REVIEW - Siglent SDS2000 series - A Comparison of Features with Rigol DS2000

[marmad]

Table of contents:

00:00:00: Unpacking
00:05:04: Initial impressions of look/feel, boot-up, and display
00:08:01: Some thoughts on physical layout
00:12:40: Comparison with Rigol DS2000 = Measurement features
00:16:28: Comparison with Rigol DS2000 = Cursor features
00:19:26: Comparison with Rigol DS2000 = Trigger packages
00:22:58: Comparison with Rigol DS2000 = Math features
00:25:54: Comparison with Rigol DS2000 = Delayed Sweep
00:29:25: Comparison with Rigol DS2000 = Pass/Fail utility
00:31:24: Color temperature grading
00:32:33: Frame (Segment) playback: history mode
00:37:02: Comparison with Rigol DS2000 = Manual and automatic frame playback
00:39:07: Frame (Segment) recording: acquire mode
00:44:15: Comparison with Rigol DS2000 = Frame analysis
00:48:10: Bug involving waveform update rate
00:49:01: Comparison with Rigol DS2000 = Bus decoding (I2C)
00:59:45: Comparison with Rigol DS2000 = Bus decoding on recorded frames (segments)
01:02:12: Comparison with Rigol DS2000 = Trigger Out delay and jitter
01:03:55: Bug (?) involving sin(x)/x - linear interpolation
01:05:12: Table of waveform update rates using vectors
01:10:41: Comparison with Rigol DS2000 = Waveform update rates @ 1ns/2ns/5ns/10ns
01:12:48: Comparison with Rigol DS2000 = Waveform update rates @ 200ns/500ns/1us/2us/5us
01:13:48: Comparison with Rigol DS2000 = Waveform intensity adjustment
01:14:25: Intensity grading anomalies
01:17:55: Final thoughts

(Siglent FW: 1.1.1.35.1 - HW: 5.3 / Rigol FW: 3.0.1.3 - HW: 1.0.1.0.0)



EDIT: Here's an expanded (and modified) version of the vector waveform update chart that appears in the video:




EDIT: Here's a very quick table I made (there might be errors) of the Siglent's frame capturing speeds (when not displaying the waveform). They're quite impressive, especially when set to linear interpolation. The problem, as mentioned in the video, is that Siglent still hasn't created reasonable tools for (at a minimum) moving around and examining large amounts of recorded frames - or better still, pass/fail or some other kind of an analysis routine:

[Ivan7enych]

Great review!

It would be interesting to test and compare FFT function, frame rate, averaging options (if any).

[marmad]

Great review!

It would be interesting to test and compare FFT function, frame rate, averaging options (if any).

Thanks!

I mention in the video that the FFT features seem completely identical to those of the Rigol - when I tested it I couldn't see any difference, so I left it out.

I'm not quite sure what you mean precisely with frame rate and averaging options:

Do you mean the rate that the Siglent can capture frames? If so, it will be faster than the waveform update rate at each timebase since the DSO doesn't display frames as they are captured (300k is the advertised fastest speed). It's possible that it's already been measured by someone else and posted, but if you give me an idea of what you'd like to know, I can check it.

Do you mean averaging measurements (in statistics)?

[nctnico]

Can you post a summary in text? Watching a >one hour video is a bit long for that.

[marmad]

Can you post a summary in text? Watching a >one hour video is a bit long for that.

That's why there's a table of contents - you can just look at the portions that are of interest to you.

[nctnico]

I lack the attention span for listening to everything so I probably skip over the part I'm interested in  Try to cut that video down to 5 minutes.

[Teneyes]

Great review, and a fair unbiased discussion.  I always learn new features to use from your Reviews.
Thanks , I hope you get to used the Siglent a bit more.

Interesting that  Decoding of recorded segments is not included in the Siglent, almost looks like reverse engineering of the earlier Rigol firmware before the bugs were fixed. 

[rf-loop]

waveform update rates
SDS2304
(in this test FW version was 1.1.1.26.5 what is today totally obsolete)

[marmad]

waveform update rates
SDS2304
(in this test FW version was 1.1.1.26.5 what is today totally obsolete)

My table is comparing the waveform update rates between the two DSOs while using vectors, not dots. Update rates while using dots are substantially higher at some timebases.

[Ivan7enych]

Do you mean averaging measurements (in statistics)?

I have FFT averaging in old TEC, it makes FFT noise floor much lower.

