Siglent SDS1000X HD 12bit 2GSa/s Oscilloscope Unboxing & Test

[2N3055]

I am not sure what restrictions result from having ERES and Averaging as Math functions only. The fact that they take up one Math channel per measurement channel when used seems like a clear disadvantage. And I don't follow in which way it would be better to have them as Math function only?

I keep repeating that these math channels have complex math. So you can have ERES, Filter and FFT for one channel (and more) applied as a complex formula, all simultaneously on single math channel.

Using ERES as a math have you keep original data as sampled - so you can have original data and 4 different transformations....

You might want to run FFT on original data on 2 channels , an ERES in time domain on one noisy channel multiplied with other channel that has Average on it to calculate power... that would be 3 math channels.
You can have a sampled buffer (stopped mode) and apply math ERES after the fact, changing ERES depth interactively, until it is just right..

ERES and average as sample mode is global setting for all channels.  That limits you to what can you do.

I know that I mostly use ERES as math function for quite some time. And don't even use so much ERES because of 12 bit and low noise. You use it only when you deliberately want to filter out higher frequencies in signal, and not to work around scope own noise or lack of resolution.

These scopes also have proper vertical zoom too.

[Martin72]

Hi,
With my SDS2104Xplus, I was annoyed to have to "sacrifice" a math channel for average or eres, because the scope only has two and one is already gone.
Then there was the fact that you then had both signals on the screen, which was also annoying.
This was solved quite quickly with the "Hide" function.

When the 2000X HD came out, it also only had two math channels, but average and eres as acquisition mode.
I thought it was great not to have to use either of the two precious math channels for this, despite the advantages mentioned above, if you do it via them.
Well, now I have both.
The use of eres and average via the acquisitionmode also has a small disadvantage:
This then applies to ALL channels, but sometimes you don't want this.

[Martin72]

I don't think so.
The SDS2000Xplus has neither 4 math channels nor digital filters and the necessary memory management, precisely for limiting reasons.
The 1000X HD has these functions.

Edit:
User "idolclub" could post a few more pictures with the scope in action.
To finally answer the bandwidth question, it would be enough if he activates 2 (direct) channels at the same time, which is then shown in their infoboxes.
Then we would know one hundred percent.

[2N3055]

I appreciate the strong brand loyalty and the protective reflexes. I would appreciate even more if someone could answer my questions re. what the SDS1000X HD actually does...

In the meantime I found an updated Chinese datasheet, https://www.siglent.com/u_file/download/23_12_25/SDS1000X%20HD_Datasheet_CN01E.pdf. The "200 MHz in 2-channel mode only" restriction has disappeared; it was probably just a mistake in the earlier datasheet edition, carried over from the old 1 GSa/s version of the scope.

There is still no mention of ERES and Averaging as acquisition modes; but again, maybe that is just an omission in the datasheet. Given how close the 1000X HD is to the 2000X HD in almost all software aspects, I was simply wondering about the difference. Maybe when @idolclub comes back online he can have a quick look at the actual product.

EDIT: Found the Chinese user manual and convinced Google to translate the relevant section. Averaging and ERES are indeed not supported as acquisition modes, it seems; only Normal and Peak Detect are on offer. See the attached figure from the manual.

For context, in case you have missed it: I have just returned my DHO1074 and am eagerly awaiting the release of the SDS1000X HD, hoping that it will come at a price I can justify to myself for hobby use. Trying to learn as much as I can about that scope in the meantime.

Please don't take this the wrong way, but would you please stop with these comments about brand loyalties and protective reflexes. You are completely off with that to the point to start being offensive. It starts to bother me. It is completely uncalled for.

You asked for someone to confirm if SDS1000XHD has ERES acquisition mode. I don't know because I don't have scope nor crystal ball. So I'm not going to answer what I don't know. Since I have no interest in that particular scope, I'm not going to make further research. Sorry, cannot help you with that.

You also made confused/inaccurate comments on usage of ERES on math channels and asked what would the benefits be. I answered that, because I know it, and actually use it every day. I have not 1 but 2 Siglent scopes of this family as daily drivers and am very familiar how it works.

Maybe you would trust my facts better if they came from Rigol user?

We are trying here to give you good data so you can make educated decision.
I'm not trying to convince you anything. Just to help you make  purchasing decision that is appropriate for you.
What you buy in the end makes no profit to me in any way, one way or another.

Those little (humorous ?) jabs are not appropriate if people are trying to help you.
If you don't want help from RF or me or anybody else, please just say so. So we don't waste our time just to be insulted as we are trying to help.

