Which oscilloscope is better?

[AV9]

Hello!

I´m a studient from Spain and I want to buy an oscilloscope. I look these three models but I don´t know which is better:

1º - RIGOL DS1052E. This can be hacked as you know to get 100 MHz bandwith.
link:http://www.rigolna.com/products/digital-oscilloscopes/ds1000e/

2º - SIGLENT SDS1102CM. This model has been recommended to me from a HAMEG´s worker. (It seems to the Atten model).
link:http://www.siglent.com/en/product_show.asp?articleid=159

3º - ATTEN ADS1102CAL. I like this one only for the screen. It seems so big.
link:http://www.dealexcel.com/atten-ads1102cal-100mhz-digital-oscilloscope-1102cal_p3714.html

I just want to have another opinion of wich one has better quality and wich has better technicals specifications. I prefer both first...what do you think about??

Thanks, all your opinion will be very important for choose one!!!  

[tyblu]

Looks to me like the Siglent SDS1102CM is the best one: 100MHz, 2M memory depth. Try it out before you buy, though, to test the interface.

[AV9]

That´s what I thought, but the memory depth is the last diference,isn´t? because when you hack the Rigol the other parameters like rise time or sample rate, will be the same than SIGLENT?.

Thanks!

[tyblu]

That's right. Plus, you have to actually hack the Rigol. Memory depth is more important to me than bandwidth, to a point.

[saturation]

I rarely find any reviews for Atten, and almost nothing for Siglent.  If these scopes live up to their specs, they would be good machines, but until they do, many Chinese test gear tend to overstate their spec sheets.

Rigol has come to the forefront because it has lived up to its spec sheet and reported good build quality, because many folks have reviewed it, here and elsewhere.

Rigol too isn't immune from overstating or making errors in its specs.  Rigol revise its users manual quietly adding an important omitted item: using the maximum record length will halve the sampling rate and more if 2 channels are used, will halve it again! this was a heavy critique made by user rf-loop on eevblog and his other name, InstrumentVu on youtube.com.  Also, there is some controversy on how accurate the sinx/x function was implement by Rigol.  I also find the 1052E automated measurements are off, compared to manually measuring values, I find it disagrees with my measurement by 5-10%.

Recently, a user called Tinhead has been giving good marks to Tekway/Hantek, and also we've heard good reports from Instek, but so far nothing Atten and Siglent.

Some Atten reviews, very cursory, like 'it works ok', 'fine for me', nothing remotely what Dave or others here have done detailing its electrical capacities.

http://www.sonsivri.com/forum/index.php?topic=22076.msg76518#msg76518

http://www.electro-tech-online.com/repairing-electronics/105860-help-needed-my-digital-oscilloscope-gets-freeze.html

Alas, what little of Atten reviews are on youtube.com are also very poorly done.  All those on Rigol are more revealing.

So, in the end Rigol has had a lot of exposure, that makes me give them higher marks than other brands in this speed and price range.

[AV9]

Thanks for all your answers. So I have to decide between a sure election like the Rigol, that we know that it works so good, or Siglent, that has more depth memory but maybe they has less experience manufacturing and anyone knows him...Anyway I don´t how much important is this depth memory diference in the practice. The question is, Is it enough to buy the Siglent instead Rigol?

I thought in the Siglent DSO because a worker from HAMEG called me to sell it to me. At the begining he wanted to sell me a HAMEG DSO but it was too expensive for me, and then he told me that they have a chinese model that they recommending me for their quality and technicals specs, so I thought if HAMEG sell this, it must to be good...

It should be a great idea to find someone who has a Siglent...but it seems a little dificult, so I think that finally I will buy a Rigol and I will change his firmware to get the 100 Mhz. 

Anyway if someone has another idea it would be perfect!

Thanks again!

[saturation]

Just an idea, it sounds like you have a local dealer for Siglent, and so this purchase need not be mail order?  If so, that's a good opportunity to try something new because it can be easy to return, but I'd need assurance I can test and return it if I'm not happy.  It may come out to be a very good buy, but someone has to be the first one to give it a review.

For a DSO, bandwidth, sampling rate, and memory depth are all the critical components for its overall performance, a weakness is one hampers the whole performance.  For example, if you use the Rigol in long record mode with 2 channels active, the sampling rate drops to 250Ms/s, this means its useful frequency response is ~ 25 MHz before aliasing become a problem, even if the scope is rated at 50 MHz.  If you hack the Rigol to 100 MHz, it won't matter, it won't change the sampling rate and aliasing will still occur.

Thanks for all your answers. So I have to decide between a sure election like the Rigol, that we know that it works so good, or Siglent, that has more depth memory but maybe they has less experience manufacturing and anyone knows him...Anyway I don´t how much important is this depth memory diference in the practice. The question is, Is it enough to buy the Siglent instead Rigol?

I thought in the Siglent DSO because a worker from HAMEG called me to sell it to me. At the begining he wanted to sell me a HAMEG DSO but it was too expensive for me, and then he told me that they have a chinese model that they recommending me for their quality and technicals specs, so I thought if HAMEG sell this, it must to be good...

It should be a great idea to find someone who has a Siglent...but it seems a little dificult, so I think that finally I will buy a Rigol and I will change his firmware to get the 100 Mhz.  

Anyway if someone has another idea it would be perfect!

Thanks again!

[AV9]

But, Dave Jones said in one of this forums that the diference betwen Rigol DS1052E and Rigol DS1102E 100MHz is just the software, so if this its true, when you increase the 50 Mhz to the double, it does also the sample rate and the rise time convert into the half, isn´t?

Sorry if I understood it wrong, I´m from Spain and my English is not perfect.

Thanks.

[alm]

For a DSO, bandwidth, sampling rate, and memory depth are all the critical components for its overall performance, a weakness is one hampers the whole performance.

Don't forget the analog part, this is not your standard RF amplifier (no matched impedances and much wider bandwidth). The Rigol doesn't appear to be particularly great at this, but I haven't seen a comprehensive test. Analog performance is just not as apparent from a spec sheet as something like sample rate. Other properties that don't stand out from specs are user interface, responsiveness, number of waveforms per second (especially when enabling measurements/math/FFT), contrast/viewing angle of the screen and noise (in the waveform, not fan noise). Tons of ways to ruin the user experience of a scope .

[tyblu]

It apparently has an analog front end that drops off around 160MHz.. sufficient for antialiasing samples rates of 320MSps. Hopefully the DSP stage doesn't 'throw away' too many samples, and actually does interpolating decimation. (Is this the question about the sinc(x) stage?)

[saturation]

More later, but here's a clear demo of what the sample rate does to standard waveforms.

Source, Hantek 3x25.
1052E, 50 MHz, unmodded, with Rigol RP 2200 probes.

The first 3 show a 25 MHz sine wave, sampled at 250 Ms/s, 500 Gs/s, and 1Gs/s.  The yellow dots are the actual waveform samples, and the white tracing interpolates the samples.

Note the shape of the sine wave, particulary the jagged edges and how they are smoothened.

The next three images are 50 MHz sine waves, at the bandwidth limit of the Rigol 1052E.  Sinc off.

White tracing is the interpolated 250 Ms/s sine.  The yellow tracing is 500 Ms/s, so you can easily detect the differences.

The next tracing, White tracing is 500 Ms/s compared to yellow, 1Gs/s.

The next tracing, is both 500Ms/s and 1Gs/s with sin(x)/x turned on.

The last 3 images are 700 kHz square waves sampled at 250Ms, 500Ms and 1Gs, all per sec.  Note the rise time and compare to the next post against 7MHz.

[saturation]

Last set of scope grabs.

The yellow curve shows the rise time of a 7MHz signal from the Hantek 3x25.  You can see that as the sampling rate drops, the rise time is prolonged.  This 'artifact' is independent of frequency, see previous message for samples at only 700kHz, since a signals' rise time is typically independent of frequency.

The Rigol was set identically. The differences were:

For 250 Ms/s, both channels were activated, but the second channel in blue has no signal, set for long record length.

For 500 Ms/s, single channel activated, long record length.

For 1Gs/s, single channel activated, normal record length.

[Mechatrommer]

But, Dave Jones said in one of this forums that the diference betwen Rigol DS1052E and Rigol DS1102E 100MHz is just the software, so if this its true, when you increase the 50 Mhz to the double, it does also the sample rate and the rise time convert into the half, isn´t?

by software mod, rigol will unleash some of 1102E capability like 2ns/div (in software) and disabling some frontend circuitry (analog) IIRC. but there is no prove it will double sample rate to 2Gs/s or halving it to 500Ms/s (digital side), the rigol stays at 1Gs/s max as speced.

