Author Topic: Siglent SDS1104X-E In-Depth Review  (Read 120611 times)

0 Members and 3 Guests are viewing this topic.

Offline Performa01Topic starter

  • Super Contributor
  • ***
  • Posts: 1686
  • Country: at
Re: Siglent SDS1104X-E In-Depth Review
« Reply #225 on: January 12, 2024, 07:54:08 pm »
I can see ringing of the scope vertical amplifier there. It gets worse with more offset.
No, what you actually see is the response from a severly overloaded split path input buffer. We cannot rely on the results of any measurement when a input is overloaded, particularly in such an excessive way.

This has been descussed numerous times, e.g. here:

https://www.eevblog.com/forum/testgear/siglent-sds1104x-e-anomaly-on-what-should-be-a-simple-task/msg4603873/#msg4603873

I double-checked with another scope (RTM3004) and it is quite flat as expected:
Well, this is not expected, but is rather baffling how R&S managed to get such an excellent overload behavior. Alternatively, we could wonder why the SDS1000X-E is so sensitive to overloading even at very low frequencies - you didn't tell about the rise time of the square wave, so I cannot estimate the bandwidth of your signal.

 

Offline bdunham7

  • Super Contributor
  • ***
  • Posts: 7972
  • Country: us
Re: Siglent SDS1104X-E In-Depth Review
« Reply #226 on: January 12, 2024, 09:08:07 pm »
Alternatively, we could wonder why the SDS1000X-E is so sensitive to overloading even at very low frequencies - you didn't tell about the rise time of the square wave, so I cannot estimate the bandwidth of your signal.

It's been a while since I looked at this, but I actually don't think the phenomenon is due to the overloading/saturation/recovery as much as it is to the size of the step.  Of course it is most severe when you actually have an overload--wherever that point is--but you'll get progressively larger disturbances as you increase that step size even not overloaded, in fact even with signals that are totally on the screen with no offset.  IIRC, rise time doesn't matter at all (within reason) as it is a very low speed phenomenon.
A 3.5 digit 4.5 digit 5 digit 5.5 digit 6.5 digit 7.5 digit DMM is good enough for most people.
 

Offline cichmen

  • Newbie
  • Posts: 6
  • Country: cz
Re: Siglent SDS1104X-E In-Depth Review
« Reply #227 on: January 13, 2024, 12:44:06 pm »
even not overloaded, in fact even with signals that are totally on the screen with no offset.

Exactly, I can see there this "ringing" even in this case:
 

Offline Performa01Topic starter

  • Super Contributor
  • ***
  • Posts: 1686
  • Country: at
Re: Siglent SDS1104X-E In-Depth Review
« Reply #228 on: January 14, 2024, 12:34:35 pm »
Input overload, Signal detail & fidelity

As the previous discussion shows, it would be desirable to have some hard facts. This is why I’ve done some tests using my old Siglent SDS1104X-E, Hardware 00-02, Firmware 6.1.37R8.

As always, we need to characterize the test signal first; it is a 1 Vpp 10 Hz square wave with 50 ns rise time (arbitrarily set). The reference measurement has been made with a Picoscope 4262, which has at least two unique advantages over other instruments:

•   True 16 bit performance
•   No split path input buffer, hence none of its inherent problems

As a consequence, we can capture the signal correctly without overloading the scope input and then use zoom to inspect the details:


Pico4262_Square_10Hz_Zoom

As can be seen, the square wave is not perfect indeed. The top is not entirely flat and has an aberration of ~9.7 mV, resulting in a maximum error of 0.97%.
 
In order to avoid overloading the input of an SDS1104X-E, the signal amplitude must not exceed 1 Vpp. At 100 mV/div, the 1 Vpp signal already exceeds the screen height, but this is still no problem. It is a problem though that the SDS1000X-E series is a strict 8 bit DSO. Consequently, it doesn’t provide vertical zoom, because even without any zoom half the screen height would already be enough to show all the captured detail.

There is a little trick though, (with regards to LeCroy who used the Identity math function to provide zoom). Unfortunately, the SDS1104X-E has just the most basic math operations and certainly no Identity function. But we can substitute that by an addition: if we use the Ch.X + Ch.X operation, we get 2 x Ch.X, i.e. twice the amplitude of the otherwise unchanged input signal. Taking this into account, we can just double the intended vertical scale setting for zoom, e.g. use 20 mV/div when we originally intended to use 10 mV/div:


SDS1104X-E_Square_10Hz_Math

Yet that’s certainly not what we have hoped for. The math result is still only 8 bits and at 10 times magnification, noise also becomes noticable. Even HIRES and Average acquisition modes are truncated to just 8 bits, but at least they help to get rid of the noise.

Average is the preferred mode as it doesn’t affect the bandwidth, yet it requires a repetitive signal, as is the case here.

