Rigol DS1054Z vs Siglent SDS1202X-E

[rf-loop]

What is the update cycle on the Rigols? I am leaning towards the DS1054Z at the moment, but fear a newer model might pop up months after purchase. Even though my scope will be just as good as before, it would be nice if a new scope could improve upon some of the shortcomings of the current model. Having a more reponsive UI and better FFT would both be improvements that I would appreciate and those would inevitably be part of a new model.

If you want better responsiveness, better vertical resolution and better FFT the Siglent SDS1202X-E is what you're looking for. But limited to two channels. Also, the Siglent is not yet supported by software such as Sigrok. If you want the best bang for the buck with four channels, then get the Rigol. Both have shortcomings but both are useful tools as long as you are aware of their limitations.

I'm not sure they will release anything in the low end soon. Maybe they'll jump on the Zynq 7000 bandwagon following Siglent and GW Instek? No idea. They recently made a vague announcement of some new electronics for oscilloscopes but it seems to be targetted to a higher cathegory.

At the end of the day you will have to decide wether one of the alternatives in the market suit you and it does at an acceptable price. In the future of course there will be a new release. That is true especially with cut throat competition between the newcomers to the instrumentation market.

Commen mainly trigged due to this colored part.
They have same vertical resolution. But, Rigol have expanded from 5mV/div  to 2 and 1mV/div  reducing vertical resolution. These 1 and 2mV/div are derived from 5mV/div.  Siglent do not use this trick in SDS1000X/X-E series. Down to 0.5mV/div it use full ADC range.

But then ther eis one other resolution thing and this is big difference.
Rigol reduce horizontal mesurement resolution. I do not know exactly what is amount of data what it use. Propably something under 1k or even less.
Siglent X-E use full data. It do not decimate at all. If there is 14M sample length on the RFT and it was 1GSa/s sampled, automatic measuremets resolution 1ns. 
This can test easy and some result is also here in forum.
If some can do test, set horizontal to 1ms/div and one channel. Input 500ns period pulses and pulse width example 50ns and risetime 8ns and fall time 25ns. Now turn measurements on. Rise, fall, width and period.
For Siglent this is easy piece of bisquit. Rigol - err

Rigol do not have waveform history buffer what works justbacround with current wfm/s speed storing every acquisition to memory, amount depending t/div and memory settings. Up to 54M with some settings.
Stop scope and look what just previously happend.

Rigl do not have high speed sequence mode.

Rigol destroy real samople data (well known Sinc joke.
Siglent can always display true raw sample data. In run more and also afterwards linear/Sinc can select, as also dots/lines.
In many cases this is important with digital type signals for go over Sinc function bad effects.

Motly in all things Siglent X-E and X beats Rigol Z in everything. Perhaps Rigol some things on display looks more "finished and polished". Also it is more mature perhaps, or least it have been on market quite long time.  But Siglent is much more severe real tool.
Also I have previously owned and used Rigol 1kZ,  so there is also some real behind my words instead of rumors.

There is one major show stopper thing.  If user want/need 4 channel and this is "must".  Rigol have 4 channel. Siglent 1kX  series have only 2 channels --- for now...

[MrW0lf]

Rigol reduce horizontal mesurement resolution. I do not know exactly what is amount of data what it use. Propably something under 1k or even less.

It would be 600 points (screen) but end-result suggests around 300 points effective. 16kpts FFT (if configured not to use "screen") is only exception AFAIK.

[Mr. Scram]

If you want better responsiveness, better vertical resolution and better FFT the Siglent SDS1202X-E is what you're looking for. But limited to two channels. Also, the Siglent is not yet supported by software such as Sigrok. If you want the best bang for the buck with four channels, then get the Rigol. Both have shortcomings but both are useful tools as long as you are aware of their limitations.

Yeah, I have to remind myself of that occasionally. Whatever scope one picks, it is going to be a neat and very useful tool regardless, and it is still going to be a great tool when a new champion has emerged.

Currently I am leaning towards the Rigol due to the four channels.

That line of thinking never works.

If it did, nobody would ever buy a car (for example).

I don't intend to start a huge discussion, but that is how it often works For instance, in computer, phone and car sales, update cycles are an important factor.

Automotive update cycles are quite predictable and you can be throwing away money if you do not take that into account. Not only consumers, but lease and rental companies also take heed. The drop in sales at the end of the life cycle of a certain model can be an issue for car makers, so they sometimes entice consumers to buy with luxury extras. Sometimes they even explicitly release 'goodbye editions', a sort of ultimate edition at the end of the model cycle. Consumers can choose to have a kitted out model now, or a more bare bones new model later for the same money.

Another example would be the sales of Apple products. If we look at iPhone sales, they are highly cyclical in nature, with sales peaking right after release of the new model.

https://images.dazeinfo.com/wp-content/uploads/2016/04/Apple-iPhone-sales-figures-q2-2016.png

[Fungus]

Another example would be the sales of Apple products. If we look at iPhone sales, they are highly cyclical in nature, with sales peaking right after release of the new model.

Did Apple ever release a new model at the exact same price point as the previous one? Usually they put the price up.

The oscilloscope market is much slower moving than the Apple or automobile market though.

[Mr. Scram]

Rather than starting yet another thread, my question might be more suited here. I am currently looking for an entry level scope. The work I want to do with it is very mixed. Analog, digital, microcontrollers,  switching and linear power supplies on PCBs, everything to do with remote controlled vehicles, CPLDs and FPGAs and a small amount of RF stuff. In most cases, I think more channels would serve me better, like when decoding digital signals, working on brushless motor drivers or viewing two sets of input and output signals. In a few cases, like FPGAs, the extra bandwidth might serve me well. Unlocking bandwidth or features to get the most bang for buck is not an issue.

In my specific case, are there any reasons to talk me into or out of either model that have not been discussed at length already?

Did Apple ever release a new model at the exact same price point as the previous one? Usually they put the price up.

The posted graph shows units shipped.

[nctnico]

@MR Scram: if you have not already done this: also take a look at the GW Instek GDS-2000E (4 analog) or MSO2000 (4 analog + 16 digital) series. These are more expensive compared to the scopes in the topic title but they have more features besides displaying a signal and the firmware is mature.

[Mr. Scram]

@MR Scram: if you have not already done this: also take a look at the GW Instek GDS-2000E (4 analog) or MSO2000 (4 analog + 16 digital) series. These are more expensive compared to the scopes in the topic title but they have more features besides displaying a signal and the firmware is mature.

Thanks for the advice! The cheapest 4 channel GDS-2000E seems to be three times as expensive as the DS1054Z. That might be a bit too rich for my blood right now, even if it can be unlocked to 200 MHz. The MSO2000 does not seem to be for sale around these parts, which would make warranty a huge hassle.

If it means a significant upgrade, I am willing to consider increasing the budget, or if it turns out some of the things I would want to do with a scope are simply out of reach on the models mentioned in this thread. I don't mind spending money on a good tool, but at the same time I have to be realistic about my budget. Spending more here means less options elsewhere.

[Mr. Scram]

If you want better responsiveness, better vertical resolution and better FFT the Siglent SDS1202X-E is what you're looking for. But limited to two channels.

Regarding FFT, I just found the project linked below. It provides better FFT on the DS1054Z, allowing your computer to do the heavy number crunching and yielding much better results. If you use FFT a lot it might be a bit of a hassle, but for occasional use, it is a great option to have.

http://hackaday.com/2015/09/22/a-better-spectrum-analyzer-for-your-rigol-scope/

[nctnico]

@MR Scram: if you have not already done this: also take a look at the GW Instek GDS-2000E (4 analog) or MSO2000 (4 analog + 16 digital) series. These are more expensive compared to the scopes in the topic title but they have more features besides displaying a signal and the firmware is mature.

Thanks for the advice! The cheapest 4 channel GDS-2000E seems to be three times as expensive as the DS1054Z. That might be a bit too rich for my blood right now, even if it can be unlocked to 200 MHz. The MSO2000 does not seem to be for sale around these parts, which would make warranty a huge hassle.

If it means a significant upgrade, I am willing to consider increasing the budget, or if it turns out some of the things I would want to do with a scope are simply out of reach on the models mentioned in this thread. I don't mind spending money on a good tool, but at the same time I have to be realistic about my budget. Spending more here means less options elsewhere.

It also depends on whether you need this scope to make money. If you can't afford to wait for bug fixes and/or the limitations make work go slower then spending extra money is the only option.

[MrW0lf]

Note that with DS1054Z you cannot work live with substantially different frequencies at same time. Memory depth allows for high sample rate capture but auto-measurements do not work on hf part while zoomed out. Might be important with motor work (low freq physical process + high freq signaling). This limitation did come as surprise and annoyed me so much that started sort of "holy war" back in the day
https://www.eevblog.com/forum/testgear/testing-dso-auto-measurements-accuracy-across-timebases/
This is common with scopes that use screen buffer for measurements to speed things up, but comes at cost.

[Mr. Scram]

Note that with DS1054Z you cannot work live with substantially different frequencies at same time. Memory depth allows for high sample rate capture but auto-measurements do not work on hf part while zoomed out. Might be important with motor work (low freq physical process + high freq signaling). This limitation did come as surprise and annoyed me so much that started sort of "holy war" back in the day
https://www.eevblog.com/forum/testgear/testing-dso-auto-measurements-accuracy-across-timebases/
This is common with scopes that use screen buffer for measurements to speed things up, but this comes at cost.

If I understand you correctly, the DS1054Z uses a software optimisation by using the data on the screen, rather than the full recorded dataset. Because of this, detecting multiple signals that vary wildly in frequency cannot be accurately automatically detected.

Would that be an accurate summary?

[MrW0lf]

Would that be an accurate summary?

Yep. Now finding cheap scope that does all well currently impossible if you need it all:
- good analog part (4ch) with full-memory automation
- good FFT
- digital channels
- serial triggering
- serial decoding
I would be looking into combining devices if proper do-it-all stuff out of range.

[2N3055]

Note that with DS1054Z you cannot work live with substantially different frequencies at same time. Memory depth allows for high sample rate capture but auto-measurements do not work on hf part while zoomed out. Might be important with motor work (low freq physical process + high freq signaling). This limitation did come as surprise and annoyed me so much that started sort of "holy war" back in the day
https://www.eevblog.com/forum/testgear/testing-dso-auto-measurements-accuracy-across-timebases/
This is common with scopes that use screen buffer for measurements to speed things up, but this comes at cost.

At a risk to start it again (hope not) I do have to again reiterate: DS100Z uses decimated data (from screen) for measurements. So if you are watching a signal that is zoomed in (set timebase ) to a part of signal you want to measure it will measure correctly.  If you set timebase so you can't legibly see waveform, it doesn't measure correctly. That is completely true.

But most of the time that is not important.  That type of measurement is only applicable to repetitive, monotonic, waveform, and is actually statistics of measurements across the capture (single trigger event). That is useful to analyze quality of clock , for instance..  But, in order for that to be useful, scope has to have high end clock stability etc..

For instance when you measure "Rise time" on Rigol DS1000Z, it measures first detected positive slope and measures that.. It wont measure any other, even if you have 10 periods on screen. It works same as auto cursor measurements. And statistics shown are between different trigger events, not between different waveform periods in same trigger event.

So basically, before measurement, you need to "aim" to what part of waveform you want to measure..
That is not very sophisticated, but works.. And is used most of the time in my work. And most other people. And is only way to measure something if you are not looking into continuous clock or sine wave..

If I have 10ms of empty capture, and in the middle of it there is 1us burst of something, I will zoom in to that. To see what is going on there.. To see how it looks, to see the shape of the edges.
And surprisingly, if you "zoom in" to it (on top, long timebase, on bottom zoomed in) it will measure with zoomed pixels.. With good resolution and accuracy. And since it does that even on stopped signal, it means it keeps whole sampled memory, and decimates only for screen. They might even be able to make measurements work on a whole 24Mpoints (or a smaller subset, but still more that now, like some other scopes.) but they don't think it's important..  And most of the time it's not.

Statistical signal analyzer it is not. But as a simple scope works just fine.

DS1000Z is simple cheap machine that serves well if you respect it's specifications. It's not perfect, not even very good, but more than good enough for most part. It is certainly worth it's money.

There are many thing on DS1000Z that are not perfect.
 
I'm looking to get better scope, partly because it doesn't have decode on segmented memory (which is joke in a first place) and tools to search segmented memory.

For instance, GW Instek GDS-2000E Series seems to have both decode from segmented memory and capability to search through segmented memory with pretty much everything you can trigger on... If that works well, that is probably cheapest good scope that can be used professionally for simpler tasks not requiring mid or high end scope...

