Products > Test Equipment
Using (and living with) Siglent SDS1000X-E Series Oscilloscopes
bdunham7:
--- Quote from: SimonM on June 30, 2021, 05:41:28 pm ---6. The scope can measure skew and offsets between channels, it just treats them separately e.g. scope is 2 + 2 channels.
--- End quote ---
The scope is 2+2 channels as far as its ADC architecture goes, but they should be aligned. This would be a huge issue if you were looking at 4 logic channels, for example. Did you contact Siglent?
SimonM:
--- Quote from: bdunham7 on June 30, 2021, 05:46:53 pm ---
--- Quote from: SimonM on June 30, 2021, 05:41:28 pm ---6. The scope can measure skew and offsets between channels, it just treats them separately e.g. scope is 2 + 2 channels.
--- End quote ---
The scope is 2+2 channels as far as its ADC architecture goes, but they should be aligned. This would be a huge issue if you were looking at 4 logic channels, for example. Did you contact Siglent?
--- End quote ---
Agreed, they should be aligned and the spec for horizontal alignment is < 100ps across the scope (page 9 of the spec sheet). Agreed, it would be a huge issue for looking at anything that's 4 channel and would imply phase shifts that don't exist. It might also affect the decoding of a serial bus if the clock and data were split across channels 1 and 3.
As I get to know the scope, I'm starting to use channels 1 and 3 for dual channel use. This provides the full 1GSa/s sampling rate. If it's not aligned between the first and second pair of channels e.g. separate ADC chips, then the extra sampling can't be exploited. One of the reasons I bought the SDG1032X FG was the relative ease of evaluating bode plots. The outcome of this is to show frequency and phase shift, so I would like the measured phase to be accurate. I haven't tried it so far, but I would expect to use the best pair of channels e.g. 1 & 3, and not simply 1 & 2.
Simon
rf-loop:
--- Quote from: SimonM on June 30, 2021, 05:41:28 pm ---
--- Quote from: tautech on June 30, 2021, 04:35:06 am ---Channel skew is specified on P9 in the datasheet:
Channel Skew <100 ps
https://www.siglenteu.com/wp-content/uploads/dlm_uploads/2020/12/SDS1000X-E_DataSheet_DS0101E-E04C.pdf
Whenever you strike something that doesn't look right use Default to bring settings back to factory then run Self Cal and revisit the measurement.
For BNC connections an unmatched pair of cables can often produce different results.
--- End quote ---
Thanks.
I am returning the scope to the factory setup and running Self Cal before revisiting the measurements.
The BNC connections are with/without the 50-ohm termination. Results are unchanged.
The cables are pretty well matched - they are the 4x PP510 probes. Swapping the probes around, the results are unchanged.
Here is what I did to simplify the setup (one channel of FG directly connected via two probes to the scope)
I have removed the 50-ohm terminator and setting the FG to Hi-Z. The setup is about as simple as it can be.
There is a setting for Quick-Cal on the scope. I hadn’t noticed it before and it was switched ON. I switched it OFF because the condition of varying temperature doesn’t apply.
I have reset the scope as follows:
1. Reset to factory defaults;
2. Use the scope Self Cal calibrate operation (takes about 5 minutes);
3. Press default button;
4. Set all four probes to 10X;
5. Set all four channels to the same 2V/div, 0V offsets;
Without anything connected, the four channels are overlain at 0V.
Set the FG to 10MHz and Hi-Z. Channel 2 is off. There is no 50-ohm terminator.
1. Channel 1 is triggered and displays the waveform with a probe point directly off the FG;
2. Channel 2 is added and the probe is connected to the same FG output. Both channels are in phase;
3. Move the 2nd probe from channel 2 to channel 3. The scope is out of phase;
4. Move the 2nd probe from channel 3 to channel 4. The scope is out of phase;
5. Move the 2nd probe from channel 4 back to channel 1. The scope is in phase;
When the scope is in phase the 2nd trace covers the first trace, otherwise, they are out of phase.
I can manually skew a second channel to overlay the first trace. -29.35ns is required.
NB without the 50-ohm terminator the results are as before, there is one difference that is expected: displayed trace doubles (Vp-p) because there is a change in the load impedance when connecting/disconnecting a second probe.
At 1MHz and probes set to 1X, timebase altered and scope input channels set to 1X, 2V/div, 0V offsets, it shows the same waveform.
Adding the second channel, overlays in phase, moving 2nd probe to channel 3 and then to channel 4 also shows the channels out of phase.
At 1MHz the amount of phase shift on the screen is reduced e.g. to 1/10. Skew to correct this is again best at -29.35ns.
Switching between probes doesn’t vary/alter the result.
Observations and conclusions
1. Removing the 50-ohm terminator doesn’t vary the out of phase outcome;
2. Varying which probes are used doesn’t vary the outcome;
3. Changing frequency e.g. between 10MHz and 1MHz and using probes 1X doesn’t vary the outcome;
4. The scope doesn’t show phase alignment between channels (1 & 2) with channels (3 & 4) after calibration;
5. The scope is remarkably aligned within the channels (1 & 2) or channels (3 & 4);
6. The scope can measure skew and offsets between channels, it just treats them separately e.g. scope is 2 + 2 channels.
rf_loop provided a review of the scope and his analysis of skew indicated a very high degree of alignment (see page 76):
(Attachment Link)
NB the only mismatch is with attenuation (might be expected) and not with skew - each channel is very well aligned.
Simon
--- End quote ---
Not me. I think it was @Performa01
I do not remember/know with most new FW if there is still total deep reset.
During whole power up boot time toggling very fast Math button. If it works you know it because you can see it have also reset calibration. If not work first time try again... As far as I remember factory reset (via menu) do not deep reset so that also all cal data is reset. (@tautech can perhaps confirm if this method is still there today)
And of course then calibration and -nothing- must be connected to input BNC's. Nothing - it really mean nothing. (how it affect it also depends scope model, in some old model whole internal cal signal come to BNC centers but do not know SDS1004X-E, I do not have this model available here for tests, these are in my homeland, not here in China)
Now if I understand right you have skev betwee (Ch1/Ch2 pair, (ADC1)) and (Ch3/Ch4 pair (ADC2)) roughly near 30ps. Ande this is really lot of. Example 6 meter RG316/U have roughly 29ns travel time.
There is very poor simple and dirty method. "Wilderness and desert method. Just for bullet proof verify the problem is or is not."
ETA After seen your video:
Yoy can Connect your BNC T to channel 2 and trigger to CH2. Connect just one your coaxial from SDG to this T other side. Then take just very short copper wire and connect it from BNC T other side center as short as possible to CH3 input center. Now you have only one coax from SDG and very small, and roughly known, signal travel time difference between CH 2 and Ch3 input. Just 1kHz is signal where is fast edge is enough (square, pulse or 0% triangle). With this, because you do not have better things (special time matched cables and good splitter etc) available, you can infinitely proof if there is error between channels (when they are different ADC).
tautech:
--- Quote from: rf-loop on July 01, 2021, 07:24:18 am ---I do not remember/know with most new FW if there is still total deep reset.
During whole power up boot time toggling very fast Math button. If it works you know it because you can see it have also reset calibration. If not work first time try again... As far as I remember factory reset (via menu) do not deep reset so that also all cal data is reset. (@tautech can perhaps confirm if this method is still there today)
--- End quote ---
Those are enough as we should not share the other dealer secret Reset's we have. ;)
However there still is Secure Erase in the Save/Recall menu used for deleting User settings such as LAN/WiFi and Reference waveforms.
--- Quote ---If with this desert method there still is 10's of ns difference, if even ns difference, it is clear your scope have some severe problem what need Siglent service.
--- End quote ---
Soon to unpack my old prerelease SN#12 SDS1104X-E and check for channel skew and will post screenshots......
Channel skew tests.
Source SDG6202X square wave terminated with Tek feedthrough to Tee and to scope inputs with Siglent 1 GHz rated 1m BNC cables.
Dot and Vector displays.
SimonM:
--- Quote from: tautech on July 01, 2021, 08:21:49 am ---Those are enough as we should not share the other dealer secret Reset's we have. ;)
However, there still is a Secure Erase in the Save/Recall menu used for deleting User settings such as LAN/WiFi and Reference waveforms.
--- End quote ---
Thanks, I have used "Secure Erase" that removes user settings and the "Hard Reset" documented here: https://siglentna.com/operating-tip/oscilloscope-hardware-reset/. "Secure Erase" is suggested to remove the user settings e.g. if you were about to sell the scope. The "Hard Reset" is recommended for issues with start-up and whilst it certainly delays the bootup, it doesn't affect the ~ 30ns skew between pairs of channels (separate ADCs).
I sent a follow-up video (using DropBox) to my Siglent dealer showing the setup of the FG and channels on the scope:
https://www.dropbox.com/s/lv0rczm5bcguz9k/IMG_1965.MOV?dl=0
The movie shows all channels show deskew 0. At 1MHz the two traces separate (widen) using input channel 1 and moving 2 -> 3 or 2 -> 4. At 15 MHz the out of phase is obvious as input is moved across to 3 or 4. Emailed the dealer screen prints. With timebase set up for 15MHz sine wave, the 1MHz or 15MHz traces show a shift of 1 division when 2nd ADC is used. Ideally, this wouldn't need to be manually corrected by remembering to set a deskew of ~ -30ns on channels 3 & 4. Hopefully, there is an easy re-calibrate fix, or I (aka the 'rookie') overlooked something. ;)
Simon
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version