So if I'm not wrong my calc for MSOX3104T would go like this, for those conditions:
(Attachment Link)
and
(Attachment Link)
so that gives:
(Attachment Link)
Although ENOB is not single number, it is frequency/ENOB graph....
So if I'm not wrong my calc for MSOX3104T would go like this, for those conditions:
(Attachment Link)
and
(Attachment Link)
so that gives:
(Attachment Link)
Although ENOB is not single number, it is frequency/ENOB graph....
Shouldn't the ENOB be computed based upon a sine wave input of full scale peak to peak (8 divisions)? If so then the relative signal power would be (4*scalefacor/rt2)^2, or 8*(Scale Factor)^2.
Best,
So if I'm not wrong my calc for MSOX3104T would go like this, for those conditions:
(Attachment Link)
and
(Attachment Link)
so that gives:
(Attachment Link)
Although ENOB is not single number, it is frequency/ENOB graph....
Shouldn't the ENOB be computed based upon a sine wave input of full scale peak to peak (8 divisions)? If so then the relative signal power would be (4*scalefacor/rt2)^2, or 8*(Scale Factor)^2.
Best,
To be honest ENOB is supposed to be measured specified with the actual signal at actual frequency, so it includes both noise and distortions in SINAD.
In this case we assume (for 2mv/div for instance) full scale P-P sine signal of 16 mV. RMS of that (11,35mV) divided with RMS of noise (110 uV) gives 10641 ratio, convert that to dB (power ratio) gives 40,2dB, and converted to ENOB :6,397 bit.
I did quick test with R&S method and got same results.
These tests will have major source of error in scopes built in RMS function and it's limits to deal with calculating RMS of noise and otherwise very complex signal.
Please correct me if I'm wrong..
Regards,
Please note that my table was effective resolution, not ENOB; there is a difference.
TurboTom: I do still have the scopes, but I'm very time poor at the moment.
So if I'm not wrong my calc for MSOX3104T would go like this, for those conditions:
(Attachment Link)
and
(Attachment Link)
so that gives:
(Attachment Link)
Although ENOB is not single number, it is frequency/ENOB graph....
Shouldn't the ENOB be computed based upon a sine wave input of full scale peak to peak (8 divisions)? If so then the relative signal power would be (4*scalefacor/rt2)^2, or 8*(Scale Factor)^2.
Best,
To be honest ENOB is supposed to be measured specified with the actual signal at actual frequency, so it includes both noise and distortions in SINAD.
In this case we assume (for 2mv/div for instance) full scale P-P sine signal of 16 mV. RMS of that (11,35mV) divided with RMS of noise (110 uV) gives 10641 ratio, convert that to dB (power ratio) gives 40,2dB, and converted to ENOB :6,397 bit.
I did quick test with R&S method and got same results.
These tests will have major source of error in scopes built in RMS function and it's limits to deal with calculating RMS of noise and otherwise very complex signal.
Please correct me if I'm wrong..
Regards,
I had seen the IEEE definition on ENOB, and agree it's without signal distortion and not at a specified signal frequency since no signal is actually applied. So maybe a best case ENOB, or as Howard mentioned Effective Resolution.
From above "In this case we assume (for 2mv/div for instance) full scale P-P sine signal of 16 mV. RMS of that (11,35mV) divided with RMS of noise (110 uV) gives 10641 ratio, convert that to dB (power ratio) gives 40,2dB, and converted to ENOB :6,397 bit."
Since the peak to peak full scale signal would yield a sine wave of 4 divisions peak, the RMS would not be of the 8 division peak to peak value, but of the 4 division value. So this would yield 4*2mV/rt2 = 5.657mV RMS, not 8*2mV/rt2 = 11.35mV??
Just trying to follow the thinking behind these calculations, so I can supply a similar set of values that are consistent with what's been shown.
Best,
My logic is that full screen at 2 mV/div is 16 mV from top to bottom (8 mV above and 8mV below zero). If you inscribe sinewave inside, top to bottom, RMS of that sinewave would be 16 mV/1.41= 11,35 mV.
That is full scale RMS. That is also how I understood IEEE excerpt in R&S whitepaper.
But, I agree it is beside point to call upon some standard if all conditions are not observed. And I agree that we shouldn't call it ENOB but effective resolution.
We could simply define a figure of merit that would compare full scale with residual RMS and P-P noise.. It would serve well as relative comparison and would be somewhat easier to reproduce.
There were previous comparisons made, including calculating noise PSD and such...
