Author Topic: Rigol DS1054Z not as precise as hoped, or is it my power supply, or ...?  (Read 1859 times)

0 Members and 1 Guest are viewing this topic.

Offline carldrTopic starter

  • Newbie
  • Posts: 4
  • Country: gb
Hi all,

I received my DS1054Z yesterday, and was playing with it this morning.  My first test was to test the output of a simple astable 555 circuit I set up on a breadboard which is driving a LED.  The outputs are the capacitor on pin 2, and the LED on pin 3.

The traces on the 'scope don't look as precise as I was expecting.  What I don't know is whether it's my use of the oscillioscope, or the PSU, or the circuit, or just whether my expectations need resetting.  The scope is a Rigol DS1054Z as mentioned, the PSU is a Maplin N93CS (which I believe is a re-branded Manson NRP-2050.)  I've calibrated the 'scope's probes.

The output looks like this (click to view hi res) :

985042-0

I tested the power supply output directly, set at 5.0v, and zoomed in on the trace somewhat, which shows about 100mV variation.  The specs of the power supply says the output ripple ought to be <5mV.  That's whilst powering the circuit.

985046-1

What are your thoughts?

Thanks,
Carl.
« Last Edit: May 06, 2020, 08:47:49 am by carldr »
 

Offline thinkfat

  • Supporter
  • ****
  • Posts: 2150
  • Country: de
  • This is just a hobby I spend too much time on.
    • Matthias' Hackerstübchen
The DS1054Z has an 8 bit ADC, that's 256 steps. It has 400 vertical pixels in the waveform area of the screen. That means, 1 ADC LSB is about 1.5 pixels. As the screen outresolves the ADC, this is what you'd expect - clearly identifiable steps.

On the output ripple - this scope doesn't have a world-class front end so of course there's some noise to be expected, but not in the 100mV range. So what you see there is definitely something that comes in from the probe. Whether it's real or some probing issue I cannot say.
Everybody likes gadgets. Until they try to make them.
 
The following users thanked this post: carldr

Offline RoGeorge

  • Super Contributor
  • ***
  • Posts: 6145
  • Country: ro
Something is not OK, especially in the second pic.
 
The following users thanked this post: carldr

Offline newbrain

  • Super Contributor
  • ***
  • Posts: 1714
  • Country: se
Something is not OK, especially in the second pic.
In the second pic, the OP is measuring a 5V supply on a 200mV/div. This means that the 0 V level is 25 divisions below the shown trace - in this case 21 of them are off screen.

The DS1054Z is not known for having a wide range of input offset, and saturation of the front end is quite possible.
Try to repeat the same measurement with the channel AC coupled.

I see nothing wrong with the first picture, it looks quite clean to me.

Edit: found the specs, the vertical offset range is:
Quote
Offset Range:
  1 mV/div to 499 mV/div:  ± 2 V
  500   mV/div to 10 V/div: ± 100 V 
So, with -5V offset at 200mV/div the OP is quite outside the acceptable offset range.
Nope, see below!
« Last Edit: May 06, 2020, 10:26:31 am by newbrain »
Nandemo wa shiranai wa yo, shitteru koto dake.
 
The following users thanked this post: carldr

Offline thinkfat

  • Supporter
  • ****
  • Posts: 2150
  • Country: de
  • This is just a hobby I spend too much time on.
    • Matthias' Hackerstübchen
Something is not OK, especially in the second pic.
In the second pic, the OP is measuring a 5V supply on a 200mV/div. This means that the 0 V level is 25 divisions below the shown trace - in this case 21 of them are off screen.

The DS1054Z is not known for having a wide range of input offset, and saturation of the front end is quite possible.
Try to repeat the same measurement with the channel AC coupled.

I see nothing wrong with the first picture, it looks quite clean to me.

In ranges < 500mV/div the offset range is only +/- 2V, but with a 10x probe, that's still +/- 20V.

Everybody likes gadgets. Until they try to make them.
 
The following users thanked this post: newbrain, carldr

Offline newbrain

  • Super Contributor
  • ***
  • Posts: 1714
  • Country: se
In ranges < 500mV/div the offset range is only +/- 2V, but with a 10x probe, that's still +/- 20V.
You are of course right! :-[
Sorry for the misleading answer.
(the probe multiplier is only shown in the first picture, but it's quite certainly the same).
Nandemo wa shiranai wa yo, shitteru koto dake.
 
The following users thanked this post: carldr

Offline thinkfat

  • Supporter
  • ****
  • Posts: 2150
  • Country: de
  • This is just a hobby I spend too much time on.
    • Matthias' Hackerstübchen
In ranges < 500mV/div the offset range is only +/- 2V, but with a 10x probe, that's still +/- 20V.
You are of course right! :-[
Sorry for the misleading answer.
(the probe multiplier is only shown in the first picture, but it's quite certainly the same).

