SDS 1104X-E I could use some help capturing DC

[slodat]

I'm realizing how oscilloscope dumb I am.

I'm not concerned with where each channel's trace is in the gridplot. I'm only looking at +/- deviation from 1.65vdc. I move the zero's up and down so I an see all three drives at one time.

I measured the noise, as reported above, with my DMM as second harmonic (120hz) 30ish vac. I don't know how to get rid of it.. I'm sure this is easy, but it's kicking my ass.

[slodat]

Nope, some user error.

Look at the 2 images and in particular the trigger position in relation to the selected Trigger's waveform and trigger settings are also different.

IF in both these screenshots ch1 is on the same source there is indeed a mismatch of ch1's signal in relation to ch's 2 and 3.
We can examine each in more detail by going to AC coupling and/or increasing sensitivity and having the channel 0V level off the display (within datasheet limits).

All three channels are on the same signals in both screenshots.

[tautech]

I'm realizing how oscilloscope dumb I am.

Not really as diving in doing 3 channels to start with can be pretty daunting.
Just using one channel takes a little bit to understand sound scope usage let alone 3.
Ya doin alright. 

I'm not concerned with where each channel's trace is in the gridplot. I'm only looking at +/- deviation from 1.65vdc. I move the zero's up and down so I an see all three drives at one time.

Yep, when using DC channel coupling that is what you need to do.

I measured the noise, as reported above, with my DMM as second harmonic (120hz) 30ish vac. I don't know how to get rid of it.. I'm sure this is easy, but it's kicking my ass.

Have a play with Averaging or E-Res. Noise can be a curse but it can also indicate there's another issue that needs addressing.....when we see it on the display we then have to determine if it's a problem, something we can ignore or something to use the tools within the scope to clean up.

I'd be more concerned if the scope didn't display all the noise......environmental and circuit noise.

Threads like these are great resources for the DSO newbie as they give lots of hints of operation and things to try.
Please carry on............ 


BTW, get some sleep as it must be late in the US.
Others will chime in and we can have another go tomorrow.

[StillTrying]

This is what I'm going for (yes, I regret returning the other scope

Don't worry - yet. I think when you find out what's causing the noise the X-E display should should look at least as good as the Instek.

I don't think you can use AC coupling at these very low 5Hz signal and 1s/div. speeds, but you can use the DC offsets.

The noise when you connect the other channels is very strange, are you sure the GND gnd clips are alright and all connected to exactly the same GND.

As others have said you could use Hi-Res, not waveform averaging, at low sample rates some scopes use it in the background anyway, the Instek could have been, - I don't know.

I suppose you could try it with the other 3 or 4 channels switched on but without the probes connected to the BNCs.

[StillTrying]

BTW, get some sleep as it must be late in the US.

It's even later in the UK, and the clocks have gone +1 hr. so it's become early again.

[nctnico]

IIRC you have the GDS1054B as well. Use that and enable input filtering. It will get rid of the noise your probes are picking up.

That's the problem.. it already went back. So, I'm trying to figure out how to do this with the Siglent.

Ah... you have thrown the perfect tool for this job out. What is left is to turn on bandwidth limiting and hi-res mode to get a cleaner signal.

[slodat]

The ground clips are connected to their respective drive’s pin1. As I mentioned, there’s significant second harmonic voltage on these signals. When I connect the ground lead to earth ground all I see is the 120hz signal. I don’t think I can do this with this scope.

[slodat]

Ah... you have thrown the perfect tool for this job out. What is left is to turn on bandwidth limiting and hi-res mode to get a cleaner signal.

I know. I’m not impressed with my quick decision. At all. 

I have the bandwidth limit on and eres didn’t do anything to make the dc signal pop out any better.

I don’t know what else to try on this scope. Wishing I wouldn’t have chosen between the two so quickly.

[nctnico]

An alternative would be to try a differential probe or differential amplifier. That way the common mode noise gets suppressed.

Edit: something else you can try is not to use the probes but connect coax cables directly to what you are measuring and don't strip the shielding of the coax any further than you have to.

[rf-loop]

Ah... you have thrown the perfect tool for this job out. What is left is to turn on bandwidth limiting and hi-res mode to get a cleaner signal.

I know. I’m not impressed with my quick decision. At all. 

I have the bandwidth limit on and eres didn’t do anything to make the dc signal pop out any better.

I don’t know what else to try on this scope. Wishing I wouldn’t have chosen between the two so quickly.

Of course there is different scopes and different methods to filter out higher frequencies. Some scopes, like example GW have user settable low pass, band pass and high pass filters and quite useful in some situations but not all. Also some older Siglent have these. But then it need also note that these are made in digital side (DSP). In some cases, my opinion is that even they works ok many times but -  as told they are digital side. If example low pass filter freq is set for 100kHz how it know if signal is true signal or false signal produced internally before this digital filter.
So or so, I do not want start infinite unnecessary debate about this.

If you really have there high noise level in signals, in signal itself and more due to noisy environment and noisy ground and ground loops.

You can - if you have available - differential probe.
But if there is still in signal line itself high noise...

If you have some materials and tools... and can do some experimental things.

You can filter noise out _before_ scope input.

Because I do not know enough details about your signals under tesdt and what you need look, exmple what kind of risetimes/frequencies you need look and what are these what can filter out.
Because I do not know your signal true level related to ground I can not recommend next.
Set probe to 1x (and scope). Take 1kohm resistor. Other end to probe tip and other end to signal. Then connect 0.1uF capacitor from probe tip to probe ground (just this ground what is around 5mm after tip. (take this hook out first). Then look how it looks now. After then, depending result, can think more.
(this is 1kHz 1st order LPF what give 20dB attenuation for decade. It is not lot but it may give some idea what to next. You can do same with cheap 50ohm or 75 ohm coaxial.
Because, what ever skope yuu take if there is very heavy noise and it go to scope front end and to ADC it may still rotten your result. So it is some cases quite wise use filter before scope input.

Your connection to signal in one image looks really poor. It is like "please come in all noise..please come all in... "

Also your probe... (in image I can see setup with 10x) it is full band in 10x
Scope input 20M BW is also something like 1st order filter sxo even it do not attenuate even 40 - 100MHz very well.

But this same probe in 1x mode, it acts also like 1st order LPF around 5 - 8MHz, together with 20M BW you have bit bnetter attenuation but 0 - several MHz come in without attenuation. Do you need look 10MHz or 35ns risetimes. I think not... 
It do not cost lot if you make some trial with simple external RC (but also this, do it so that it also attenuate and is not  like antenna for collecting all noise.

I think whole problem is "how to connect this signal to scope" (and how to filter it before scope input)
There is no scope what have internal analog side low frequency low pass settable filters. Least I do not know any - and over 50 years have not seen any. But very often have used external filters before oscilloscope inout and in some cases they are really easy to do (exept if need  more steep that 2nd order)

You can start with one resistor and one capacitor. After then if it looks anyhow useful you can think how to do it bit more better and usable. Test do not cost nearly anything. Just think what are highest frequencies/fastest risetimes what you need see reliable without attenuation... this is starting point.

[StillTrying]

I've read the thread and the servo manual, all I can say is that this seems to prove that it's the connecting of just 2 or more grounds that causes the noise. 

[slodat]

I think so, too. But, I don't know what to do about it. It appears each drive's internal power supply is floating..

[Martin72]

As written here before, try the Enhanced Resolution mode:

Normal Mode siglent:




Hi-Res Mode (+3bits) :



Or, try to connect a little cap (1n..10n)on the bnc input of the scope ( by using a bnc t-adapter)

[rf-loop]

Is there a way to filter out the ac and just show the dc?

Because some times also other peoples may play with same kind of situation.

I will show one "poor man's" solution. Not rocket science but extremely simple (and poor) but some times still usable and may solve some problems.

Method is extremely simple and most can do it (if have BNC T and some capacitors.

Just like this:



With this I've done some simple experimental measurements.

First, PP215  1x (1:1) frequency response (source 50ohm)


Measured using SDS1104X-E and one Siglent SDG function generator. (Not SAG1021)
 SFRA dB scale is just so that signal trace left can think as "0dB" reference. (least now this FW version there can not se "offset" so that can set 0dB to my reference level,)




With this probe I measured 1.5V DC line (with extremely heavy noise)



Before filter:
You can see V/div is 20mV and offset 1.5V and heavy noise over screen but level roughly 1.5V.


Ok, now I can not shut off this somewhere coming random noise. What to do, I want see this "DC" as  OP want.
In this case I want still see DC level fluctuation and possible some 100Hz ripple and other things up to near 1kHz so that these things are still roughly level accurate. Then attenuate higher frequencies.

I made RC LPF using this BNC T method and look its frequency response.

 

Frequency response: Probe 1x, and capacitor (sum of 10nF+10nF+470nF parallel capacitors =490nF)
We can see corner is around 850Hz (Im lazy and not want anymore solder or remove)


And how this 1.5V rail now looks.
 


After filtering. Compare image before filtering. Signal is exatly same only added this BNC T between scope input and probe.
Now I can see this 100Hz 20mV ripple and also of course if this "DC" is other ways drifting or fluctuate.
(do not care right top corner frequency meter, it is nearly nonsense - noise and partially also due to method how I make this GIF animation)
Also you may note I have dropped samplerate. No problem because scope can not see so much >fNyquist frequencies due to filter before scope.  But reducing samplerate (memory length) I also can get more acquistions to history buffer (in this case around last 15 seconds what is nonsense for just show this filter method.

Of course with this method can use with different capacitors and also it can use when probe is in 10x (1:10). Example under 1Hz... or what ever.

Here 40kHz LPF (probe 1x capacitor in BNC T  10nF)



It is not good to stand hands in pocket if there is a problem that needs to be solved. Quite often, some help can be found by very simply. Simple is beautiful. Or should I say the old way can sometimes be better than whole bag of new ones.

[Martin72]

I will show one "poor man's" solution.

As I said before:

Or, try to connect a little cap (1n..10n)on the bnc input of the scope ( by using a bnc t-adapter)

Thanks for the pics ! 

Martin

[rf-loop]

I will show one "poor man's" solution.

As I said before:

Or, try to connect a little cap (1n..10n)on the bnc input of the scope ( by using a bnc t-adapter)

Thanks for the pics ! 

Martin

Yes

...and as short wires as possible...etc things specially if noise have HF components. (also there can see in my BodePlots, if want good, need also select what kind of capacitors... in this 800Hz LPF they are bit "bad" as can see.)
And of course 10x probe need _very_ different capacitance what 1x  not at all like tenth of value due to very different R before C and after C.

Old simple trick - used tens of years.

Other old trick for other purposes is using scope DC coupled but do AC coupling external for more low freg or more high freg.
Siglent AC coupling corner freq is bit over 1Hz (very low if compare many other scopes)

[Martin72]

The "trick" with the LPF we must use recently for measuring the output-ripple of a dc/dc converter.
In this case it´s not enough to set the coupling to AC, you´ll see everything but not the "true" ripple.
After using a 1n cap directly to the input, the noise disappear and you see the switching frequency of the converter, that´s the ripple and nothing else.