Siglent SDS3000X HD and upgraded SDS1000X HD

[markus_jlrb]

0.315V / SRT(2) = 0.222(7)V about 0.223V as on display (RMS(C1)-Value)

What is the point?

Sorry not caught your issue.

Markus

[Grandchuck]

I guess just verifying 0 dBm

[KungFuJosh]

I think Dave needs to add a DBAD license to the user agreement on sign up. 🙄

People answered you in multiple different ways, and now you're insulting them because you didn't like their answers?

[2N3055]

@2N3055

The sopce showed at max sensivitv (500uV/div 50 Ohm)
a noise of about 953 nV (CH1, CH2), negligible difference.

This a noise power of (953nv * 953nV)/50=18.15fW or 10xlg(18.12pW) = -107.4 dBm

The scope then has a very good (for a scope) NF (noise figure)
of (-174dBm - 107.4dBm) = 66.6dBm

At a sensivitv of 1V/div it showed 102.7 uV (CH1) and 398.3uV(CH2)
This gives noise powers of 210.9 pW or -66.76 dBm (CH1) and
3,173 nW or -54.99 dBm (CH2)


This is a abvious diffenence and my quesitions were
a) What is the possible cause of this difference
b) What differences other owners of the same scope are seeing.

The amount of work to check b) and to answer this question is minimal,
up to now no other owners have answered.

I know these difference compared to full scale (8V) are negligible.

I have made no statement regarding the quality of a scope that
shows such differences between channels nor about the influences
of measurement failures caused by such differences.

But I got a strange answser from a Siglent fanboy who showed me
how low the differeces the of the mean(!) values are.
Other thanked him for this discovery.

Firstly. I will politely point out to stop calling anybody fanboy. If you want help, insulting people will not make them want to help you. You would think that would be obvious.

Second, your English is not perfect so it is hard to understand what you mean. That also mean that you have to give people benefit of the doubt and be patient with answers. And maybe explain more before getting snarky.

Third, it is you that are wrong. You made a mistake. You are looking into wrong place.   You have Stdev(C1) measurement. For that measurement you have statistics, vertically.  Value (current), Mean, etc..
Those are statistical values. StDev on the bottom of the stat is StDev that describes how stable is the measurement of the actual StDev (AC RMS) level of the signal.

Value of AC RMS of the signal that you need here is Mean of StDev(C1).
110µV RMS of noise at 500µV/div  and cca 5 mV RMS for 1V/div.

So "fanboy" is right and you are wrong.. How that happened? By rushing to conclusions with partial data.

Why is RMS of noise not perfectly stable so there is amplitude variation?  Because source of noise is not leveled calibrated noise source. Source is internal scope noise that is complex noise with periodic and aperiodic components. And since you are measuring  with rather short timebase, by definition of noise there will be differences.
And as theory goes, noise averages to infinity time. So if you select longer timebase and let scope gather statistics long enough, the  difference in variations will be less and you will get idea of what is real difference in RMS noise between the channels. By comparing Statistical Mean of respective AC RMS (stDev) values. Any difference in stability of measurements between channels (statistical StDev on the bottom) will be because internal sources of the noise for each channel are not going to be the same. Ever.
At this point I would like to ask if I understood correctly and answered what you meant to be your question. Or I am wrong, in which case you should try to explain what you meant.

Best,

[kladit]

@2N3055

You were rigtht. I looked at the wrong values, not stdev but stdev(stdev).
I repeated the measurment after a self-calibration of the scope.

This time I have two points.
First the same a before +
CH1 mean(stdev) 5.268 mv
CH2 mean(stdev) 5.301 mv
CH3 mean(stdev) 5.701 mv
CH4 mean(stdev) 5.815 mv

I assume CH1 and CH2 share an A/D-converter and CH3 and CH4 the second one.
That is a diff 0.8dB and ok, but something one has to keep at back of his
mind.