Well, we don't know that. The screenshot doesn't give it away. I do admit the waveform looks weird. It's not what you'd typically expect to see when it comes to power supply noise.

PS: the first picture shows the second channel.
Everybody likes gadgets. Until they try to make them.
 
The following users thanked this post: carldr

Offline carldrTopic starter

  • Newbie
  • Posts: 4
  • Country: gb
Thanks all!

I have a large light sat quite low above my workbench which I just thought might be causing some interference, but I've just done some testing, and it doesn't make any difference.  The probes used were both x10 in both pictures, and the probe indicated by yellow (ch1) was the same in both pictures.

Anyway, it sounds like it's mostly expectations (I was expecting a nice, 1 pixel thick line, especially at the 1v/div setting.)

I don't have another bench power supply on hand to further test that with.  I have the likes of phone chargers and batteries, are either a good way of testing to rule out a potential issue with the scope?  I just did a quick test with a generic 12v PSU brick, and that's noisy as hell.
 

Offline Karel

  • Super Contributor
  • ***
  • Posts: 2214
  • Country: 00
The DS1054Z has an 8 bit ADC, that's 256 steps. It has 400 vertical pixels in the waveform area of the screen. That means, 1 ADC LSB is about 1.5 pixels. As the screen outresolves the ADC, this is what you'd expect - clearly identifiable steps.

To be precise, the display of the DS1000Z series maps 25 lsb's to every vertical divsion, so only 200 of the max. 256 steps are visible.
When you download the data from the scope you will receive the full 8-bit resolution (256 steps).
 
The following users thanked this post: carldr

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11535
  • Country: my
  • reassessing directives...
or maybe the PSU spec is "not as precise"? its a SMPS what do you expect? https://www.manson.com.hk/product/nrp-2050/ as others said? try AC coupled, you can get down to 10mV/div and see if the noise real.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 
The following users thanked this post: tooki, carldr

Offline newbrain

  • Super Contributor
  • ***
  • Posts: 1714
  • Country: se
I was expecting a nice, 1 pixel thick line, especially at the 1v/div setting.
This is a common, if unreasonable, expectation:

Nandemo wa shiranai wa yo, shitteru koto dake.
 
The following users thanked this post: tooki, carldr

Offline carldrTopic starter

  • Newbie
  • Posts: 4
  • Country: gb
newbrain, thanks for that video, that's exactly what I needed.

I've experimented with changing the bandwidth limits and memory depth on my scope and I can mirror what the video shows, and now understand what I'm actually seeing on the screen.  I also now appreciate that it's showing 200 levels on basically a 400 pixel screen, and can see that when horizontally stretching a captured waveform.

I'll do more digging into the power supply thing later, but I'm now confident the scope is working as it should.

Thanks to everyone, your replies are appreciated.
 
The following users thanked this post: tooki

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 8108
  • Country: fi
Something is not OK, especially in the second pic.

The scope is in STOP mode. This is a classic: my guess is, the OP has zoomed in the vertical scale AFTER stopping (single-shotting).

Obviously, as the original data is just 8 bits (which, by the way, is true with most scopes even in $10000 class; o-scopes are primarily instruments to measure time quite accurately), the quantized voltage levels become very apparent after just one or two steps of vertical scale zoom-in on the stored data.

So, always re-acquire the data after changing vertical sensitivity.

8 bits is quite acceptable for most purposes if you originally set the scale (analog amplification in the front-end) just right, but 8 bits doesn't leave any room to add gain digitally, and sucks for the math operations, as well.

Whenever the signal is repetitive, using the averaging function is advisable (if average is what you want to see!),
« Last Edit: May 06, 2020, 01:29:28 pm by Siwastaja »
 
The following users thanked this post: tooki, newbrain, carldr

Offline carldrTopic starter

  • Newbie
  • Posts: 4
  • Country: gb
Something is not OK, especially in the second pic.

The scope is in STOP mode. This is a classic: my guess is, the OP has zoomed in the vertical scale AFTER stopping (single-shotting).

That's exactly what I did, and completely explains it.  Thanks for spotting that!
 
The following users thanked this post: Siwastaja, tooki

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 9886
  • Country: us
The 555 traces look fine.  The PS trace doesn't look good.  I would be certain to remove any loads before I probed the PS.

You can eliminate some of the trash with 0.1 ufd ceramic capacitors (at least 1 for every IC) and perhaps a 10 ufd electrolytic where the power comes onto the board.  Or maybe at the far end of the traces.  Or maybe both...

Don't try to imply precision with a scope trace.  All you can really count on is the general shape of the wiggly lines.
 
The following users thanked this post: tooki


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf