9

Following this question, I purchased a DSO-2090 USB oscilloscope.

When I power it up, I see lots of small fluctuations in the waveform, whether connected to nothing, a battery source, itself (small waveform generator on the back) or by shorting the ground clip.

Oscilloscope Noise

(See the large version)

In the screenshot above, the CH1 is connected to a small, battery powered circuit of mine. CH2 does not have the probe connected at all.

I have spoken to a colleague about this and was told that I should see a flat line so I am worried I have purchased a faulty unit.

My question is whether or not background noise like this is normal in an oscilloscope?

Edit 1

Added 1kHz waveform example as per comment:

1kHz

(See the large version)

Community
  • 1
Mr. Hedgehog
  • 2,117
  • 4
  • 24
  • 30
  • What does the trace look like when connected to the waveform generator? Assuming it is a 1 KHz square wave, you should see a very clean waveform. If not, something is wrong. Also, CH1 and CH2 should both show the same thing. – tcrosley Oct 31 '10 at 20:58
  • I see a 1KHz waveform with the noise (added above). – Mr. Hedgehog Oct 31 '10 at 21:13
  • 3
    The second trace looks normal for a scope with an 8-bit resolution. As Kortuk mentioned in his answer, the trace can normally jitter back and forth 1 step of the ADC even on a stable signal. There are 256 steps for a 8-bit ADC, so that is what is causing the line not to be perfectly straight but have little bumps. This is normal for any DSO. More expensive ones have ADC's with more bits (9, 10, 14, and 16 are most common in addition to 8), so the effect isn't as noticeable. As long as CH1 shows essentially the same thing, it appears to be working okay. – tcrosley Oct 31 '10 at 21:43

4 Answers4

9

As an indicator that the second capture you displayed looks normal, here's one I just made of a 1 KHz square wave, from my ELAB-080 combination instrument (dual channel DSO, 16-channel logic analyzer, arbitrary waveform generator (AWG), and dual channel power supply) which also has a vertical 8-bit resolution in its DSO section. (Note: at $495, the ELAB-080 was over your budget in your other question where I recommended the DSO-2090).

You can see the noise at the top and bottom of each square wave, which looks almost identical to your trace.

alt text

tcrosley
  • 47,708
  • 5
  • 97
  • 161
  • 1
    +1 from me. Gives a great reference if others see the same thing. I tried to add more about how to reduce noise, but there is always a chance it is just the systems noise. – Kortuk Oct 31 '10 at 22:24
  • Thanks for the reference. Another thing I noticed is the DSO-2090 can only enter "rolling" mode at a 1s/div which isn't very helpful when looking at fast oscillations (as the resolution just isn't available.) Is this likely to become a problem in the future? (I can still return it) – Mr. Hedgehog Oct 31 '10 at 22:42
  • @Nat, don't know what you mean by "rolling mode". – tcrosley Oct 31 '10 at 22:47
  • You are going to really have to look at what you need it for. It is hard to say one way or another. – Kortuk Oct 31 '10 at 22:47
  • @tcrosley: "Rolling" in this software is a continuous acquisition mode where it draws (and scrolls) across the screen as time passes. – Mr. Hedgehog Oct 31 '10 at 22:53
  • 2
    I don't think it is a very common feature, I looked up DSO + "rolling mode" on Google and found that it is primarily for looking at changes to slow-moving signals (e.g. see "roll mode" in www.crupower.fi/DSO.pdf ) -- so it isn't intended for fast signals. – tcrosley Oct 31 '10 at 22:58
  • @Nat, note that the Rigol DS-1052E, which I believe you were also looking at, also has an 8-bit vertical resolution, so you are going to see some of the same 1-bit noise issues: http://www.rigolna.com/products/digital-oscilloscopes/ds1000e/ds1052e/ – tcrosley Oct 31 '10 at 23:09
  • If you look at the screenshot though, the noise looks to be a bit more than 1-bit. – Mr. Hedgehog Nov 01 '10 at 09:04
  • I think that is just due to the way the software is displaying the data. I enlarged my screen capture, and most of the bumps are one line width high, except on the far right of each bottom trace which might be due to my probe compnesation being slightly off. To make the trace more visible, the software is drawing the trace thicker than one pixel. So the one-bit differences are exaggerated. – tcrosley Nov 01 '10 at 09:17
9

I have one of these scopes (a DSO-2250, so a slightly faster model).

I can confirm that there is approximately ~1-2 bits worth of noise at all input settings.

Basically, these scopes are worthless for anything but digital work, which is what I bought it for anyways.

One of the few uses I have found for the thing is to use it with a laptop, which gives you a simple battery-scope, which can be useful when you have ground-loop issues.

Connor Wolf
  • 31,938
  • 6
  • 77
  • 137
  • 3
    +1 Interesting note. I am so use to working at big stations I forget you could use a laptop and neglect the issue to some extent. – Kortuk Oct 31 '10 at 22:47
8

I would have to touch the oscope to confirm something.

This amount of noise is common. There are many sources of noise, but with a noise magnitude that small, it is probably not a concern.

Let me list off a couple things.

Sources of Noise

The loop of your ground clip creates has an inductance with everything on your board. any electrical signal nearby can induce a wave on your ground loop. This ground loop also acts as a low pass filter, albeit the frequency that it cuts off is normally in the high MHz range, not normally any issues for a 40Mhz scope. I have drawn a red line around what I am referring to in the following image. oscope probe image with red line around the ground clip. In this image there is a 6 inch ground clip

The fact that you have a ground reference in one circuit, and another ground reference in the other circuit plus your ground line connecting them can create ground loops(not to be confused with the ground clip issue I mentioned above, have to love similar terminology). This can actually create quite a problem, and is also no fun to fix. One method is to wrap the probe almost the the extreme tip with aluminum foil and ground it at the o'scope. This shields the interior ground connection and greatly decreases ground loop problems. It also looks awesome(not at all).

Testing for issues

  1. To see if it is just a 1bit bounce, try changing your resolution and watching if the noise changes in the amount it is jumping by.

  2. To check if the ground loop is your issue, try using an sharp blade and touching between the ground at the very tip of your probe to the ground on your board. This gives more than an order of magnitude drop in inductance.

  3. To check for ground loops, you get to wrap your probe in foil. Do not enjoy this too much.

  4. If your noise disappears just from connecting to the waveform generator on your scope then your circuit is probably the cause.
Kortuk
  • 13,362
  • 8
  • 60
  • 85
  • As a side note, I clearly left out sources of noise that come from the circuit itself. There is not enough time in the day to cover all of that in one post someone will read. – Kortuk Oct 31 '10 at 22:31
5

I don't have a PC scope but a real bench oscilloscope.

When running normally I see around 500µV noise on all settings about 5mV/div, and about 1mV noise on the 5mV/div setting. The noise is uniform across all channels.

This noise is not significant for any of my uses so far.

It's also not unusual for the scope to read ±1mV dc offset when there is zero input; some scopes have this calibrated out but expect a slight drift over time. Mine's quite old (manufactured '93), so I get around -3mV offset.

Thomas O
  • 31,546
  • 57
  • 182
  • 320