Here is simple comparison, triangle signal 10mV peak-to-peak is passed via BNC-BNC cable from signal generator to one oscilloscope, than to another.
TEC no averaging - 4-5 harmonics are visible above noise
TEC with 8x averaging - 8 harmonics are visible
Rigol 2072 - 4-5 harmonics are visible
If Rigol have at least 10240points FFT window (instead of 2048 regardless mem depth setting), it will have much better resolution and accuracy. But now it looks like a toy, not a useful feature. That's why I'm asking how it is going on Siglent.

I mean FFT graph update framerate, in TEC it is 1-2 frames pes second, in rigol - 3-4 frames/s

[W7NGA]

marmad,

after watching your insightful and in-depth review I felt compelled to say thank you and let you know it is greatly appreciated.

daniel W7NGA
MSEE, Stanford University

[Siglent]

marmad, thanks for the review! It is helpful.

[tautech]

Can you post a summary in text? Watching a >one hour video is a bit long for that.

Worth the watch though. 

@marmad
Many questions revealed and answered. 
You should be congratulated for your impartial and honest review. 

Could we ask, how long did this review take? hours, days, weeks?

[rf-loop]

waveform update rates
SDS2304
(in this test FW version was 1.1.1.26.5 what is today totally obsolete)

My table is comparing the waveform update rates between the two DSOs while using vectors, not dots. Update rates while using dots are substantially higher at some timebases.

In normal use and if want catch glitches, what is main reason for want fast waveform update rates, there is no any advantage for use lines.  So, why select line connected dots instead of dots alone.

Of course in this case, if hunting clitches or other fast random things,  user use settings what give best reesult. Specially because dots connected fake lines do not give any markable displayed visual advantage. So, for test wfm update rate and compare, of course it need use settings what give best result. (also in Siglent table what I have is automatic measurements on and cursors).
So, it is max ~110k / s (but this is not maximum peak value, peak value is more)

About seqmented memory capturing. Here is Siglent table.

[BravoV]

Mark, thanks a lot for the review ! 

[marmad]

I have FFT averaging in old TEC, it makes FFT noise floor much lower.

Here is simple comparison, triangle signal 10mV peak-to-peak is passed via BNC-BNC cable from signal generator to one oscilloscope, than to another.
TEC no averaging - 4-5 harmonics are visible above noise
TEC with 8x averaging - 8 harmonics are visible
Rigol 2072 - 4-5 harmonics are visible
If Rigol have at least 10240points FFT window (instead of 2048 regardless mem depth setting), it will have much better resolution and accuracy. But now it looks like a toy, not a useful feature. That's why I'm asking how it is going on Siglent.

I mean FFT graph update framerate, in TEC it is 1-2 frames pes second, in rigol - 3-4 frames/s

Ok, Ivan, I understand now - I will make some tests today and post the results later.

[marmad]

marmad,

after watching your insightful and in-depth review I felt compelled to say thank you and let you know it is greatly appreciated.

daniel W7NGA
MSEE, Stanford University

Thanks for your kind words, Daniel!

Mark

You should be congratulated for your impartial and honest review. 

Could we ask, how long did this review take? hours, days, weeks?

Thank you, tautech.

I did the unpacking at the end of July, but then work came up and I had to postpone testing for about 2 weeks. I then filmed the video in many pieces over 10 days (there is a lot of material I didn't use). I tried to only use natural light to minimize the effects of artificial lighting on the displays (there is only one shot in the video using artificial light), but the effect of this is that the brightness in the room varies a lot because of sun/shade/clouds/etc. Also, because of the way the camera adjusts for exposure, it makes the brightest areas on the Siglent hard to read - although they aren't that way in reality.

I think the Siglent has a nicer display than the Rigol - but I wish I could turn the overall brightness down sometimes, just as I wish I could turn the overall brightness of the Rigol up sometimes. I'm not quite sure why the manufacturers don't include that function, but perhaps it has to do with possible inverter noise from adjustable supplies.

Mark, thanks a lot for the review ! 

You're welcome, BravoV!

[marmad]

In normal use and if want catch glitches, what is main reason for want fast waveform update rates, there is no any advantage for use lines.  So, why select line connected dots instead of dots alone.

Of course in this case, if hunting clitches or other fast random things,  user use settings what give best reesult. Specially because dots connected fake lines do not give any markable displayed visual advantage. So, for test wfm update rate and compare, of course it need use settings what give best result. (also in Siglent table what I have is automatic measurements on and cursors).

It's true that if you're specifically hunting glitches, it would make sense to use the settings that give you the fastest waveform update rate. But sometimes when I'm just looking at signals, I see a glitch that I didn't know existed beforehand.

I don't know how other people use their DSOs - I can only speak for myself - but as I mentioned in the video, normally when I'm looking at signals or doing some basic debugging, I'm using my Rigol with vectors ON 99% of the time - since I'm switching between slow and fast time bases and I want to see a full waveform, not a collection of sample points, without having to continually access the vectors/dots switch. But, yes, there are certain times I will specifically use dot mode.

In dots mode, the Siglent is much faster than the Rigol at most timebases - except those <= 20ns in single channel mode - and <=5ns in dual channel mode (as long as you have interpolation set to linear, as you did when you made your waveform update table).

As I tried to point out in the video, there is either a bug - or some strange manifestation of Siglent's implementation - that causes the following behavior: EVEN when in dots mode on the Siglent, if you have interpolation set to Sin(x)/x it drastically reduces the waveform update rates when at timebases <= 20ns:

About seqmented memory capturing. Here is Siglent table.

As noted above, I made frame capturing measurements very quickly - 25MHz sine from the AWG, Sequence SINGLE capture, go to History List and count the microseconds - so it's quite possible I made a math error - my brain can certainly do that    But in any case, the speeds are very impressive - it's a great idea by Siglent for very fast capture rates - I just think that the user should have the option to turn Display-When-Capturing ON/OFF for slow/fast occurring triggers; i.e. with slowly arriving triggers it would be good to have the frames displayed in-between captures.

[tautech]

You should be congratulated for your impartial and honest review. 

Could we ask, how long did this review take? hours, days, weeks?

I did the unpacking at the end of July, but then work came up and I had to postpone testing for about 2 weeks. I then filmed the video in many pieces over 10 days (there is a lot of material I didn't use). I tried to only use natural light to minimize the effects of artificial lighting on the displays (there is only one shot in the video using artificial light), but the effect of this is that the brightness in the room varies a lot because of sun/shade/clouds/etc. Also, because of the way the camera adjusts for exposure, it makes the brightest areas on the Siglent hard to read - although they aren't that way in reality.
I think the Siglent has a nicer display than the Rigol - but I wish I could turn the overall brightness down sometimes, just as I wish I could turn the overall brightness of the Rigol up sometimes. I'm not quite sure why the manufacturers don't include that function, but perhaps it has to do with possible inverter noise from adjustable supplies.

Just as I thought, a massive job, well done.
You have set a bench mark for others to follow.

I have my SDS2304 on a shelf at eye level, at arms length and shaded by the shelf above and consequently the brightness is never a problem, or maybe I should remove the screen protector film. hehe
As you used natural light, you may find as I have, in dark sites away from the bench, that the overall brightness need not be set quite so high, but great for my old eyes.
I imagine this adjustment would need to be controlled in HW, so maybe a feature for future models. 

A couple of points that where raised in the Video: (no order just jotted down as seen)
Button silkscreen: Mine is perfect after 9 mths hobby and loan use.
Sequence menu: There is plenty to be improved, and I observe there are 3 unused buttons in the top menu that could be possibly dedicated to navigation functions.

It is good to see Siglent acknowledge your comparison.
They have a bit to do it seems.

[marmad]

I have my SDS2304 on a shelf at eye level, at arms length and shaded by the shelf above and consequently the brightness is never a problem, or maybe I should remove the screen protector film. hehe

Yes, one point I mis-stated in the video was when I said the Siglent, like the Rigol, is a bench-top scope - not a shelf scope. But of course, as long as the shelf is around eye-level, it works perfectly that way.

As you used natural light, you may find as I have, in dark sites away from the bench, that the overall brightness need not be set quite so high, but great for my old eyes.

Yes - but I think, in general, I prefer the slightly too-bright display of the Siglent to the slightly too-dim display of the Rigol.

Button silkscreen: Mine is perfect after 9 mths hobby and loan use.

To be fair, I used a borrowed DSO from my dealer (i.e. not a brand new model straight from the factory), so it's possible that one little piece of silkscreen got damaged somehow after the fact. Since I've done a lot of silk-screening myself over the years, I can see that the quality of the silkscreen is not quite as good on the Siglent than the Rigol - but it's a small difference and not a big deal. Unfortunately, I can't really show it adequately because I don't have a lens for shooting extreme close-ups.

Sequence menu: There is plenty to be improved, and I observe there are 3 unused buttons in the top menu that could be possibly dedicated to navigation functions.