As a more constructive answer to your other questions, I also have no idea as to general performance of the "black" SDS1000XHD compared to SDS200XHD, for same reason. I find speculating about these things waste of time. We simply don't know until someone gets their "mittens" on it.
But what I know, is that I would wait for "rest of the world" release before we pass judgment. When 6000 series came out, there was lots of confusion about models because China only models had several hardware versions that never came to rest of the world.

Also all this talk about BOM optimization is not exactly correct. BOM optimization is exactly that: optimization. Difference lies in what optimization criteria is prioritized.  You must understand that parts acquisition channel in China works differently than in West. Pricing is different, volumes are different. Many times it is not about cheapest possible parts, but about availability.  Also Chinese companies don't have to have 1200% profit on BOM. They are being run differently.   
So without very detailed insight in Siglent inside strategy and structure of how things are done, we cannot make any predictions of any value. I know I don't have those details, and I doubt many outside inner circle have..
Don't get me wrong, speculate away if that makes you happy, but don't expect to be close to reality.. You simply have no relevant data...

Best

Sinisa

[rf-loop]

The use of eres and average via the acquisitionmode also has a small disadvantage:
This then applies to ALL channels, but sometimes you don't want this.

This!

( Is it even small disadvantage...    )

[2N3055]

Please don't take this the wrong way, but would you please stop with these comments about brand loyalties and protective reflexes. You are completely off with that to the point to start being offensive. It starts to bother me. It is completely uncalled for.

Sorry if I caused offense; I did not mean to. But maybe you don't realize how quickly one triggers responses which seem to "defend" Siglent products.

In the present case, I simply asked whether two particuar features were in the SDS1000X HD or not. The answer from you as well as rf-loop was "here's why what Siglent has implemented is good" -- without even knowing (or answering) what Siglent actually has implemented.

You also made confused/inaccurate comments on usage of ERES on math channels and asked what would the benefits be.

Where exactly did I do that?

Maybe you would trust my facts better if they came from Rigol user?

Look, now it is my turn to get annoyed. I am neither in the "Rigol camp" nor the "Siglent camp"; I am just trying to find the right scope for my needs. (And in the present case have already concluded that the Rigol DHO1000 is not it.) I think I have made it quite clear in  many posts that I don't take sides -- or at least don't take the same side always. But apparently that's not good enough -- "You are either with us or against us." 

We simply don't know until someone gets their "mittens" on it.

Right. And if you re-read my post #50, you will see that a simple answer from someone who has the scope already in their hands is exactly what I was hoping for.

Good. As I hoped it was not meant to be offense. But still I do inform you that I don't like it. It was uncalled for.
I was not defending or protecting anything. Simply stating facts.

I am not sure what restrictions result from having ERES and Averaging as Math functions only. The fact that they take up one Math channel per measurement channel when used seems like a clear disadvantage. And I don't follow in which way it would be better to have them as Math function only?

My math comments are a reply to this..

And you are again being adversarial.  You stated that my comment was "reflexive triggered response of defenders or Siglent products". How is that not a jab at someone answering to some of your questions.
And I rightfully asked what do you want. An answer from someone who doesn't use Siglent products or those that know it in detail?
If you think my answers are not truthful and are just propaganda, please let me know so I don't bother.... That is your right.
I didn't attack you. You attacked me, calling me names.

And enough of this Siglent/Rigol camp. I don't belong to any camps. When I was saying good things about Picoscopes I was accused of being Picoscope shill.
I was the one that asked BUD to stop with endlessly repeating stuff about Yaigol PLL problems on DS1000Z. But was also one of the people who have found, verified and demanded RMS bug on DS1000Z to be fixed. While one of my now blocked "friends" was attacking me that I should shut up and just accept that is not a bug but an "imperfection" and "in specifications" and I should shut up and questioning my "agenda". Luckily we persisted and Rigol finally fixed that one. And he kept behaving like that, constantly attacking me personally for anything he didn't like.
So I have a lot of "emotional baggage" when people attack me for no reason.
You must appreciate how I must feel like an idiot after spending 10 minutes of my time to write detailed explanation of how math works just to greeted by "it is amazing how fast one seems to trigger responses which seem to defend Siglent".  Yes, that "took a jelly out of my doughnut"..

As for your post #50, you must understand that no person will go 50 posts back and read everything you ever wrote on topic to get full context.
If I respond to post #64 it is going to be response to #64, unless you specifically referenced something in a post.. And I believe most people will be the same.
Sorry for misunderstanding if any was there.
 