My logic is that full screen at 2 mV/div is 16 mV from top to bottom (8 mV above and 8mV below zero). If you inscribe sinewave inside, top to bottom, RMS of that sinewave would be 16 mV/1.41= 11,35 mV.
That is full scale RMS. That is also how I understood IEEE excerpt in R&S whitepaper.
But, I agree it is beside point to call upon some standard if all conditions are not observed. And I agree that we shouldn't call it ENOB but effective resolution.
We could simply define a figure of merit that would compare full scale with residual RMS and P-P noise.. It would serve well as relative comparison and would be somewhat easier to reproduce.
There were previous comparisons made, including calculating noise PSD and such...
The RMS of a sinusoid is (peak value)/rt2, not (peak to peak value)/rt2. In your example you have a full scale 8 division sine wave at 2mv/div, so 16mv peak to peak and thus 8mv/rt2 RMS value.
Agree a FOM should be utilized and I'm all ears for what should be utilized/created for such.
Best,
So if I'm not wrong my calc for MSOX3104T would go like this, for those conditions:
(Attachment Link)
and
(Attachment Link)
so that gives:
(Attachment Link)
Although ENOB is not single number, it is frequency/ENOB graph....
Shouldn't the ENOB be computed based upon a sine wave input of full scale peak to peak (8 divisions)? If so then the relative signal power would be (4*scalefacor/rt2)^2, or 8*(Scale Factor)^2.
Best,
To be honest ENOB is supposed to be measured specified with the actual signal at actual frequency, so it includes both noise and distortions in SINAD.
In this case we assume (for 2mv/div for instance) full scale P-P sine signal of 16 mV. RMS of that (11,35mV) divided with RMS of noise (110 uV) gives 10641 ratio, convert that to dB (power ratio) gives 40,2dB, and converted to ENOB :6,397 bit.
I did quick test with R&S method and got same results.
These tests will have major source of error in scopes built in RMS function and it's limits to deal with calculating RMS of noise and otherwise very complex signal.
Please correct me if I'm wrong..
Regards,
I had seen the IEEE definition on ENOB, and agree it's without signal distortion and not at a specified signal frequency since no signal is actually applied. So maybe a best case ENOB, or as Howard mentioned Effective Resolution.
From above "In this case we assume (for 2mv/div for instance) full scale P-P sine signal of 16 mV. RMS of that (11,35mV) divided with RMS of noise (110 uV) gives 10641 ratio, convert that to dB (power ratio) gives 40,2dB, and converted to ENOB :6,397 bit."
Since the peak to peak full scale signal would yield a sine wave of 4 divisions peak, the RMS would not be of the 8 division peak to peak value, but of the 4 division value. So this would yield 4*2mV/rt2 = 5.657mV RMS, not 8*2mV/rt2 = 11.35mV??
Just trying to follow the thinking behind these calculations, so I can supply a similar set of values that are consistent with what's been shown.
Best,
My logic is that full screen at 2 mV/div is 16 mV from top to bottom (8 mV above and 8mV below zero). If you inscribe sinewave inside, top to bottom, RMS of that sinewave would be 16 mV/1.41= 11,35 mV.
That is full scale RMS. That is also how I understood IEEE excerpt in R&S whitepaper.
But, I agree it is beside point to call upon some standard if all conditions are not observed. And I agree that we shouldn't call it ENOB but effective resolution.
We could simply define a figure of merit that would compare full scale with residual RMS and P-P noise.. It would serve well as relative comparison and would be somewhat easier to reproduce.
There were previous comparisons made, including calculating noise PSD and such...
You're right to question them, as did I at the time.
The figures are from the scope's own measurements.
This your RMS is corrected after this original msg, so I do not take this accidental human mistake on table.
But I want ask one think.
There are these..
"My logic is that full screen...blablabla.."
Then also
"That is full scale RMS."
Not all scopes have ADC full scale same is display full height.
Of course it do not make big difference but... small error there and other error here and there... and we talk finally perhaps big error.
Many scopes (but not all) what I have handled have ADC full scale around or over 10 vertical div and displayed part of whole vertical is 8 div. But different scopes may be different in this.
If 2mv/div and ADC FS is example bit over 10 div. For simplify think 10div.
In this case FS sinewave is 20mVp-p so 7.071 mVrms. I think there need use this ADC full scale in calculations what is used in individual scope instead of displayed part of signal if it is different.
You're right to question them, as did I at the time.
The figures are from the scope's own measurements.
Did you do a self-cal immediately before making those measurements?
Please note that my table was effective resolution, not ENOB; there is a difference.