[saturation]

Yes, the front end analog amp/filter is less a limiting factor, but as my simple test shows the rate limiting performance is the sampling rate and memory length.

What I've seen as a limitation of the front end amp is mostly amplitude roll off with frequency, so far, no distortion.

If Rigol were to exploit its -3dB 160 MHz analog bandwidth, it would have to raise the sampling rate to at >= 1.6Gs/s; which they do in the next model line above the 1000E series.

Nyquist-Shannon 2x limit applies only to the fundamental, not the harmonics, so at least 160 MHz square waves into a 320Ms/s will look line a sine wave, with the harmonics filtered away by aliasing.

My images demonstrate that, 25 MHz sine sampled at 250Ms/s is already showing distortion, note the tops of the sine waveform:




A 50MHz sine wave is sampled at 250Ms/s, tracing in white, with clear distortion, at 500Ms/s, tracing in yellow, its reduced simiar to the 25 MHz sine at 250 Ms/s but its even better at 1Gs/s, last image.



White = 500Ms/s, Yellow 1Gs/s, sampling 50 MHz sine wave, sinc off

It apparently has an analog front end that drops off around 160MHz.. sufficient for antialiasing samples rates of 320MSps. Hopefully the DSP stage doesn't 'throw away' too many samples, and actually does interpolating decimation. (Is this the question about the sinc(x) stage?)

[saturation]

Couldn't agree more alm, but these traits are the eidos of the DSO, for lack of a word.  I was referring to the analog bandwidth in my short summary.

My experience with other DSOs are limited, but Rigol's analog front end hasn't played a substantial part in affecting the scopes capacity to show waveforms properly, beyond reducing amplitude.  Even with the unmodded 1052E, which has its 160 MHz front analog end crippled.

I think Rigol engineered 50 MHz as alimit as it really is a best tradeoff of the scopes capability.  Removing the analog limit won't reduce the limitations imposed by Nyquist without improving sampling rate and/or memory too.

The Agilent version of the Rigol, DSO-3062, does not have the front end filter, and will go to 160 MHz, but its was sold as a 60 MHz scope by Agilent because it had the sampling rate system of Rigol, and it only has 8kpts memory, no long memory length mode.

Thus, its seems more important for Rigol users to know how to control the factors that change the sampling rate and memory depth for best effect, rather than hack the analog amp off its 50 MHz limitation.

Its good that eevblog has a cadre of folks who continue to dissect its capabilities far more than anywhere on the net, and it shows you do get more, and help you wisely choose, when you pay more for a class of scope from reputable folks like Tek, LeCroy and Agilent, including mundane items too like their passive probes.

For a DSO, bandwidth, sampling rate, and memory depth are all the critical components for its overall performance, a weakness is one hampers the whole performance.

Don't forget the analog part, this is not your standard RF amplifier (no matched impedances and much wider bandwidth). The Rigol doesn't appear to be particularly great at this, but I haven't seen a comprehensive test. Analog performance is just not as apparent from a spec sheet as something like sample rate. Other properties that don't stand out from specs are user interface, responsiveness, number of waveforms per second (especially when enabling measurements/math/FFT), contrast/viewing angle of the screen and noise (in the waveform, not fan noise). Tons of ways to ruin the user experience of a scope .

[alm]

My experience with other DSOs are limited, but Rigol's analog front end hasn't played a substantial part in affecting the scopes capacity to show waveforms properly, beyond reducing amplitude.  Even with the unmodded 1052E, which has its 160 MHz front analog end crippled.

Remember that it's a frequency-dependent reduction of amplitude, so for anything but sine waves, this will result in distortion. There may also be a phase shift at higher frequencies, which will result in even more distortion.

The 160 MHz number you guys keep quoting was from a single scope on a single vertical attenuation setting. I've seen some other results that appeared to indicate that the front-end attenuator was way too low above 70 MHz or so, but that person didn't have the proper test equipment to verify, and I don't have a Rigol scope. Someone much more knowledgeable than me claimed that it was unrealistic to expect cheap scopes with standard SMT components on FR-4 to meet their frequency specs over the vertical attenuation range, and that it was perfectly normal that the gain was way too low at 70 MHz. But I am careful with drawing conclusions from these results, since they were from just one person, and somewhat incomplete. Unlike some people that base their claims on a single bandwidth measurement .

[saturation]

Yes, assuming its at its best to stay far before the -3dB mark to avoid it frequency response related distortion.  If this scope were really 160 MHz, making it limited to 50 MHz would give a lot of 'head room' for harmonics.  Not to mention your skepticism about the true frequency limits of the front end amp.

It again points that the Rigol 1052E is at its best in the 50 MHz range but to realize it well and stay at least 3x lower than the alleged un-crippled amp frequency response, its best the 50MHz filter be removed, like the Agilent version, 3062, which is rated by Agilent as 60 MHz but external testers have reported its -3dB point is at ~ 160 MHz [ recently reported at hp_agilent_equipment@yahoogroups.com].  As for phase shifts etc., alas they have specified any, but it would be good to check if it did occur.

I too was going to post a query about this 150/160 MHz limit but I found at least 3 separate eevblog posters who made measures at different times, it doesnt' seem like its the same device, so there maybe some truth to it:

gandalf8
https://www.eevblog.com/forum/index.php?topic=553.msg29449#msg29449

dimlow
https://www.eevblog.com/forum/index.php?topic=1495.msg19819#msg19819

Rossmoffett:
https://www.eevblog.com/forum/index.php?topic=30.msg2059#msg2059

My experience with other DSOs are limited, but Rigol's analog front end hasn't played a substantial part in affecting the scopes capacity to show waveforms properly, beyond reducing amplitude.  Even with the unmodded 1052E, which has its 160 MHz front analog end crippled.

Remember that it's a frequency-dependent reduction of amplitude, so for anything but sine waves, this will result in distortion. There may also be a phase shift at higher frequencies, which will result in even more distortion.

The 160 MHz number you guys keep quoting was from a single scope on a single vertical attenuation setting. I've seen some other results that appeared to indicate that the front-end attenuator was way too low above 70 MHz or so, but that person didn't have the proper test equipment to verify, and I don't have a Rigol scope. Someone much more knowledgeable than me claimed that it was unrealistic to expect cheap scopes with standard SMT components on FR-4 to meet their frequency specs over the vertical attenuation range, and that it was perfectly normal that the gain was way too low at 70 MHz. But I am careful with drawing conclusions from these results, since they were from just one person, and somewhat incomplete. Unlike some people that base their claims on a single bandwidth measurement .

[alm]

Yes, assuming its at its best to stay far before the -3dB mark to avoid it frequency response related distortion.  If this scope were really 160 MHz, making it limited to 50 MHz would give a lot of 'head room' for harmonics.  Not to mention your skepticism about the true frequency limits of the front end amp.

And aliasing, too much bandwidth is a mixed blessing, which is why the brick-wall response is getting popular.

I too was going to post a query about this 150/160 MHz limit but I found at least 3 separate eevblog posters who made measures at different times, it doesnt' seem like its the same device, so there maybe some truth to it:

I don't deny it, I just don't accept it as fact.

gandalf8
https://www.eevblog.com/forum/index.php?topic=553.msg29449#msg29449

This one seems valid, but only at one attenuator setting.

dimlow
https://www.eevblog.com/forum/index.php?topic=1495.msg19819#msg19819

It appears that he concluded that his edge rate was way too slow to actually verify 150 MHz bandwidth. You also need to show that it's a Gaussian response before you can calculate bandwidth from rise time.

Rossmoffett:
https://www.eevblog.com/forum/index.php?topic=30.msg2059#msg2059

This one is relative to the unmodified frequency response, and refers to a frequency response curve made by a different person, on a different scope, with different equipment/methods. Also, I wouldn't trust a BNC-to-alligator cable for these measurements, I would go 50 ohms all the way to the scope input (feed-through terminator if the scope doesn't have a 50 ohm setting). I believe Tektronix sold a special coax cable with precisely controlled impedance with their low-250 MHz leveled sine wave generator, and the 250 MHz to 1 GHz generator even had a special leveling head directly connected to the scope input.

IMO not enough evidence to state that the front-end has a bandwidth of 160MHz, since that statement suggests that it applies to all samples of their scopes at all vertical setting.