ERES on the other hand can be used on single shot recordings, but it limits the bandwidth – which might even be desirable in certain scenarios.

The same zoom result with Average 64 acquisition:


SDS1104X-E_Avg64_Square_10Hz_Math

Now that the noise is gone, we can measure the aberration: it is about 18 mV. Since we’re measuring 2 x Ch.4, we need to divide by two and get ~9 mV, which conforms to the reference measurement quite nicely.

Yes, the waveform is not quite the same as the reference, because we see a kind of overshoot at the beginning. This is one of the problems of the split path input buffer; it is in the recombination of the LF and HF path, which have significant runtime differences. Such effects will be visible whenever the crossover frequency of the split path buffer falls somewhere in the middle of the signal spectrum, causing pulse distortions because of the non-constant group delay.

All that said we should never forget about the dimensions we’re talking here: an aberration of less than one percent of the signal amplitude – this is much better than e.g. the DC accuracy specification of the instrument. And even expensive scopes can show overshoots up to 10% - not many manufacturers are brave enough to put these figures in their spec sheets though.

We can now try overloading the input step by step and see if we still get correct results, e.g. five times overloading by using 20 mV/div:


SDS1104X-E_Avg64_Square_10Hz_Math_OVD5

Now there should be enough detail already and the measurement is still correct.

We can take it one step further, get rid of the math channel and switch back to normal acquisition mode:


SDS1104X-E_Square_10Hz_OVD10

Conclusions: Even though in this special case the waveform isn’t reproduced 100% accurately, the aberration is less than one percent, hence nothing to complain about, especially not for a cheap bottom entry level scope like the SDS1104X-E.

By contrast, an SDS2000X HD is much closer to the truth (<0.1% aberration). It needs not be overdriven as it has 12 bits from the outset, together with high resolution math and vertical zoom. The SDS2000X Plus is only 8 bits physically, but it has high-res math, thus can provide similar results.

In particular the upcoming SDS800X HD will have all the relevant features of the SDS2000X HD, thus being much less restricted than the SDS1000X-E series and that for a really low budget.


SDS2504X_HD_Square_10Hz_Zoom_Avg64
« Last Edit: January 14, 2024, 12:38:40 pm by Performa01 »
 
The following users thanked this post: 807, 2N3055, Mortymore, Svgeesus, nez, core, sylvandb, cichmen

Offline cichmen

  • Newbie
  • Posts: 6
  • Country: cz
Re: Siglent SDS1104X-E In-Depth Review
« Reply #229 on: January 15, 2024, 10:22:45 am »
Thanks a lot Performa01 for your time and effort to measure and explain this!

For my further study, would be possible to direct me where I can get more info about how the analog frontend is designed, i.e. what is the split path input buffer and why 4262 does not have it?

In order to avoid overloading the input of an SDS1104X-E, the signal amplitude must not exceed 1 Vpp.
Is this somwhere specified?
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 28931
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Siglent SDS1104X-E In-Depth Review
« Reply #230 on: January 15, 2024, 10:36:55 am »
In order to avoid overloading the input of an SDS1104X-E, the signal amplitude must not exceed 1 Vpp.
Is this somewhere specified?
Datasheet P8
https://int.siglent.com/u_file/document/SDS1000X-E_DataSheet_EN04D.pdf

Vertical System>Offset Range
Avid Rabid Hobbyist.
Siglent Youtube channel: https://www.youtube.com/@SiglentVideo/videos
 

Offline Performa01Topic starter

  • Super Contributor
  • ***
  • Posts: 1686
  • Country: at
Re: Siglent SDS1104X-E In-Depth Review
« Reply #231 on: January 15, 2024, 11:58:37 am »
For my further study, would be possible to direct me where I can get more info about how the analog frontend is designed, i.e. what is the split path input buffer and why 4262 does not have it?
David Hess has attached a very informative article about this here (reply #115):

https://www.eevblog.com/forum/testgear/how-much-noise-floor-and-other-things-matter-in-oscilloscope-usability/msg3899150/#msg3899150

In general, there is a discussion about this centered around his posting. Typical unwanted features of the split path design like very pronounced 1/f noise and awful overload recovery should have been touched there, as far as I remember.

The 4262 has only 5 MHz bandwidth. The main reason for the split path design is the increasing difficulty to make very high frequency wideband amplifiers with high input impedance and low offset drift. For only 5 MHz, classical OpAmps with reasonably good offset drift can be used.

In order to avoid overloading the input of an SDS1104X-E, the signal amplitude must not exceed 1 Vpp.
Is this somwhere specified?
No, it's not (offset compensation range is something completely unrelated), but we can test it - for instance, my tests above have proven that a signal that reaches 940 mV below the lower border of the visible screen area is still processed without oveload distortion.

We can do some calculations to estimate it; the maximum regular input voltage can be found by following these steps:
- the highest input range without attenuator is 118 mV/div.
- full screen is 8 divisions, which in turn is equivalent to 200 LSB of the ADC.
- one division is 25 LSB
- full range of an 8 bit ADC is 256 steps; 256/25 = 10.24 divisions.
- 118 mV * 10.24 div. = 1.208 Vpp for full scale.

From this, we can conclude that the internal clamps in the input buffer must be higher than 1.2 Vpp. Furthermore, we can know that the PGA (Programmable Gain Amplifier) following the input buffer will have 0 dB gain at the lowest sensitivity (=118 mV/div) and the average ADC will have at least 2 V input range. Any clamps would be set according to this.

Therefore it's safe to assume a 1 Vpp signal, which can be moved up and down by another volt by means of the offset control, will still not exceed the 2 Vpp unclamped input range.

 
The following users thanked this post: 2N3055, Mortymore, core

Offline cichmen

  • Newbie
  • Posts: 6
  • Country: cz
Re: Siglent SDS1104X-E In-Depth Review
« Reply #232 on: May 23, 2024, 12:43:21 pm »
Again me with question regarding this scope.

I noticed, that when I export (save) samples as binary data, I get minimum 0x03 and maximum 0xFC. I expected min. 0x00 and max. 0xFF for 8bit ADC. I tried to overload the adc, but never got anything lower/higher.

Any thoughts why these 6LSBs are lost somewhere?
« Last Edit: May 23, 2024, 02:10:34 pm by cichmen »
 

Offline kimera

  • Contributor
  • Posts: 11
  • Country: la
Re: Siglent SDS1104X-E In-Depth Review
« Reply #233 on: July 07, 2024, 04:03:38 am »
my oscilloscope  SDS1104X-E
Can I read THD from SDS1104X-E  FFT mode? I have some information about THD calculation from dbc calculation, but SDS1104X-E does not have dbc type. Can I convert dbm to dbc? And how to calculate? Thank you.
 

Offline Performa01Topic starter

  • Super Contributor
  • ***
  • Posts: 1686
  • Country: at
Re: Siglent SDS1104X-E In-Depth Review
« Reply #234 on: July 07, 2024, 05:34:35 am »
Can I convert dbm to dbc?

Basically, dB is a relative measure and the additional letters just signify the reference.

dBc means “dB below Carrier”; as a consequence, it is just the difference between the spectral lines of the fundamental and any harmonic signal.

That means, you can measure in whatever units you find convenient (dBm, dBV) and can always easily calculate the difference between carrier (which is the fundamental in the case of distortion measurements) and any of the distortion products (harmonics).

Here is an old screenshot that demonstrates that you can even make the scope calculate the differences for you:


SDS1104X-E_FFT_Marker_Peak_T

In the screenshot above, you can see a Marker table with Delta Amplitude/Frequency enabled. Unfortunately, this screenshot is from a quite old (2019) firmware where the DC component was still listed in the table – it shouldn’t be there nowadays.

Base line is, you need to ensure that the fundamental frequency (or carrier in other applications) is the first entry in the table. You can always use manual marker control to set marker 1 to the fundamental frequency in order to get correct delta readings for the harmonics, like in the following example with SDS2354X Plus, where the distortion products of a 50 MHz sine wave are measured.


SDS2354X_Plus_Sine_-29dBV_50MHz_2GSa

Caution: the oscilloscope frontend isn’t distortion-free either. While the screenshot above suggests that you can measure at least down to -60 dBc (= 0.1%), this depends on the settings, particularly on the exact vertical gain. Whenever you want to make a distortion measurement, you should first find an optimum setting by measuring a known low-distortion source and finding an appropriate vertical gain (using fine adjust!), where the distortion products of the DSO-fronted are at a minimum. If then your actual measurement is close to this, you know that the test object is probably better than your measurement.

 
The following users thanked this post: Mortymore

Offline kimera

  • Contributor
  • Posts: 11
  • Country: la
Re: Siglent SDS1104X-E In-Depth Review
« Reply #235 on: July 07, 2024, 06:50:02 am »
Can I convert dbm to dbc?

Basically, dB is a relative measure and the additional letters just signify the reference.

dBc means “dB below Carrier”; as a consequence, it is just the difference between the spectral lines of the fundamental and any harmonic signal.

That means, you can measure in whatever units you find convenient (dBm, dBV) and can always easily calculate the difference between carrier (which is the fundamental in the case of distortion measurements) and any of the distortion products (harmonics).

Here is an old screenshot that demonstrates that you can even make the scope calculate the differences for you:


Can you please show me an example of how to calculate it in THD%?

Thank you
« Last Edit: July 07, 2024, 07:01:22 am by kimera »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf