Siglent SDS1104X-E and SDS1204X-E Mixed Signal Oscilloscopes

[mroek]

And lastly. With the coupling set to ground. Do they read better or worse (suspicious procedure may have changed)

If I set the coupling to ground, the offset goes to around +10 mV. Go figure...

Now I need to sleep (it's nearly 3 am here), but I can try other tests tomorrow if need be.

[lordvader88]

My heart is set on 4-ch 200MHz or more scope by the end of the Summer, so I'm budgeting for the SDS1204

Sorry for asking again, but are lot's of people hacking the 100MHz version to the 200MHz without any problems ? There's a big price diff for me.

I want a 4-ch scope, but I don't want to settle for 50-70-100MHz like the nice Rigol's. I want to be higher than that level, as my "base" scope. And then some day I'll save up for a 2-ch thats a lot faster than my base scope.

[Rerouter]

still no issues as of the latest known firmware,

To tautech, I would like to find out what causes GND coupling and DC coupling to have an offset, I'm assuming its hardware (resistor tolerance or something like that), Any chance of some pointers from the factory? e.g. the component references that are suspect.

[mroek]

It appears the offsets are all over the place on my scope. I just did a new warm-up and self calibration, and then recorded all the offsets for all the channels, and here's what I got:

*	Channel 1	Channel 2	Channel 3	Channel 4
5 mV/div	-837.58 μV	-1.79 mV	-5.24 mV	-553.21 μV
10 mV/div	-1.17 mV	-2.83 mV	-6.25 mV	-751.54 μV
20 mV/div	-1.50 mV	-3.20 mV	-6.40 mV	-1.31 mV
50 mV/div	-3.74 mV	-4.00 mV	-4.31 mV	0.0 pV
100 mV/div	0.0 pV	-4.00 mV	-4.19 mV	-4.00 mV
200 mV/div	-8.00 mV	-45.71 nV	0.0 pV	-8.00 mV
500 mV/div	-6.34 mV	-6.83 μV	-114.29 nV	-15.43 mV
1 V/div	-4.21 mV	0.0 pV	0.0 pV	-649.60 μV
2 V/div	-159.94 mV	150.17 μV	-616.46 μV	0.0 pV
5 V/div	-185.12 mV	-47.52 mV	-20.29 μV	-28.29 μV
10 V/div	-392.78 mV	0.0 pV	0.0 pV	0.0 pV
20 V/div	-1.56 V	85.96 mV	25.14 μV	-8.00 μV
50 V/div	-2.86 μV	0.0 pV	-91.43 μV	-148.57 μV
100 V/div	-776.38 mV	0.0 pV	-3.42 mV	-1.17 mV

Probes was connected and shorted, but it doesn't really matter if they're disconnected either. I used averaging in the acquire menu as well, to smooth out the values as much as possible. Despite this, I also observed some unexplainable large jumps in the mean values, but this table is made from a snapshot I made on each setting. I've also attached these snapshots to this post.

If this is indeed to be expected, then I am quite disappointed. I know this is a low end scope, but at least I would expect it to be able to zero out these DC offsets to acceptable levels. My OCD does kick in when I observe the trace being far off from the zero position when there is no signal.

[tautech]

Not heard from factory yet but some thought has me look at the datasheet spec and for old posts.

Offset Accuracy
±(1%* Offset+1.5%*8*div+2 mV): ≥2 mV/div
±(1%* Offset+1.5%*8*div+500 uV): ≤1 mv/div

Further explanation about this is here:
https://www.eevblog.com/forum/testgear/siglent-sds1000x-series-oscilloscopes/msg1138775/#msg1138775

Simply, in 10X input attenuation mode, the displayed offset is 10x worse than it really is, so we see it reduce to very low #'s as the V/div is increased.
Try 1x input attenuation to check if it meets specification and with input to BNC short. (0 ohm)

[mroek]

Changing the attenuation to 1X doesn't really do anything to improve the situation. The range automatically changes from 2V/div to 200 mV/div when I go from 10X to 1X, but if I re-adjust the range back to 2V/div, I observe the same large offset (on channel 1) of around -160 mV.

[tautech]

Changing the attenuation to 1X doesn't really do anything to improve the situation. The range automatically changes from 2V/div to 200 mV/div when I go from 10X to 1X, but if I re-adjust the range back to 2V/div, I observe the same large offset (on channel 1) of around -160 mV.

Which aligns closely to screenshot #12, 1.56V @ 20V/div.

Let's see what the factory says.

[BillB]

My heart is set on 4-ch 200MHz or more scope by the end of the Summer, so I'm budgeting for the SDS1204

Sorry for asking again, but are lot's of people hacking the 100MHz version to the 200MHz without any problems ? There's a big price diff for me.

I want a 4-ch scope, but I don't want to settle for 50-70-100MHz like the nice Rigol's. I want to be higher than that level, as my "base" scope. And then some day I'll save up for a 2-ch thats a lot faster than my base scope.

lordvader88, there aren't any problems after performing the hack.  If calibration certification isn't an issue for you, you can save a lot of money by getting the SDS1104X-E. 

[lordvader88]

Yeah and another member PM'ed me about it too, and the 1104 is 200-300USD less here, so I could get 1 around June if I keep my party money to a minimum.

Excellent. Having a nice modern, small, quiet, fast, 4ch scope will be a big step up from my current scopes. The only thing I don't like about these and similar scopes, is the LCD resolution is low, so some stuff looks jaggie.

[ian.ameline]

My heart is set on 4-ch 200MHz or more scope by the end of the Summer, so I'm budgeting for the SDS1204

Sorry for asking again, but are lot's of people hacking the 100MHz version to the 200MHz without any problems ? There's a big price diff for me.

I want a 4-ch scope, but I don't want to settle for 50-70-100MHz like the nice Rigol's. I want to be higher than that level, as my "base" scope. And then some day I'll save up for a 2-ch thats a lot faster than my base scope.

It is super easy to hack to 200Mhz - no problems at all, other than the probes, which others have shown to be a non-issue.

[lordvader88]

Wait a minute, the SDS1104 only has the single 1GS/s ADC , but the SDS1204 has 2 of them ? Is the memory and the rest of it the same ?

[Rerouter]

Sds1104 and sds1204 both have 2x 1GSPS ADC.

[tautech]

Wait a minute, the SDS1104 only has the single 1GS/s ADC , but the SDS1204 has 2 of them ? Is the memory and the rest of it the same ?

SDS1104X-E and SDS1204X-E both have dual 1 GSa/s ADC's and 14 Mpts of memory serving each.
Four channels used allows for 500MSa/s and 7 Mpts per channel whereas only using 2 channels and with each on an ADC, 1 GSa/s and 14 Mpts are available.

[mroek]

@tautech: You haven't heard back from the factory yet, right?

I'm a bit ashamed to admit, but I'm having trouble reading the specification for DC offset accuracy in the datasheet, which states:

±(1%* Offset+1.5%*8*div+2 mV): ≥2 mV/div
±(1%* Offset+1.5%*8*div+500 uV): ≤1 mv/div

I don't fully understand how to calculate the maximum permissible DC offset, for example for the 2V/div range (which is: ±(1%* Offset+1.5%*8*div+2 mV)) . Could someone enlighten me? When the channel is shorted, the input signal is 0 V, does that mean the offset should not exceed 2 mV?

[radiolistener]

The only exact way i can phrase it is later firmware versions may tweak it.

Thats possible. I'm using old model SDS1102X and when I flashed firmware to the latest version, I got unexpected result.
Usually I'm using oscilloscope for spectrum analysis, by recording large raw sequence and then processing it with FFT on the PC. I'm using FFT with 1-4 million points, so it very sensitive. With latest firmware I got a lot of distortions.
It was below -80 dB, so in the time domain the difference is not significant.

Here is firmaware 2.13R5 (which works good):


And here is firmware 2.15R10:


But in the frequency domain it looks much worse:


I'm not sure why it happens, but these artifacts was added at the time when Siglent released new X-E model. So, I think they both may share the same signal processing algorithm.
After all, I didn't find the way to fix it and since I need clean spectrum, I rollback to the old 2.13R5 firmware. It helps and distortions disappears 

So, the firmware change may lead to significant change in measurement results.
It will be very interest if someone can check if this issue exists on X-E model with latest firmware.

[2N3055]

That is good information. Could you please tell me what software is that ?

Regards,
Siniša

[bugi]

I'm not sure why it happens, but these artifacts was added at the time when Siglent released new X-E model. So, I think they both may share the same signal processing algorithm.

Not only those marked large bumps, but the whole noise floor seems to have smaller bumbs, and lifted in general by about 10dB. Assuming that the measurements between versions were otherwise fully comparable, and that my mind is not tricking me to see those smaller "bumps" in the noise.

[bugi]

I'm a bit ashamed to admit, but I'm having trouble reading the specification for DC offset accuracy in the datasheet, which states:

±(1%* Offset+1.5%*8*div+2 mV): ≥2 mV/div
±(1%* Offset+1.5%*8*div+500 uV): ≤1 mv/div

I don't fully understand how to calculate the maximum permissible DC offset, for example for the 2V/div range (which is: ±(1%* Offset+1.5%*8*div+2 mV)) . Could someone enlighten me? When the channel is shorted, the input signal is 0 V, does that mean the offset should not exceed 2 mV?

Not a pro here, but... replace the "Offset" with current offset (position) value, and the "div" with current volts/div (gain/sensitivity) value. So, when checking calibration with 2V/div and zero offset, the value would be ±(1%*0V + 1.5%*8*2V + 2mV), or about ±242mV. That would be the range of where the offset error should stay within (with those particular settings). If I estimated things right in my head, that corresponds to about 6 LSBs.

[mroek]

@bugi: Thanks, that makes sense. And if that's indeed correct, then my ~ -120 mV is well within spec, even if it triggers my OCD to see that the trace is clearly separated from the zero position marker. I'm not used to that with my old TDS210, where the trace seems to always align pretty close to the position marker.

[tinhead]

trace is clearly separated from the zero position marker.

check PM

EDIT: i got beta firmware and was hoping it will help him, but it was not the case

[mroek]

The beta I got from tinhead did seem to change the problem a little. It appears to improve on the self calibration in some ranges (so my 2V/div range is now quite OK), and worsen it in the high sensitivity ranges.

I now have a different question. There seems to be some visual artifact at the point where the trigger is. Take a look at the following screenshot, where I am just feeding in a DC voltage (around 7.5 V) to channel 1, and with the channel in AC coupling and at 10 mV/div:



In the center of the screen, there's a very visual notch/stair in the waveform, just at the trigger level. Is this to be expected? I can't really understand why, but it does appear to be just a visual artifact. For the record, I can input a single clean sine from a function generator, and the notch is visible also in that scenario.

[rf-loop]

In the center of the screen, there's a very visual notch/stair in the waveform, just at the trigger level. Is this to be expected? I can't really understand why, but it does appear to be just a visual artifact.

This is normal (and visual effect is normal with noisy signal). One TFT frame image you have perhaps (depending case) tens or even thousands of "horizontal sweeps" aka acquistions overlaid. (example if true wfm/s is around 25000 then every TFT frame have around 1000 waveform ("horizontal sweeps") overlaid. Every single horizontal sweep need adjust as exactly to trigger time position as possible including also "fine interpolation" (what is not main reason in this case, due to t/div and samplerate). After last acquistion there is some blind time and after then it do least fixed length pre trigger acquistion and after then it start waiting trigger and after trigger it continue to rest acquistion up to max length. And this acquistion is also horiozontally fine positioned so that trigger position is exactly in trigger time position. (here center of screen)

You can also look it more. Stop scope after running some time and turn on History and then look every separate acquisitions or run history more fast.  Also you can watch it running scope normally but switch acquistion mode Fast or Slow. In Slow mode one TFT frame show only one horizontal acquisition. To fully explain how it all works need write one book. But I think it can understand enough without going more deep what happen there. Only need know, it works just ok. If it do not this then we can think trigger engine is poor.


Here this "notch/stair"

[n3mmr]

What i was after was a command that would put it in a bode plot mode with the parameters exposed so I could set the center, span, etc via scpi,

Officially; Bode plot SCPI command set: totally unsupported at this time.

Rerouter would like it, please raise hands who else.

 

[mroek]

@rf-loop: Thanks for taking the time to write down that explanation. As I wrote, I thought it was just a visual effect, and you explained why.

I have yet another question on the offset issue. On my scope, the offset is different for each coupling type (DC/AC or GND). Ideally, the offset should be zero for any coupling (when there is no input signal), and even if it may not be a real problem, it would have felt better to see that the offsets was zero. Some channels are "better" than others, though.

I guess there are many owners of this scope that reads this thread, and I am curious if this is normal for this model.

[tinhead]

I am curious if this is normal for this model.

actually it might depend on model, hw revision, etc., i do have same hw as yours, so let me check later what results are on my SDS

EDIT: yep, that seems to be model specific thing, tested with various settings (but always inputs set to 20MHz and GND, x1, hw1.4), i can get always on some channel 0V and on other some uV offset, worst case was 199uV for CH3 at 1mv/DIV. The good news, with increasing DIV/sec (from 20mv/DIV) it remains low in compare to DIV setting (e.g. 550uV for CH3 at 100mV), 0->0,25%->0,5% of DIV/sec setting, so can be ignored. For the 500uV it is on my SDS better as with 1mV/DIV, 2mv/DIV and 5mv/DIV are "equally bad" on some channels and differs on others (e.g. 200uV at 2mV/DIV, and 200uV at 5mV/DIV).

For DC/AC offsets, they of course depends on environment noise, coupling, DIV setting, so it does not matter to compare anything.