektronix designers did was because it was so easy to do on these early DSOs.  They all included high speed level triggering which operates very much like a data slicer, so comparing the processed trigger signal to 1/2 of the sample rate will immediately reveal aliasing.  The only thing required was a simplified frequency counter.
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1827
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: 12 bit Keysight... when?
« Reply #73 on: January 18, 2024, 11:27:39 am »
- There are never more bits than an ADC can provide. Everything else is done by boxcar averaging and oversampling. While this can reduce noise, it will only make the inherent nonlinearities of the ADC more visible  >:D
These nonlinearities create spikes in the spectrum, automatically reducing dynamic range and also the resulting ENOB.

Nothing prevents having a converter with greater linearity than its resolution will support, but they are rare and except as describes below, I am not aware of any fast ones.

Modern fast ADCs all seem to be pipelined subranging types with 3 or 4 stages in series, with the earlier stages having a linearity much better than their low 3 to 5 bit resolution will support because it directly affects the later stages.  Error correction then discards some of the generated bits to produce the specified resolution at the interface.  Multiples of these subranging converters are interleaved for even higher sample rates and their self calibration encompasses this as well.

I think that current DSOs could show something related on the screen. A popup of some sort can light up whenever sampling rate is <2x of scope nominal BW (or with selected 20MHz BW filter 40MS/s or less). This is left to the user AFAIK - actual sampling rate is shown, but it is not immediately obvious if it is sufficient.

Some of the appeal of what the Tektronix designers did was because it was so easy to do on these early DSOs.  They all included high speed level triggering which operates very much like a data slicer, so comparing the processed trigger signal to 1/2 of the sample rate will immediately reveal aliasing.  The only thing required was a simplified frequency counter.

Agreed, but none of these tricks can be used in scope front ends with high sampling speeds. The "naked" linearity is what counts there, and there seems to be no way around that, AFAIK.
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: 12 bit Keysight... when?
« Reply #74 on: January 18, 2024, 12:38:56 pm »
During the Tektronix 22xx and 24xx era of DSOs, engineering came up with a way for the DSO to indicate that aliasing was likely present; just compare the trigger frequency with the sample rate.  So they had a prototype with an indicator to show when aliasing was present.  Leaving aside how useful this might be to inexperienced users, and keep in mind that everybody was an inexperienced user as far as DSOs back then, this feature was obviously dropped because it implied that Tektronix DSOs suffered from aliasing when their competitors did not.
Interesting story. Need to think how useful (or not) this feature would be.
I agree. A light goes on saying what you see is not what you got. And then???  :)  IMHO peak-detect is a much more useful feature to prevent aliasing. At least you can see the extremes of the signal.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf