Siglent SDS1000X HD 12bit DSO's

[robert.rozee]

hopefully folks won't deride this as a silly question... but what practical usage advantage does a 12-bit scope have over an 8-bit scope?

it seems to me that 12-bits gets you a 16x vertical zoom advantage, but i can't see that as being something that would be of much value in most usage cases. and how much useful information do those extra 4 bits really carry?

a distinct disadvantage seems to be the loss of features at a given price point, as manufacturers need to cut costs elsewhere to pay for the 12-bit hardware. for instance: all else being equal, would you rather your 4-channel scope have a single shared 12-bit ADC, or 4 independent 8-bit ADC?

i also wonder if much of the advantage of a 12-bit ADC could be achieved with front-end improvements that allow precise DC offsetting (even if only on 1 of the 4 channels) that gives you a view into a narrow vertical band (for example, dial in a 5v offset so you you can look at a 4.95v to 5.05v band without distortions due to over-driving of the front end).


cheers,
rob   :-)

[Martin72]

Hi,

a distinct disadvantage seems to be the loss of features at a given price point, as manufacturers need to cut costs elsewhere to pay for the 12-bit hardware.

I think this only applies to the low-cost variants in order to be able to offer a 12-bit scope in the segment at all.
The lecroy HDO6034B for example has the full sample rate and the full memory even if all channels are active.
But it also costs 18000$ as an entry in this series.
That's why the models from Siglent (2000X+HD) and Rigol (DHO4000) are also low cost, with corresponding concessions.
These become even more significant if you make it even cheaper (DHO1000 and SDS1000X HD, respectively).
So what are the advantages, a small insight gives this article:
https://www.electronicdesign.com/technologies/test-measurement/article/21805339/whats-really-the-difference-between-a-12bit-and-8bit-oscilloscope

I took last year nearly 3600€ in the hand and stood before the decision, I buy the new 12 bit model of siglent, with max 2GSa/s and 100Mhz(max. 500Mhz).
Or the SDS5000X with 350Mhz (max 1Ghz) and 5GSa/s.
Both cost about the same.
I decided then for 12bit, because I would like to use its advantages in the middle future.
For my current measurements in practice it is irrelevant whether 8bit or 12bit.
You have to think about it, do I really need a higher accuracy (now or later) yes or no.
Because currently you can get a better equipped 8 bit scope for the same money or even less.
E.g. for a DHO1000 a MSO5000 from rigol or a 2000X+ from Siglent.
Or for a DHO4000 or 2000+HD a SDS5000X.

[robert.rozee]

https://www.electronicdesign.com/technologies/test-measurement/article/21805339/whats-really-the-difference-between-a-12bit-and-8bit-oscilloscope

good article, although i was a little puzzled by this part:
"Most modern, mid-range oscilloscopes include a high-resolution mode, which increases the number of bits by filtering. In essence, bandwidth is traded off for the increased dynamic range. Basically you get one additional bit for each halving of the scope’s bandwidth. Therefore, to achieve a 4-bit increase, it would reduce the scope’s bandwidth by 16:1. So if you start with a 1-GHz scope, increasing the resolution to 12 bits with the high-resolution mode would reduce the scope’s bandwidth to below 100 MHz."

my understand was that for each ONE bit increase, it required 4x oversampling, leading to a bandwidth [addendum: i should probably be talking about 'sample rate' here, not 'bandwidth'] reduction of 256:1 to get the extra 4 bits. perhaps we are instead seeing 10-bit ADCs being used with a further 2 bits then obtained in software to give us the final 12 bits. i shall need to research this point some more - if the author is correct then i have a few (abandoned) projects i need to revisit!

he does present one example, a decaying ringing, that nicely illustrates a useful application, and i guess if this sort of example dominates your area of interest then a 12-bit scope would be a time-saver. however, i could see that with the right setup the same level of detail could be captured with two 8-bit channels each set for different vertical scales (once over-driving issues were eliminated).

you are right, that in the end it comes down to balancing budgets and individual user cases. and as always, any advances will filter down over time into even the lowest end of the market. essentially, a 12-bit scope is one step towards replacing the venerable 3½-digit multimeter on one's workbench. but for today, at what cost?


cheers,
rob   :-)


addendum (a day later) see text in bold red above

[baldurn]

If you are going to do math, especially FFT, those extra bits will make a huge difference.

In addition the noise floor of the SDS2000X-HD is specified as 50 mikrovolt. The SDS2000X+ specifies "sensitivity" of 0.5 div with minimum div of 500 mikrovolt = 250 mikrovolt. So I expect you do get less noise?

Also looks like the new SDS1000X-HD is getting the new "glass cockpit". I know it is the same UI etc but I think it looks nicer :-)

[tautech]

Also looks like the new SDS1000X-HD is getting the new "glass cockpit". I know it is the same UI etc but I think it looks nicer :-)

What does that actually mean, you've lost me what a glass cockpit has to do with a DSO ? 

SDS1000X HD is in the exact same enclosure as SDS2000X HD with the same physical measurements 317.2 mm × 236.0 mm × 149.0 mm and same weight too 4.1 kg.

This newish front panel layout is a consolidation and improvement of SDS2000X Plus with smarter use of the GUI and menu structure however I am saddened SDS1000X HD is not apparently supplied with a wireless mouse as SDS2000X HD and SDS6000A are as this UI does benefit from use with a mouse although it certainly doesn't require one to be fully functional.

[2N3055]

https://www.electronicdesign.com/technologies/test-measurement/article/21805339/whats-really-the-difference-between-a-12bit-and-8bit-oscilloscope

good article, although i was a little puzzled by this part:
"Most modern, mid-range oscilloscopes include a high-resolution mode, which increases the number of bits by filtering. In essence, bandwidth is traded off for the increased dynamic range. Basically you get one additional bit for each halving of the scope’s bandwidth. Therefore, to achieve a 4-bit increase, it would reduce the scope’s bandwidth by 16:1. So if you start with a 1-GHz scope, increasing the resolution to 12 bits with the high-resolution mode would reduce the scope’s bandwidth to below 100 MHz."

my understand was that for each ONE bit increase, it required 4x oversampling, leading to a bandwidth [addendum: i should probably be talking about 'sample rate' here, not 'bandwidth'] reduction of 256:1 to get the extra 4 bits. perhaps we are instead seeing 10-bit ADCs being used with a further 2 bits then obtained in software to give us the final 12 bits. i shall need to research this point some more - if the author is correct then i have a few (abandoned) projects i need to revisit!

he does present one example, a decaying ringing, that nicely illustrates a useful application, and i guess if this sort of example dominates your area of interest then a 12-bit scope would be a time-saver. however, i could see that with the right setup the same level of detail could be captured with two 8-bit channels each set for different vertical scales (once over-driving issues were eliminated).

you are right, that in the end it comes down to balancing budgets and individual user cases. and as always, any advances will filter down over time into even the lowest end of the market. essentially, a 12-bit scope is one step towards replacing the venerable 3½-digit multimeter on one's workbench. but for today, at what cost?


cheers,
rob   :-)


addendum (a day later) see text in bold red above

When you do Hi Res in software, actual BW is reduced. It works by basically enabling digital low pass filter. So you get smaller effective sample rate (less points in time interval) and digitally filtered lower BW...
Both is less. That is opposed to simply reducing sample rate where you only get less points.

[thm_w]

The resolution of the screen is pretty low, no hdmi or video output of any kind, no wifi.
Having to use an external signal generator makes it more expensive, not cheaper.
I was waiting for Siglent to release the competition of the rigol dho1000 but this is not very tempting.
As with the dho1000 the only good thing are the 12 bit.

Anyone considering a 12 bit scope would likely already own an external sig gen of better capability than you can bundle in. If this were a $400 entry level scope I would understand your point.

Also looks like the new SDS1000X-HD is getting the new "glass cockpit". I know it is the same UI etc but I think it looks nicer :-)

What does that actually mean, you've lost me what a glass cockpit has to do with a DSO ? 

I think they just mean glass cockpit = less physical buttons and more use of the touch screen.
Compared to sds1000x 36 buttons this has cut it to 23 buttons. Same 6 knobs but placed closer together.

Just personal preference. I would probably go with the larger screen.

[Martin72]

If the housing is really the same as the 2000X HD, then it will be in superb quality.
I like it a lot on my SDS2000X HD, also the rotary knobs/encoders with their "clicking sound".
Display should be also the same, then the probesensed 1M/50ohm inputs...
I´m very curious about the entry-price.

[tautech]

Also looks like the new SDS1000X-HD is getting the new "glass cockpit". I know it is the same UI etc but I think it looks nicer :-)

What does that actually mean, you've lost me what a glass cockpit has to do with a DSO ? 

I think they just mean glass cockpit = less physical buttons and more use of the touch screen.
Compared to sds1000x 36 buttons this has cut it to 23 buttons. Same 6 knobs but placed closer together.

Just personal preference. I would probably go with the larger screen.

Yep, that makes sense I guess.

I use the 1000X-E and these touch screen DSO lots and each has ways to best utilise them however nothing compares with the larger touch screen display when you add a mouse.

When you have 3 user input methods after some little practice to find/discover most efficient usage methods throughput is dramatically improved but it does take a little while to discover all the features the GUI offers like hidden virtual keyboards and mouse scroll wheel support.

I imagine this new 1kX HD model will include all the GUI features of its big brothers, SDS5000X, SDS2000X Plus, SDS2000X HD and SDS6000A.

[BillyO]

I want one.  It would make a perfect side-kick to my SDS2504X-P.

Is anyone interested in s slightly used 6 moth old SDS1104X-E (fully liberated)?

[Performa01]

hopefully folks won't deride this as a silly question... but what practical usage advantage does a 12-bit scope have over an 8-bit scope?

it seems to me that 12-bits gets you a 16x vertical zoom advantage, but i can't see that as being something that would be of much value in most usage cases. and how much useful information do those extra 4 bits really carry?

First we should consider the fact that even on an entry level DSO, the vertical screen height (in pixels) is at least twice as much as the net number of levels in an 8-bit ADC. Thus, every data point is actually drawn as a line that is two (or even more) pixels high if you utilize a full screen waveform display.

This is an issue especially for the XY-mode, where this problem exists not only for the vertical direction, but also horizontally. Each display dot is actually a square of 2 x 2 pixels on an 8-bit scope. The sole reason why it still looks decent is the DPO technology, mixing a picture from thousands of single acquisitions.

The above have been only the visual effects so far.

Dynamic range is another important consideration, especially when using math, as has already been mentioned before. It makes a difference if we have a guaranteed genuine dynamic range of 49 dB or 72 dB – and this difference can be demonstrated quite easily. I have once shown a few examples in the SDS2000X HD thread…

a distinct disadvantage seems to be the loss of features at a given price point, as manufacturers need to cut costs elsewhere to pay for the 12-bit hardware. for instance: all else being equal, would you rather your 4-channel scope have a single shared 12-bit ADC, or 4 independent 8-bit ADC?

Well, this is not a valid question. If you need (or even only want) 12 bits, then you don't care.

Apart from that: few DSOs have independent ADCs per channel these days. Most of the better ones have two ADCs for four channels. At Siglent, the only exception currently is the SDS6000A (8-bit) and the SDS6000 H10/12 Pro (10/12 bit, available in China only), which use four independent 5 GSa/s ADCs.

i also wonder if much of the advantage of a 12-bit ADC could be achieved with front-end improvements that allow precise DC offsetting (even if only on 1 of the 4 channels) that gives you a view into a narrow vertical band (for example, dial in a 5v offset so you you can look at a 4.95v to 5.05v band without distortions due to over-driving of the front end).
 

A precise offset would be nothing new. I've already demonstrated several times, how we can measure for instance 205V with better than 0.08% accuracy even with the entry level SDS1104X-E (8-bit).

The overload issues on the other hand, are inherent to the split path input buffer design, universally used in all higher bandwidth scopes. This won't change anytime soon, because it is the prerequisite for accurate and stable DC offsets.

[Performa01]

https://www.electronicdesign.com/technologies/test-measurement/article/21805339/whats-really-the-difference-between-a-12bit-and-8bit-oscilloscope

good article, although i was a little puzzled by this part:
"Most modern, mid-range oscilloscopes include a high-resolution mode, which increases the number of bits by filtering. In essence, bandwidth is traded off for the increased dynamic range. Basically you get one additional bit for each halving of the scope’s bandwidth. Therefore, to achieve a 4-bit increase, it would reduce the scope’s bandwidth by 16:1. So if you start with a 1-GHz scope, increasing the resolution to 12 bits with the high-resolution mode would reduce the scope’s bandwidth to below 100 MHz."

my understand was that for each ONE bit increase, it required 4x oversampling, leading to a bandwidth [addendum: i should probably be talking about 'sample rate' here, not 'bandwidth'] reduction of 256:1 to get the extra 4 bits. perhaps we are instead seeing 10-bit ADCs being used with a further 2 bits then obtained in software to give us the final 12 bits. i shall need to research this point some more - if the author is correct then i have a few (abandoned) projects i need to revisit!

The resolution enhancement goes along with the oversampling. For instance, four times oversampling results in four times more resolution = 2 bits of additional resolution and at the same time only one fourth of the bandwidth.

Best example for this is the Siglent SDS2000X Plus in 10-bit mode. It utilizes four times oversampling for two additional bits of resolution and at the same time the bandwidth drops to about 100 MHz, because with all four channels in use, the sample rate is limited to 1 GSa/s per channel, therefore Nyquist is at 500 MHz and one fourth of that is 125 MHz.

The confusing part might be that the noise reduction is only 3 dB per doubling of the sample frequency. So it is only 6 dB for four times oversampling, thus resulting in an ENOB enhancement of only one bit.

Long story short: we have to distinguish resolution and ENOB – and this is why physical 12-bit ADCs are the superior solution compared to HiRes/ERES oversampling approaches, even when we ignore the bandwidth reduction caused by oversampling.

[nctnico]

Remember: In the end you can't turn an 8 bit ADC into a 12 bit ADC just by oversampling. Getting extra bits relies on linearity and having enough & good noise to work with. Neither is guaranteed (and more likely not there in practical situation).

Also keep in mind that an oscilloscope is not a precission instrument. A 12 bit ADC gives extra resolution to look at small wiggles in a signal (which typically get removed by hires / eres BTW) but it doesn't give extra accuracy.

[electr_peter]

SDS1000X HD would be very welcome seeing positive feedback on SDS2000X HD. According to datasheets, SDS1000X HD is the same as SDS2000X HD with lower parts count but with almost all capabilities of bigger brother (missing zone trigger, lower sample rate).
SDS2000X HD firmware currently is better developed than Rigol DHO(HDO)1000/4000, likely SDS1000X HD will be similar. Main attraction is 12 bit ADC in nice form factor and many features. If price is expected to be in 800-1000 USD range or even lower, this will be a no brainer.

[Martin72]

Yepp, and if this price will come true, I wonder what else is "slimfasted" in comparison to the 2kHD...
One instead two ADCs, ok.
No inbuild awg, ok.
But same solid case, screen, maybe the same good encoders.
Where else could they have done to make it (if the price will come) appx 2500 bucks cheaper.
But that´s only guessing, time will tell and the first launched one in the hands of a teardowner...

[switchabl]

The datasheet lists SFDR ≥35 dB (same as 1000X-E) which would be subpar for a 12-bit instrument. But we will need to wait for proper tests to see what that means in practice.

[Martin72]

Other models:
5000X >32dB
2000X+ >40dB
2000X HD >45dB
6000A >45dB

[Neutrion]

Is it possible to run these 12bit scopes in an 8bit mode but with higher sample rate, and more pages in the memory?

[rf-loop]

Is it possible to run these 12bit scopes in an 8bit mode but with higher sample rate, and more pages in the memory?

No.

[Martin72]

Any guesses on cost ?

My guess is, the 1104X HD will cost appx the same as the 2104X+....
Because the prices in China for the 2000 X HD model are almost identical to the prices here without VAT.
So I assume it won't be any different with the 1000X HD.
If you can remember, with the 2000X HD, we also thought the prices would be lower at first, but they stayed the same.

[tautech]

Any guesses on cost ?

My guess is, the 1104X HD will cost appx the same as the 2104X+....

We'll see, my pick is SDS2102X Plus cost but no more.

Because the prices in China for the 2000 X HD model are almost identical to the prices here without VAT.

IIRC early in the 2kX HD thread rf-loop stated it was priced too cheap in his opinion due to the cost of the ADC's.

So I assume it won't be any different with the 1000X HD.
If you can remember, with the 2000X HD, we also thought the prices would be lower at first, but they stayed the same.

Yet it's been trimmed down to use existing external HW which will reduce BOM and cost.

But, we will see. 
 

[Martin72]

We'll see, my pick is SDS2102X Plus cost but no more.

Deal! 

Let´s see and remember these posts..
On the chinese website, datasheet and quick start guide are avaible for download, with pics from the backside.
Only difference to the 2000 are the missing wavegen output and the fan seems to be another.
And 4 mathtraces....boy thats killing me. 

[tautech]

We'll see, my pick is SDS2102X Plus cost but no more.

Deal! 

Let´s see and remember these posts..
On the chinese website, datasheet and quick start guide are avaible for download, with pics from the backside.
Only difference to the 2000 are the missing wavegen output and the fan seems to be another.
And 4 mathtraces....boy thats killing me. 

There is every chance the 4 Math channels in this HD is new SW development that will later be rolled out into older models. To the observant there are constant little changes in many models.
But of course development costs have been reduced by using the 2kX HD enclosure just as they used the 2000X rear case for SDS2kX Plus.
I can see a target price set to continue to attract buyers by the # of savings from using another existing case and external option HW in the development of this DSO.

[Martin72]

I am curious whether it should come so.
Actually, it was said that only 2 go for performance reasons - even the lecroy wavesurfer 4000HD produced by siglent also has only 2 channels.
Not that one decided now because of the competition, we make but 4 channels possible.

As for the 12bit, with the 1000 and 2000:
Another speculation of mine, this is an ADC from the shelf of lecroy.
They had once brought a Waverunner 6Zi on the market, with 12bit resolution and, unusual for Waverunner, a maximum sample rate of "only" 2GSa/s...

[ojete]

Martin72 you have the SDS2104X HD, after watching the quick guide of the SDS1000X can you say if everything is the same?

https://www.siglent.com/upload_file/document/SDS1000X%20HD_Quick%20Guide.pdf