TurboTom: I do still have the scopes, but I'm very time poor at the moment.
Could you show how you calculated the Effective Resolution in your table so we can compare results on an "Apples to Apples" basis.
BTW thanks for showing the table and taking the time to post these results
Best,
Just for completeness:
The SDS2000X+ and also SDS5000X have 30 LSB/div, that is 240 steps for full screen. Very little overhead.
The SDS1000X-E (and most likely also SDS2000X-E) have 25 steps per division, hence only 200 counts full screen.
The SDS2000X+ can maintain the full extended bandwidth of 500 MHz down to 500 µV/div.
All contemporary Siglent DSOs prvide fine adjust of the vertical gain, and these are true hardware PGA steps.
Please note that my table was effective resolution, not ENOB; there is a difference.
TurboTom: I do still have the scopes, but I'm very time poor at the moment.
Could you show how you calculated the Effective Resolution in your table so we can compare results on an "Apples to Apples" basis.
BTW thanks for showing the table and taking the time to post these results
Best,
Here's the spreadsheet I used.
Other than the brand look and feel and having similar interfaces (I would hope) are there other advantages of going with the same manufacturer?
Please note that my table was effective resolution, not ENOB; there is a difference.
TurboTom: I do still have the scopes, but I'm very time poor at the moment.
Could you show how you calculated the Effective Resolution in your table so we can compare results on an "Apples to Apples" basis.
BTW thanks for showing the table and taking the time to post these results
Best,
Here's the spreadsheet I used.
Thanks. If I interpret the spreadsheet correctly you are using Effective Resolution Bit as Log(FS/N)/Log(2) and not the standard ENOB of (SINAD -1.76)/6.02, where in our case SINAD would be 10*Log(FS/N +1) using RMS values and no distortion term, since no input signal is actually applied.
Using what I did which includes the ADC quantizing error term but no distortion term I get similar but slightly 0.3 bit lower (not surprising since 1.76/6.02 ~0.3 bit) results, for example.
MSO8104A
2mv 5.26 vs. 5.55
100mv 6.71 vs. 7.01
200mv 6.71 vs. 7.01
500mv 6.74 vs. 7.03
5V 6.70 vs. 6.99
Best,
Back to the original question...QuoteOther than the brand look and feel and having similar interfaces (I would hope) are there other advantages of going with the same manufacturer?
I have several Siglent 'x' devices, but they do not really have the same look and feel, as you (rightfully) hope. Their user interfaces differs in many ways, and the 'deeper' you go (e.g. configuring the communication interfaces), the larger the differences. While the screens with their underlying 5 or 6 soft buttons suggests a consistent UI approach, this is not the case, and the overall button 'logic' varies. Physical button placement varies quite a bit. For instance, the SDG AWG and SDL electronic load both have 10 digit keypads, but they keys are just the other way around. Also the PC software for these devices seems to have nothing in common, entirely different programs. Some have 'real' on/of buttons, some soft buttons. The cases stack reasonably well, but as others have noted, their depth is sometimes different.
Having that said, I selected them for what I needed for specific use cases, and feel happy for what they do at their price point. They are quite nice devices and I would buy them again.
Siglent should take note of this uniform UI experience if they expect to move up the instrument food chain.
The equipment decision makers are older and tend to use tried and true instruments they know and have experience with.
Please note that my table was effective resolution, not ENOB; there is a difference.
TurboTom: I do still have the scopes, but I'm very time poor at the moment.
Could you show how you calculated the Effective Resolution in your table so we can compare results on an "Apples to Apples" basis.
BTW thanks for showing the table and taking the time to post these results
Best,
Here's the spreadsheet I used.
Thanks. If I interpret the spreadsheet correctly you are using Effective Resolution Bit as Log(FS/N)/Log(2) and not the standard ENOB of (SINAD -1.76)/6.02, where in our case SINAD would be 10*Log(FS/N +1) using RMS values and no distortion term, since no input signal is actually applied.
Using what I did which includes the ADC quantizing error term but no distortion term I get similar but slightly 0.3 bit lower (not surprising since 1.76/6.02 ~0.3 bit) results, for example.
MSO8104A
2mv 5.26 vs. 5.55
100mv 6.71 vs. 7.01
200mv 6.71 vs. 7.01
500mv 6.74 vs. 7.03
5V 6.70 vs. 6.99
Best,
Yes, as I stated before, effective resolution isn’t the same as ENOB. As you say, effective resolution gives an apparently “better” result than ENOB.
It is however possible the Vrms calculation in the MSO5000 is wrong