[saturation]

I didn't check in detail rossmoffett sources, so lets remove his input then for now; in that long Rigol 1052E thread, there more folks who also tested 1052Es >= 100 MHz, and the -3B roll off, suggests it likely Gaussian. 

Jahonen [ -3dB @ 170 MHz] and rf-loop, doing a frequency sweep too:

https://www.eevblog.com/forum/index.php?topic=30.msg2838#msg2838
https://www.eevblog.com/forum/index.php?topic=30.msg2160#msg2160

and way back in 1/10, the newsgroup notorious Phil Allison, FWIW:
https://groups.google.com/group/aus.electronics/browse_frm/thread/8db8df6c7f9f710/6edf366688b538db?lnk=gst&q=rigol+phil&hl=en&pli=1#6edf366688b538db

There is also an ersatz 1152E Rigol, sold only in China, not listed on any Rigol product pages, that aside from frequency response, has identical specs to the 2 scopes in the 1000E series, suggesting ~150 MHz as the the upper limit of the 1052E's basic design.

http://www.91sbq.cn/RIGOL/DS1152E-EDU.htm


In :

http://cp.literature.agilent.com/litweb/pdf/5989-5733EN.pdf

Agilent suggests most <= 1 GHz scopes are Gaussian.

Like Agilent, your note and rf-loop suggests its better to evaluate frequency response by actually doing a frequency sweep to the -3dB mark, and not take the rise time estimate as ~ frequency response.  Since ADC can introduce non-linear gains vs frequency, its prudent to test this explicitly.  This is how I tested the probes in another thread, and the scope originally. 

However, I'm not sure doing a frequency sweep across different attenuator setting is necessary since the attenuator is calibrated in v/div and its accurate to within its spec, suggesting one setting is approximate good for all settings.


In the end, its all just food for thought.

This one seems valid, but only at one attenuator setting.

It appears that he concluded that his edge rate was way too slow to actually verify 150 MHz bandwidth. You also need to show that it's a Gaussian response before you can calculate bandwidth from rise time.

Rossmoffett:
https://www.eevblog.com/forum/index.php?topic=30.msg2059#msg2059

This one is relative to the unmodified frequency response,
IMO not enough evidence to state that the front-end has a bandwidth of 160MHz, since that statement suggests that it applies to all samples of their scopes at all vertical setting.

[rf-loop]

Btw afaik Atten = Siglent (I do not know why they do this, maybe some Asian marketing reasons or maybe this old case between Atten - Rigol copy dispute what Rigol win in Chinese cord room)

ADS1102CAL  with small capture memory 7" 480 h.pixel
ADS1102CML  with 2M (or 2x1M)  and 7" 480 h.pixel 

[saturation]

I like the bigger screens, when the menus pop up in the Rigol, the screen is too crowded.  Now, the question is how good are they as electronics instruments?  Have you tried any of them, rf-loop?  I know you were the first to show the sampling problem and sinx/x problem in the Rigol 1000E series.

The only somewhat reputable review I've read on anything Atten is by tinhead on his Tekway thread.

Btw afaik Atten = Siglent (I do not know why they do this, maybe some Asian marketing reasons or maybe this old case between Atten - Rigol copy dispute what Rigol win in Chinese cord room)

ADS1102CAL  with small capture memory 7" 480 h.pixel
ADS1102CML  with 2M (or 2x1M)  and 7" 480 h.pixel 

[cseb]

Hi,

This is my view of the SDS1102CM after 2 weeks:

http://cseb.hu/siglent

Regards,
Csaba     

[saturation]

A most excellent review.

Anyone considering a DSO should read your report.  My take on the data you present is the Siglent has enough misleading outputs that I would not bother with Siglent when there is a Rigol 1102e or better.  You get what you pay for but with Siglent its seems like you get far less than the money you save, IMHO.  A measuring device has to be impeccable and its limits clearly defined, when you start measuring unknowns you have to have full faith in the reading to act on it.

Hi,

This is my view of the SDS1102CM after 2 weeks:

http://cseb.hu/siglent

Regards,
Csaba     

[wkb]

Hi,

This is my view of the SDS1102CM after 2 weeks:

http://cseb.hu/siglent

Regards,
Csaba     

Hm, IMHO this is not a very flattering outcome for the SDS1102CM.