Since I use the frame (segment) capture feature of my Rigol quite a lot, this is the one area of operation that's of primary concern to me. All of the other differences between the two DSOs, in terms of features and performance, are negligible, IMO, compared to this one area. If Siglent can improve the operation and tools of the Sequence function to something more similar to the Rigol (obviously they don't have the Navigation knob of the Rigol, but other firmware tools can be added), I think it would be a much more compelling competitor (for people that only need 2 analog channels).

[Teneyes]

A Bit expanded topic, Mark, did you think of testing the Siglent to PC interface?
What changes would it take for RUU  to be SUU and captures frames from the SDS2000.

[Carrington]

Finally, a detailed comparison, great!
As I said long ago, Marmad is the ideal candidate to do that.

Many people were expecting for something like that.
Now to see if Siglent takes note and improves/modifies the firmware...

Thanks Mark. 

[marmad]

I have FFT averaging in old TEC, it makes FFT noise floor much lower.

Here is simple comparison, triangle signal 10mV peak-to-peak is passed via BNC-BNC cable from signal generator to one oscilloscope, than to another.
TEC no averaging - 4-5 harmonics are visible above noise
TEC with 8x averaging - 8 harmonics are visible
Rigol 2072 - 4-5 harmonics are visible
If Rigol have at least 10240points FFT window (instead of 2048 regardless mem depth setting), it will have much better resolution and accuracy. But now it looks like a toy, not a useful feature. That's why I'm asking how it is going on Siglent.

I mean FFT graph update framerate, in TEC it is 1-2 frames pes second, in rigol - 3-4 frames/s

Ivan,

First of all, I'm not sure if you're aware of the following, but you can use the Rigol's (or any DSO that has it) Average acquisition mode to the same effect as the Average setting in Math on your Tek - see the first image - I'm using Averaging of 128, and there 9 harmonics visible.

Also, Rigol has the anti-aliasing function available inside of FFT, and although they never explain exactly what it's doing in the manual, I think it's some kind of averaging (perhaps both sample AND waveform) - see the second image.

The updating of the FFT overlay seems about the same speed on the Siglent as on the Rigol.

Go to this post.

Next, I don't know what I'm doing wrong, but I can't get the Siglent to give me images with the harmonics - perhaps someone with more experience with it can help out?

The third image shows first trying with the same 10mVpp Triangle wave as used on the Rigol (I tried using Normal acquire as well as Average-128 - both looked the same).

The fourth image shows trying with a larger 3Vpp Triangle wave and Average-128.

Can any Siglent users/dealers help out?

[marmad]

A Bit expanded topic, Mark, did you think of testing the Siglent to PC interface?
What changes would it take for RUU  to be SUU and captures frames from the SDS2000.

I haven't looked at the PC interface, Teneyes; perhaps I'll do it over this weekend. It's a little early to know about the possibility of modifying RUU to work with it - we're still waiting for Siglent to publish the SCPI documentation.

Thanks Mark. 

My pleasure, Carrington.

[Ivan7enych]

First of all, I'm not sure if you're aware of the following, but you can use the Rigol's (or any DSO that has it) Average acquisition mode to the same effect as the Average setting in Math on your Tek - see the first image - I'm using Averaging of 128, and there 9 harmonics visible.

Also, Rigol has the anti-aliasing function available inside of FFT, and although they never explain exactly what it's doing in the manual, I think it's some kind of averaging (perhaps both sample AND waveform) - see the second image.

The updating of the FFT overlay seems a little faster on the Siglent than the Rigol (like all of the overlays) - perhaps 5-6 frames/s?

Next, I don't know what I'm doing wrong, but I can't get the Siglent to give me images with the harmonics - perhaps someone with more experience with it can help out?

The third image shows first trying with the same 10mVpp Triangle wave as used on the Rigol (I tried using Normal acquire as well as Average-128 - both looked the same).

The fourth image shows trying with a larger 3Vpp Triangle wave and Average-128.

Can any Siglent users/dealers help out?

Thank you!

Average-16 acquisition mode really helps to see more details. Also I've found that HiRes mode helps as well, but if I turn on antialiasing in acquire menu - it produces many false spikes on FFT. It looks like FFT gets points after aliasing processing.

On Siglent images there is much different FFT scale (100MHz per division), thats why it shows nothing with 2.5KHz signal. May be Siglent samples data with 2Gs/sec and sends to FFT only first 2048 points from the captured 14Mpoints frame, while Rigol get 2048 points uniformly distributed in the sampled frame (14Mpoints).