Best,

[rf-loop]

So please elaborate. Which datasheet exactly and where?

I was referring to the machine-translated English version in tautech's post, https://www.eevblog.com/forum/testgear/siglent-sds1000x-hd-coming/. That datahseet describes the updated 2 GSa/s version of course; otherwise I would not have asked about the bandwidth restriction.

It's a little awkward when these are published a little too early and then peoples read these like infinite truth.

In addition, it should be emphasized that  these (early info picked-up data from chinese web sides) need read with big reservations.
Especially in a situation like this, where this model has been just released over older version.
(But naturally root reason is Siglent when they publish data sheet before enough carefully revieved.)

I have one definite opinion. These must be removed from distribution. Just so that these false notions don't circulate, which then remain around forever. There IS an error in that data sheet.

According to my understanding (read: my current knowledge), the correct one is this (for next generation, dark grey  "black", SDS1000X HD):

Vertical (analog)
Channel
4 + EXT (4CH Series: SDSxxx4X HD)
2 + EXT (2CH Series: SDSxxx2X HD)

Bandwidth
(-3 dB) @ 50 Ω
SDS1104X HD, SDS1102X HD: 100 MHz
SDS1204X HD, SDS1202X HD: 200 MHz

Flatness@50 Ω
DC- 10% (BW): ± 1 dB
10%- 50% (BW): ± 2 dB
50%- 100% (BW): + 2 dB/-3 dB

Rise time@50 Ω
(typical)
Typical 3.5 ns (SDS1104X HD, SDS1102X HD)
Typical 1.8 ns (SDS1204X HD, SDS1202X HD)
------------------
And this note: *1: 200 MHz in single and dual channel modes, 100 MHz in other modes is already removed from preliminary english data sheet because it was (and is) only true in 1st generation version SDS1000X HD. This note do not belong at all to second generation SDS1000X HD. But because this model is not released for western markets there is also not current english  data sheet published.



And simplified:

SDS1204X HD, SDS1204X HD (dark grey "black" version)

BW:  200MHz when 1, 2, 3 or 4 channels are used simultaneously.

[rf-loop]

Thanks, rf-loop. As mentioned above, I found the updated Chinese data sheet (Rev. E) in the meantime, and it no longer mentions the bandwidth limitation for > 2 channels. Attaching a Google-translated copy here; maybe Tautech can update his first post in the SDS1000X HD introduction thread when he is back home.

This is afaik, just reason why he have not updated it. 

[idolclub]

Two more questions to those with early access to the SDS1000X HD (or maybe those who can read a Chinese user manual):

The datasheet mentions that 200 MHz are only available with up to two channels, and the scope drops back to 100 MHz with three or four channels active. Is that really how Siglent implemented it? That would be a rather conservative approach.

Also, the dataheet mentions ERES and Averaging only as Math functions; nothing about these modes in the Acquisition section. Have they in fact restricted this vs. the 2000X HD, or is it just an omission in the datasheet?

Thanks for any insights you might have!

Hi,

In SDS1000X HD, the ERES and Average  do not exist in Acquisition modes, but are placed in MATH Menu.

[idolclub]

I don't think so.
The SDS2000Xplus has neither 4 math channels nor digital filters and the necessary memory management, precisely for limiting reasons.
The 1000X HD has these functions.

Edit:
User "idolclub" could post a few more pictures with the scope in action.
To finally answer the bandwidth question, it would be enough if he activates 2 (direct) channels at the same time, which is then shown in their infoboxes.
Then we would know one hundred percent.

10MHz Leo Bodnar testing with two and four channels enabled on the SDS1000X HD.

[Martin72]

Maybe yes, maybe no, who knows.
It is what it is and ultimately only a teardown including removed heat sinks can shed any light on this.
We already know the 2000X HD board without heat sink.

[tautech]

Thanks, rf-loop. As mentioned above, I found the updated Chinese data sheet (Rev. E) in the meantime, and it no longer mentions the bandwidth limitation for > 2 channels. Attaching a Google-translated copy here; maybe Tautech can update his first post in the SDS1000X HD introduction thread when he is back home.

I will when we have the EN release version in a couple weeks.

Until then attached is the older and now outdated earlier white model version.
As has been discussed the sampling rates are now improved with all channels active for the forthcoming black models.

[2N3055]

Many thanks for the answers and additional data, @idolclub!

