I recently purchased a Rigol DS1054Z, I have been very pleased with it so far and it has helped me diagnose problems on numerous occasions already.
However I was recently working on a project that had a potentiometer between Arduino 5V and GND and the center pin went straight to A0. Unfortunately the input was rather noisy and our ADC value would jump around quite a bit (not huge jumps but enough to make it not work well for our application since we were converting the potentiomer reading to an angle which controlled a motor. We tried capacitors (small and large, ceramic and electrolytic) but that didn't seem to help much.
I think in hind sight I should have used a rotary encoder and kept the angle digital only that way our motor didn't jerk back and forth at certain positions. But it is too late for that now.
So I decided I would take the potentiometer and wire it up to my oscilloscope and figure out how noisy it was. I found when I turned on AC coupling there was quite a bit of noise (like 50-100mV peak to peak at times). This certain explained why our ADC reading was jumping around.
But it occurred to me that maybe I was measuring the noise of the o-scope itself (probe etc.) and not just the potentiometer.
So I took ground clip and attached it straight to the probe and I was getting about 20mV peak to peak which is better but still not great.
I then attached probe gnd clip to Arduino ground and probe tip to ground post on potentiomer. I was back up to my 70mV p-p
So my question is am I doing something wrong in my measurement techniques or have I just reached the limits of my scope? Datasheet says it goes down to 1mV per division which seems useless if idle noise if up around 20mV p-p...
And if the 70mV peak to peak noise exists when going from gnd to gnd them if I connect gnd to a0 and read about the same is there even any noise on the analog output or am I just seeing noise from scope probe etc.?
It sounds like you've done a fair bit of the process of elimination yourself.
Out of interest, when you attached the scope probe to GND, you wouldnt happen to have noticed the frequency of the noise you where picking up on the scope would you?
Also what type of cables are you using on the scope? Shielded? standard coax? High/low impedance?. I can't say a lot personally about the Rigol scope as I've mainly only used Agilent tech scopes though the ones we have in our labs are of a similar calibre to the one youve bought, we get very little noise picked up typically, i dont think you'd be too picky complaining about 20mv at all.
The scope aside, i can't think straight away what there could be in the circuit itself picking up that level of noise, when you say you'd tried capacitors, how exactly do you mean?
Regards
Sean
wes000000:
So my question is am I doing something wrong in my measurement techniques or have I just reached the limits of my scope? Datasheet says it goes down to 1mV per division which seems useless if idle noise if up around 20mV p-p...
It is a noise floor, from 0 to 20 mV.
So anything over >~20mv can be measured with 1mv resolution.
Couldn't see the noise figures on the Rigol site.
Do a google, there will be so many threads about the bargain of the year scope!
Out of interest, when you attached the scope probe to GND, you wouldnt happen to have noticed the frequency of the noise you where picking up on the scope would you?
Also what type of cables are you using on the scope? Shielded? standard coax? High/low impedance?
when you say you'd tried capacitors, how exactly do you mean?
The frequency of the noise was too high to be anything measurable it was just all over the place. I took time/div right down to its lowest time base and it was still just a block of noise.
I am using the standard probes that came with the unit, probably not the highest quality but standard probes with BNC connector and probe and ground clip on the end.
I tried putting .1uF and 100uF capacitors across both power and ground as well as between analog 0 and ground to eliminate any noise, but it had no effect or at least it didn't seem to.
It is a noise floor, from 0 to 20 mV.
So anything over >~20mv can be measured with 1mv resolution.
Ya that does make sense then, and 20mV seems a pretty high noise floor but yes it is a very much so entry level scope so maybe that is just where the noise floor is. If so, I still can't complain cause I have loved the scope so far. I Googled as well but couldn't find anything other than 'low noise floor' which is vague and non-descript but oh well.
i only asked about the frequency as a few of our scopes have been known to pick up mains noise far more than most scopes do, usually when coupled with the crappiest length of coax available.
I may not have quite visualized it right but i can't imagine the caps doing a lot as it sounds as though from where you've put them they would simply act as dc blockers as apposed to noise filters given the erratic behavior of the noise it'd be unlikely the ones you tried would provide a suitable reactance, anyway, that aside, i'm going to agree with LarryD on this one, i apologize, i was more hoping for an error elsewhere in the system, something a little easier to get at.
Cheap digital scopes are great, but the input amps are nothing to write home about.
Mine has a noise floor (with probe set to x1, note) of around 200uV pk-to-pk, plus a
set of spikes at 125kHz or so of amplitude 1mV from the 'scopes LCD screen driver
(as far as I can make out, bringing the probe near the screen makes it worse.
Be aware too that any analog high bandwidth amp (such as a scope input amp)
is going to amplify digital noise from processors, including its own, unless careful
screening is used.
Also note that a high impedance scope input has 1M resistor to generate voltage
noise as part of its design, and without limiting the bandwidth perhaps 100MHz
of effective bandwidth, and voltage noise is proportional to the square root of
resistance x bandwidth... Worth playing with input filter bandwidth and averaging
to reduce noise that's faster than the screen can display.
If you are trying to measure ANYTHING connected to a PC you are trying to deal with common mode noise.
The first and easiest method to reduce this noise is to use a 1:1 isolation transformer in the line from scope to mains supply. Try to buy one with a Faraday shield between primary and secondary windings.
Large clip-on bead type chokes are good for this type of noise..
For the cleanest measurements you will need to use batteries to power your device without any connection to your PC or anything else that has the power line in common.
The real issue is that the ground (earth) is a wire and for RF noise or broad spectrum noise a wire is an inductive reactance that will always have some RF noise on it as it is developed across the ground (earth) wire.
You should be able to get 10 DBM (0 DBM is 1 mW/50 ohms) or 1/10th reduction in noise with the isolation transformer.
I've had good results with both the transformer and one or more of the 25 mm clip-on beads..
The isolation transformer is also a good idea if you work on transformerless devices, like Mains driven switchers for LED power supplies.
Nearby AM and MW/FM radio stations are also a big issue, in that the noise is picked up by any wiring of any length at all.
Older HP and Tek analog scopes had a bandwidth reduction switch that was generally helpful when I lived/worked near a local AM radio transmitter.
Unfortunately most if not all measurement situations are different enough for the Common Mode noise to be an issue of one kind or another and it's source/reduction is always a challenge.
A 100W 220/120 and typically 120/120V isolation isolation transformers can be had from Amazon for $19.95 or so.. I use one on my soldering iron as a matter of choice and safety too.
FWIW I've measured 200 + mV of noise between two outlets 2 meters apart. on my last test bench..
In all honesty, I had my scope on one outlet, my PC and both a synthesized sig gen as well as a spectrum analyzer on the other.
Battery powered scopes operating from batteries only (no mains connection) were/are the lowest noise measurement devices available..
However the battery powered devices had their own shortcomings in bandwidth sweep speeds and input gain issues... Then as now cheap test equipment gives poor results.
You might also try common mode chokes, Typically two identical winding's on a common core, usually a ferrite material. They are commonly found in AC/mains noise filter/power switch assemblies.