Second is the difference between RMS and stdev.
The scope was warmed up half an hour before self-calibration was started.
I owned a Siglent SDS2354X Plus before and I remember that past
a self-calibration the differences between rms and stdev were
remarkable mininal.
So I assume the self-calibration routine could need some improvement.


So I am still interested in the values others see repeating
the same measurement at same conditions at their scopes.

Another observation is that were the rms differences to stdev
values indicate an dc offset (past input condenser, I measured AC)
the stdev(stdev) is high.
Has anybody idea what could cause this?

At 0.5mV the lowest noise is at CH1 with 109.4 uV ant 50 Ohm.
This result in a noise power of P = 239.37pW or 10xlog(P x 1000) = -66.2 dBm
at  full bandwith.

Divided by the bandwidth of 750MHz (datasheet) this gives -154.95 dBm/Hz
If the bandwith is ca. 1.2 GHz (meassured) this gives -157 dBm/Hz.
Both are very good noisefigures of 19 and 17.


I deleted my previous post and with it the fanboy statement.

[egonotto]

Hello,

according to my observations, the values for other SDS3104X HD are very similar.
I think the noise power is not evenly distributed in the frequency range. Otherwise you would see a halving of the noise with a bandwidth restriction to 200 MHz. But that is not the case. Even the use of ERES does not provide a particularly good noise reduction.
My further assumption is that a relatively large proportion of the noise is in the low-frequency range.

Best regards
egonotto

[2N3055]

@2N3055

You were rigtht. I looked at the wrong values, not stdev but stdev(stdev).
I repeated the measurment after a self-calibration of the scope.

This time I have two points.
First the same a before +
CH1 mean(stdev) 5.268 mv
CH2 mean(stdev) 5.301 mv
CH3 mean(stdev) 5.701 mv
CH4 mean(stdev) 5.815 mv

I assume CH1 and CH2 share an A/D-converter and CH3 and CH4 the second one.
That is a diff 0.8dB and ok, but something one has to keep at back of his
mind.

Second is the difference between RMS and stdev.
The scope was warmed up half an hour before self-calibration was started.
I owned a Siglent SDS2354X Plus before and I remember that past
a self-calibration the differences between rms and stdev were
remarkable mininal.
So I assume the self-calibration routine could need some improvement.


So I am still interested in the values others see repeating
the same measurement at same conditions at their scopes.

Another observation is that were the rms differences to stdev
values indicate an dc offset (past input condenser, I measured AC)
the stdev(stdev) is high.
Has anybody idea what could cause this?

At 0.5mV the lowest noise is at CH1 with 109.4 uV ant 50 Ohm.
This result in a noise power of P = 239.37pW or 10xlog(P x 1000) = -66.2 dBm
at  full bandwith.

Divided by the bandwidth of 750MHz (datasheet) this gives 154.95 dBm/Hz
If the bandwith is ca. 1.2 GHz (meassured) this gives 157 dBm/Hz.
Both are very good noisefigures of 19 and 17.


I deleted my previous post and with it the fanboy statement.

Thank you. Respect.

As I said before, difference between the channels is because the channels are not identical. Different front ends, consisting of many individual components in a different physical location. Even the channel pairs signals that end up in the same ADC chip are going to two separate ADCs inside chip...

Again that is a cca 500µV RMS difference in noise floor on a 8V P-P dynamic range... If we input full scale sinewave on that range it would have 2.828 V RMS value.  That means that variation in noise floor in relation to signal would be cca 193 ppm (parts per million) or 0,019 %..

Scopes are not 6.5 digit DMMs. 

As for difference between StDev and RMS, difference is there because StDev is AC RMS, and RMS is AC+DC RMS. So RMS will include any channel offset.

I will take the moment to point out this is documented in User Manual.

As for comparison with your previous scope, because it was 8 bit scope there was limit to discrete steps it could resolve. Meaning it was rounding to nice clean numbers.... Hence no such details..

[KungFuJosh]

I will take the moment to point out this is documented in User Manual.

I think that might be the most polite way anybody on this site has ever said RTFM. 😉

[kladit]

@egonotto

I discovered a higher frequency signal inside the noise.
No clue where it comes from.

[egonotto]

Hello,

it is not clear to me whether the signal determined by average is genuine or not.

Best regards
egonotto

[2N3055]

@egonotto

I discovered a higher frequency signal inside the noise.
No clue where it comes from.

If you explained what are you trying to accomplish here maybe someone could help..

Gathering averages on randomly triggered signal will yield nonsense..

You already used FFT that have shown you the noise floor, and all individual spurs..
If FFT does not see it, it is not there...

[egonotto]

Hello,

inspired by the new thread about a video about the R&S MXO58 by Linus, I watched a few videos about the MXO5 at Batronix. Among them the video “Superior Spectrum”.  I wanted to try out what was shown there with the SDS3000X HD.

Here is a screenshot of the video and my results with the SDS3000X HD.

Best regards
egonotto

[hpw]

@egonotto

How many averages did you used? Or simple none?

Hp

[2N3055]

@egonotto

How many averages did you used? Or simple none?

Hp

That is without averaging. When you enable averaging, it will say on the screen.

[Martin72]

Egonotto's last FFT image, recreated with my 3104X HD, once without(1) and once with averaging(2).
Plus one with averaging and a 200Mhz bandwidth limit(3).
No real comparison, because not the same bandwith range:
Magnova BM350 pics.

[egonotto]

Hello,

I don't understand the -160 dBm.

Best regards
egonotto

[Martin72]

Hi egonotto,
You can set the “load” for the dBm unit in the menu, so I have set it to 1M.

[mawyatt]

That can be misleading as the noise power of free space is ~-174dBm (kTB) in 1Hz bandwidth, so anything less than -174dBm is either in less than 1Hz bandwidth or lower temperature.

Best

[hpw]

Hi egonotto,
You can set the “load” for the dBm unit in the menu, so I have set it to 1M.

Known references to me are
50E, 75E, 600E. 1M never ever seen so far. So -160 is misleading by 1M IMHO.

[Martin72]

If you do not have a 50, 75 or 600 Ohm load on it, but 1M, it is just as misleading to pretend that you have a 50, 75 or 600 Ohm load on it.

[hpw]

If you do not have a 50, 75 or 600 Ohm load on it, but 1M, it is just as misleading to pretend that you have a 50, 75 or 600 Ohm load on it.

Just look on a professional meter.. see attached.

[egonotto]

Hello,

I have difficulties with the display of slow signals.

Suppose I want to measure the discharge curve of a battery over a period of 10,000 seconds.
I would then like to save the result.

How can I do this with the SDS3000X HD?
I would also like to see the signal during the measurement.

Best regards
egonotto

[2N3055]

Hello,

I have difficulties with the display of slow signals.

Suppose I want to measure the discharge curve of a battery over a period of 10,000 seconds.
I would then like to save the result.

How can I do this with the SDS3000X HD?
I would also like to see the signal during the measurement.

Best regards
egonotto

You should be able to use ROLL mode and then save data.
If you use trigger (capture start condition with trigger) and start ROLL with single, it will Roll until end of screen.

Then save BIN or CSV.

I tried on 3000xHD and it worked fine.

[egonotto]

Hello,

thank you, but then I have the following problem. If I have no signal to trigger, the roll mode runs continuously in both normal and single mode.

Best regards
egonotto

PS: What I have just noticed is the following. In roll mode you can have a maximum of 100 MSamples.

[2N3055]

Hello,

thank you, but then I have the following problem. If I have no signal to trigger, the roll mode runs continuously in both normal and single mode.

Best regards
egonotto

PS: What I have just noticed is the following. In roll mode you can have a maximum of 100 MSamples.

One other thing you could do is do Trend plot. That will log data in long term.
In which case you set scope to measure voltage on some faster timebase, and your battery discharge plot will be visible in Math window, not main window. I need to check if that can be saved. But there you will not have time limits.