I couldn't help myself and scrutinized the datasheet a bit more, comparing to the SDS2000X HD. I notice that the ERES and Averaging are no longer advertised as being "hardware accelerated", and assume that is the underlying reason for having to remove these as acquisition modes: Presumably the (global) acquisition modes are hardware-accelerated in the 2000X HD, while the Math operations are done in software, at least in the 1000X HD?

Maybe I am reading too much into this, but it is all consistent with the assumption that the 1000X HD -- while having similar CPU horsepower as its bigger brother, and hence able to e.g. support 4 Math channels -- has a smaller FPGA on board, and is hence limited in its real-time processing capabilities. Maybe the "fixed memory depth" option in the memory management settings had to be left out for the same reason?

If you take a look at that translated Chinese V2 hardware datasheet (the actual scope that is coming to EU), and compare it to SDS2000XHD there is some interesting data inside.

First, math is accelerated. In addition to FPGA acceleration there is a DSP block in CPU for just such occasions too.
FFT is FPGA accelerated, for instance.

One interesting number inside datasheet is Waveform capture rate... V2 SDS1000XHD has higher numbers that are in a ballpark with 2000XHD. Even faster a bit, because it has less memory. That certainly does not point out to a less capable FPGA..
Or does it? I don't know.
I guess we will know soon enough...

[2N3055]

Thanks, rf-loop. As mentioned above, I found the updated Chinese data sheet (Rev. E) in the meantime, and it no longer mentions the bandwidth limitation for > 2 channels. Attaching a Google-translated copy here; maybe Tautech can update his first post in the SDS1000X HD introduction thread when he is back home.

I will when we have the EN release version in a couple weeks.

Until then attached is the older and now outdated earlier white model version.
As has been discussed the sampling rates are now improved with all channels active for the forthcoming black models.

Tautech,

If you don't mind me asking, is there a reason for this datasheet here?

It is a datasheet to a scope nobody will ever see, that will only add to confusion, if you ask me...
It is not applicable to this scope in this topic... It is literally something else despite name being same.

There was enough confusion, speculations, flaming and all kinds of non-useful stuff in that original topic based on people being confused by specs..
I would rather we mark as "Do not use" all the versions of V1 d.s. in circulation.

If it is really important to discuss something I think it would be better to use an translated unofficial Chinese d.s. for actual hardware V2. That one might have errors (by virtue of not being official and officially released yet) but is at least addressing right hardware...
I personally would not circulate even that, until there is official d.s. by Siglent in English, but that is me..

Don't get me wrong, I'm not policing what you do, just saying..

Best,

[2N3055]

First, math is accelerated. In addition to FPGA acceleration there is a DSP block in CPU for just such occasions too.
FFT is FPGA accelerated, for instance.

Alright -- "software accelerated" is not to be sneezed at.   As mentioned, my guess is that the CPU capability of the 1000X HD is the same as in the 2000X HD, and that the Math capabilities (and speed) are the same.

One interesting number inside datasheet is Waveform capture rate... V2 SDS1000XHD has higher numbers that are in a ballpark with 2000XHD. Even faster a bit, because it has less memory. That certainly does not point out to a less capable FPGA..
Or does it? I don't know.

Not a slower, but a smaller FPGA is my guess. The lack of a built-in AWG and logic analyzer have probably saved some gates already, and I assume that Siglent trimmed some more functionality so they can get away with a lower-cost part.

Why else would they leave out fixed memory and ERES/Averaging acquisition modes? It's not the type of functionality one would deliberately disable in a lower-end model to differentiate the higher-end 2000X HD, but rather "deep in the datasheet" functions.

I guess we will know soon enough...

Yes, I agree with you (and Martin before) -- it's all a bit idle speculation. But what else is a hobbyist keen on a new scope going to do? I'm not good at waiting, it seems. 

Yeah, it all good fun, just not to take it too seriously...

[Performa01]

Scientific scopes from LeCroy never had any acquisition modes that would destroy the original sample data, hence not even Peak Detect (but sufficient sample memory instead). Average and ERES always have been math functions only.

Siglent, on the other hand, even though cooperating with LeCroy, used to follow different philosophies and offered Average and ERES as acquisition modes. Up to late 2016, these modes were barely usable, because they were only available with short memory like 14 kpts, strongly reminding on the memoryless storage scopes from Tek at that time. Starting with the SDS1202X-E, we got these acquisition modes with 1.4 Mpts memory, yet not HW accelerated. Even more important, these models don’t have 16 bit processing, thus not actually giving any resolution enhancement with either Average or ERES.

The SDS2000X Plus has limited resources, yet it has a HW-accelerated 10-bit mode, which is equivalent to ERES1.0 – and the internal 16 bit interface. There is no noticeable speed penalty compared with 8 bit mode, yet there are no FPGA resources left for generic ERES or Average acquisition modes. As a consequence, we finally got these as math functions, which has a number of advantages, even though it’s considerably slower; we can apply it only to the channels we actually need it, can combine it with other operations – for instance we could use average and ERES at the same time for the same trace.

The SDS6000 and later the SDS2000X HD have sufficient resources, thus providing both the HW-accelerated acquisition modes and the slower math functions for Average and ERES.

[2N3055]

Thank you for the excellent overview, performa01!

The SDS6000 and later the SDS2000X HD have sufficient resources, thus providing both the HW-accelerated acquisition modes and the slower math functions for Average and ERES.

So speed-wise, the SDS1000X HD implementation will be a step back, since it will offer the (slower) Math functions only, right?  I do understand that flexibility-wise, being able to apply ERES and/or Averaging to specific channels only is an advantage.

I appreciate the "LeCroy philosophy" of leaving the original sample data unchanged at all times, to be able to apply various analysis operations on them.  On the other hand, a lot of my oscilloscope use is not highly analytical, but interactive -- so I also appreciate Siglent's efforts to strike a good balance between analytical accuracy and fast interactive response. Offering two ways to apply ERES and Averaging in the SDS2000X HD is a great way of supporting both use cases. But I can understand that "something has to give" in the 1000X HD to meet cost targets.

On Keysight MSOX3104 I use Hi-Res a lot. It is 8 bit, noisy (1GHz BW), and no less than 5mV/DIV in hardware.

On SDS2000X-HD there is hardly need for ERES for interactive work, unlike 8 bits scopes.
Visibly, on a screen there is barely a difference.
SDS1000X HD are very low noise and 12 bit, and only 200 MHz nominal BW.
For slow signals, where you want to filter out high frequency noise you simply enable 20 MHz BW limit and that is it.
I actually very rarely use ERES as acquisition mode. It is of no use for FFT, and even with the vertical zoom, scope will produce very detailed, thin traces without ERES.
ERES helps more with scopes that have large BW, suppressing more of the noise in higher frequencies.

[Wintel]

Does the SDS1000X HD have AC50 ohm inputs?

[2N3055]

Does the SDS1000X HD have AC50 ohm inputs?

According to datasheet, yes..

[idolclub]

Does the SDS1000X HD have AC50 ohm inputs?

Yes. There is DC50R and AC50R.

[roy_eedreku]

This scope seems really good and seems like a much cheaper option compared to SDS2000x HD. Does it make sense to buy 2000xHD if I don't need > 200 MHz and I already have a 60 MHz external AWG? I am not sure if I would be using the MSO feature much and Bode plot would be covered by the external AWG.

[idolclub]

This scope seems really good and seems like a much cheaper option compared to SDS2000x HD. Does it make sense to buy 2000xHD if I don't need > 200 MHz and I already have a 60 MHz external AWG? I am not sure if I would be using the MSO feature much and Bode plot would be covered by the external AWG.

SDS1000X HD is a good oscilloscope. You can also check the future pricing of your local dealer before making a decision.

[hpw]

SDS1000X HD is a good oscilloscope. You can also check the future pricing of your local dealer before making a decision.

How does the noise compare to the SDS2000x Plus as picture below running on 10bits... ??

- FFT using "none FFT window) as noise
- 32 AVG
- Input 1M or 50 open shorted as idle spurs to see

Hp

[idolclub]

SDS1000X HD is a good oscilloscope. You can also check the future pricing of your local dealer before making a decision.

How does the noise compare to the SDS2000x Plus as picture below running on 10bits... ??

- FFT using "none FFT window) as noise
- 32 AVG
- Input 1M or 50 open shorted as idle spurs to see

Hp

Hi,

Input 1M and 50 open on SDS1204X HD.

[hpw]

SDS1000X HD is a good oscilloscope. You can also check the future pricing of your local dealer before making a decision.

How does the noise compare to the SDS2000x Plus as picture below running on 10bits... ??

- FFT using "none FFT window) as noise
- 32 AVG
- Input 1M or 50 open shorted as idle spurs to see

Hp

Hi,

Input 1M and 50 open on SDS1204X HD.

Thank you for the measurements 

Conclusions: Much less spurs, as may the result of different used ADC and may also some better RF shielding 

Hp