But any 4ch scope that is SUBSTANTIALLY better (meaning: work giving money for) is more than 4x expensive.. Or was so far.. If that changes, people will start recommending something else.

[MrW0lf]

But most of the time that is not important.  That type of measurement is only applicable to repetitive, monotonic, waveform, and is actually statistics of measurements across the capture (single trigger event). That is useful to analyze quality of clock , for instance..  But, in order for that to be useful, scope has to have high end clock stability etc..

Motors. Slow-changing statistically smoothed measurements on physical part, while keeping eye on control part. But ok, suppose this is niche application, but second implication is cumbersome serial decoding. Some think that manually browsing thru megabytes (because it only decodes zoomed part) is normal practice. Actually normal practice is having navigable event table over full memory decode.

I'm quite sure one can get all the modern features under $1000 if combine devices.

[bd139]

I spent a week trading off the SDS1202X-E and 1054Z and ended up buying the 1054Z because of the following things:

1. 4 channels
2. Ecosystem around it. For example PyDSA, forum posts galore, well established limitations and workarounds.
3. Direct copy of captured waveform to DS1022Z synthesizer
4. Slightly cheaper
5. UK reseller is decent with plenty of stock and covered by SOGA in UK
6. If I want an FFT I want one that will hit 1GHz and understands RF power so I'm going to buy (or build) a spectrum analyser anyway.

MOST of what I do is either audio, RF, power, digital and the RF side of things is better served with an SA. The power, audio, digital are well served by the DS1054Z.

Bear in mind I've done fine with a 50MHz analogue crate, a pattern trigger and a logic probe for 20 years

[Mr. Scram]

Yep. Now finding cheap scope that does all well currently impossible if you need it all:
- good analog part (4ch) with full-memory automation
- good FFT
- digital channels
- serial triggering
- serial decoding
I would be looking into combining devices if proper do-it-all stuff out of range.

That surprises me. I remember Dave showing us in the features review that the DS1054Z has a hardware frequency counter in addition to the on-screen software frequency counter. I would think that mitigates this issue.

[2N3055]

But most of the time that is not important.  That type of measurement is only applicable to repetitive, monotonic, waveform, and is actually statistics of measurements across the capture (single trigger event). That is useful to analyze quality of clock , for instance..  But, in order for that to be useful, scope has to have high end clock stability etc..

Motors. Slow-changing statistically smoothed measurements on physical part, while keeping eye on control part. But ok, suppose this is niche application, but second implication is cumbersome serial decoding. Some think that manually browsing thru megabytes (because it only decodes zoomed part) is normal practice. Actually normal practice is having navigable event table over full memory decode.

I'm quite sure one can get all the modern features under $1000 if combine devices.

Agree.. manually browsing thru megabytes is not a way to go.. That's why I also have logic analyzer separately.. But it's not the same.
For me, normal practice is search and navigable event table over multiple segmented captures..
I have many products that are idle most of the time and emit data bursts occasionally. I need segmented memory, and a usable one..

That just show how all of us have different needs. And no single instrument will be able to make everybody happy. For your work, I would recommend you Picoscope.. But you already know that and have one AFAIK...

Regards,

Sinisa

[MrW0lf]

DS1054Z has a hardware frequency counter in addition to the on-screen software frequency counter. I would think that mitigates this issue.

Maybe be of interest, counter performance mapped:
https://www.eevblog.com/forum/testgear/rigol-ds1000z-series-(ds1054z-ds1074z-ds1104z-and-s-models)-bugswish-list/msg1080859/#msg1080859

Edit: not only freq, but all horizontal stuff is basically non-functional when not at exactly right timebase, or zoom level. PWM duty for example.

[borjam]

What I have found so far, using both of them, is that the UART decoder in the Siglent is a bit sensitive to the timing settings.

I have been trying with a RS485 bus, capturing Modbus packets at 9600 bps. And packets for which all the bytes decoded successfully at 10 ms/div had some errors when capturing at 50 or 100 ms/div.

Unfortunately I've done the test at work and my main goal wasn't to compare oscilloscopes but to check something else, so I can't do a more comprehensive testing nor begin capturing examples.

I wil try to reproduce it at home if possible. But I found that, for an initial packet beginning with 01h-04h the first byte wasn't decoded despite looking perfectly fine on screen.

[tautech]

What I have found so far, using both of them, is that the UART decoder in the Siglent is a bit sensitive to the timing settings.

I have been trying with a RS485 bus, capturing Modbus packets at 9600 bps. And packets for which all the bytes decoded successfully at 10 ms/div had some errors when capturing at 50 or 100 ms/div.

Unfortunately I've done the test at work and my main goal wasn't to compare oscilloscopes but to check something else, so I can't do a more comprehensive testing nor begin capturing examples.

I wil try to reproduce it at home if possible. But I found that, for an initial packet beginning with 01h-04h the first byte wasn't decoded despite looking perfectly fine on screen.

When you get to test this in greater depth please include whether or not you're using the appropriate serial triggering suite or the normal edge trigger.
I demo'ed a X-E unit yesterday using the normal edge triggering and despite having a rising edge trigger it decoded without error.

Anyway the more info you can provide the better it can be replicated if indeed there is an issue.
TIA.

[borjam]

What I have found so far, using both of them, is that the UART decoder in the Siglent is a bit sensitive to the timing settings.

I have been trying with a RS485 bus, capturing Modbus packets at 9600 bps. And packets for which all the bytes decoded successfully at 10 ms/div had some errors when capturing at 50 or 100 ms/div.

Unfortunately I've done the test at work and my main goal wasn't to compare oscilloscopes but to check something else, so I can't do a more comprehensive testing nor begin capturing examples.

I wil try to reproduce it at home if possible. But I found that, for an initial packet beginning with 01h-04h the first byte wasn't decoded despite looking perfectly fine on screen.

When you get to test this in greater depth please include whether or not you're using the appropriate serial triggering suite or the normal edge trigger.
I demo'ed a X-E unit yesterday using the normal edge triggering and despite having a rising edge trigger it decoded without error.

Anyway the more info you can provide the better it can be replicated if indeed there is an issue.

I used serial trigger, on the first start bit.

[borjam]

I had to try that again, this time I captured an example of the SDS1202X-E decoding oddity. The oscilloscope is running the latest firmware version.


I set up the oscilloscope to 100 ms/div (It says it will fill 14 Ms at 10 MS/s which is more than enough to decode a 9600 bps signal, I guess), 1 V/div. Using a x10 differential probe (Hameg HZ109). It's triggering in serial mode, start bit.

There is some noise which could be due to probe malfunction or to the fact that there are around 40 metres of twisted pair cable going to a power consumption monitor.

I have complete files available (the oscilloscope setup and a binary capture). Both screenshots are the same Modbus packet with different levels of detail. Send me a private message with an email address if you want the complete capture (14 MB in binary format) and the setup.xml file.

[tautech]

@borjam
Got your PM.

What concerns me is even while you're DC coupled the idle level is undetermined, it's midway neither Hi or Lo. 
How can that be ?



BTW to capture more OSD info including the menu just use the blue Print button rather than the Save menu to capture a screenshot.

[2N3055]

@borjam
Got your PM.

What concerns me is even while you're DC coupled the idle level is undetermined, it's midway neither Hi or Lo. 
How can that be ?



BTW to capture more OSD info including the menu just use the blue Print button rather than the Save menu to capture a screenshot.

It's not undetermined, it is RS485, measured with differential probe... When idle drivers go mid supply... differentially that's  zero volts..

[borjam]

@borjam
Got your PM.

What concerns me is even while you're DC coupled the idle level is undetermined, it's midway neither Hi or Lo. 
How can that be ?

It's a two wire RS485 bus, which means it's half duplex. Devices are in high impedance mode when receving. In order to transmit, a device changes the port mode to transmission (which makes it reach the set the right idle mode), it transmits and after transmission it goes back to high impedance. That's why you see a 0 volt idle level.

Some devices can be a bit different. This one for example is a ModbusTCP-RS485 gateway and it doesn't assert the serial line before transmitting but it just begins with the first start bit. In another example I tried, using a different RS485 device, the line is "activated" in idle mode (so, the voltage goes high) and it waits for several ms before sending the first stop bit.

I am attaching a sample of that other device, this time captured on a Rigol DS1000Z. I tried this same device with the Siglent and it failed to decode the first byte successfully depending on the time base I configured.

In this case you see the idle value is around 2 V (positive).