Spikes and noise on Rigol DHO814

[eTobey]

Hi there,

i bought a Rigol DHO814, which is my first > 250 bucks Oscilloscope. While trying it out on my prototype, i discovered spikes (and noise?) coming from the power supply. It has a frequency of 54-70 Khz. Connecting earth does help, if its not to far away.

I connected a coil on the leads and held it to the cable. Interestingly, the amplitude changed with position on the cable.

I found out, that i have to connect all GND, in order for the spikes to be lower than with the yellow and blue curve on the picture. But it was still about 22mV. With a little bit of aluminium and ferrite, i could lower this to just under 10mV. One turn in the ferrite does a little, two turns does better, but three turns makes it worse.
Now i wonder if i have a bad power supply?

Is there also something i can do / could try, to clean up the signal?

[ebastler]

What probes were connected to the scope inputs when you captured those screenshots, and what were the probes connected to on the other end? (Including ground connections of rhe probes, if any?)

[eTobey]

I had 2 probes connected, but only one GND of them. Another probe was laying unconnected.

[ebastler]

Does the noise go away when you disconnect the probes from the scope , i.e. is it picked up by them rather than within the scope? Does it help if you move the power supply far away from the probes and your prototype under test?

[eTobey]

The noise goes away from the probe that is beeing disconnected, but it then adds up on the other channel!

The noise does not change, if i put the powersupply with the foil near the probe, but if the blue probe has a disconnected GND, then the noise gets from about 25mv to about 20mV, when the powersupply is near... wtf?

[ebastler]

Assuming that you use the little ground strap on the probe (rather than the short ground spring), the probe & strap form a loop which can pick up external fields. Depending on how your prototype is built up and wired, it may enlarge that loop cross section.

It is good practice to move the power supply as far away from the scope and the device under test as possible, and to definitely ground the scope (both for EMC and safety reasons). If you do that and start from "no probes connected at all" to "probe connected and grounded, with minimal loop area" to "probe connected to the device under test, grounded via ground spring", what does the noise look like? Still unacceptably high? 

[eTobey]

I now measured 13mVpp with GND of the probe connected wthout ground spring. With additionally Earth closer to the probing point, it goes down to 4mVpp. With groundspring i got 0.9mVpp.

Connecting earth pretty much elimates the spikes from the power supply, without really affecting the lower noise. But removing the ferrite makes it worse again (the spikes).

[ebastler]

So, with the scope grounded and proper probing technique the power supply noise is negligible? That sounds good -- although it is a bit strange that the additional ferrite is required. Maybe some other DHO owner can chime in here with comparison data?

If you want to test the intrisic noise of the scope, and check whether any PSU noise couples into the scope directly via the power line, you can disconnect all probes and shorten the inputs, or connect a 50 Ohm terminator right to the input BNC.

[jim_griff]

Hi there. My DHO804 arrived today. I found the same issue with the switching USB-C PD PSU supplied with the Rigol. Haven't tried any other supplies as I don't own any, but will be figuring out a way of dealing with it. It shouldn't be this way.

I use 20MHz BW limit (doing audio work), but pics are my measurements *NOTE: noise is only slightly higher without 20MHz BW limit.*

All tests done with max vertical magnification (500µV/div, 1ms/div, 10M memory depth, 20MHz BW limit). Persistence time: Infinite. Probe set to 1x and properly compensated.

Pic 1: Probe connected to scope; has "clamp" cover on it. [~4000µVp-p]
Pic 2: Probe connected to scope; no clamp cover on it, no ground spring.  [~900µVp-p]
Pic 3: Probe connected to scope; has ground spring connected.  [~850µVp-p]
Pic 4: Probe connected to scope; has ground spring shorted to probe pin.  [~250µVp-p]
**Test 4 is lower in noise than with an unterminated (floating) BNC with no probe connected, which makes sense.

All tests were done with the probe sitting in exactly the same position. Probe technique will count for a lot, but the switching noise is still visible in some instances. No other hardware in the vicinity (switching supplies, computer, TV, monitor, speakers, studio equipment, psu) is causing any noticeable noise. I have a fair bit of equipment. ONLY the supplied Rigol PSU is imparting visible spikes/ringing artifacts.

I suspect the psu is built down to a price point and will be finding a linear USB-C PD supply for future use.

[Fungus]

Try switching off the lights and unplug all the wall-warts in the area:

I suspect the psu is built down to a price point and will be finding a linear USB-C PD supply for future use.

Not unless it's defective. The majority of people don't get this noise.

[Fungus]

It is good practice to move the power supply as far away from the scope and the device under test as possible, and to definitely ground the scope (both for EMC and safety reasons).

Ground won't make any difference to this.

It's usually something external.

[eTobey]

Ground won't make any difference to this.

Wrong! It made it better. Altough it wasnt huge.

I have already proven, that the problem is the powersupply, so turning anything else would make no difference. 

[Fungus]

I have already proven, that the problem is the powersupply, so turning anything else would make no difference. 

Where did you prove that?

[eTobey]

Just in my first post! I found the same frequency on the cable, and doing measures for this, got rid of that spikes quite a bit!

Did you actually read what i wrote, and looked at the pictures???

[eTobey]

Hi there. My DHO804 arrived today. I found the same issue with the switching USB-C PD PSU supplied with the Rigol. Haven't tried any other supplies as I don't own any, but will be figuring out a way of dealing with it. It shouldn't be this way.

I only just watched your pictures, but there are no spikes like mine. You should scale it like i had, to see if this is really the same problem. You could also measure on the cable with a coil.

[Fungus]

Just in my first post! I found the same frequency on the cable, and doing measures for this, got rid of that spikes quite a bit!

Did you actually read what i wrote, and looked at the pictures???

That doesn't prove anything.

It can be coming through the power supply from the mains.

[eTobey]

I happily would try to investigate this, if you tell me how i could track the sourc of it.

[ebastler]

I happily would try to investigate this, if you tell me how i could track the sourc of it.

You could try and block interference on the primary (mains) side, before the power supply: Use a plug-in mains filter on the 230V side, or put ferrite cores on the 230V input cable instead of the 15V output. Or move the scope to a different mains phase (different room), or better yet to a friend's house, for comparative testing.

[eTobey]

Shouldnt the powersupply filter this all out and not generate it either?

[ebastler]

Shouldnt the powersupply filter this all out and not generate it either?

It's not black and white...

All switching power supplies generate some level of switching noise on their outputs. (There is a reason why good laboratory power supplies, which you may want to use e.g. for sensitive analog prototype circuits, still use big fat transformers running at 50 or 60 Hz and linear regulators, rather than the much cheaper switchers.) The output of a switching supply does not have to be perfectly clean if the device that you power from it does some filtering of its own.

The power supply may or may not have some internal input filters. In any case it will suppress most of the incoming noise by charging a large capacitor, then chopping the voltage up again for transfomation. But it does not suppress 100% of the incoming noise: If a particularly nasty source of voltage spikes is operating on your mains, some of that may come through.

(A motor with bad interference suppression is a classical source of powerline interference. Very cheap switching supplies are a more modern, but rather common source. E.g. I originally used a cheap no-name 12V supply to power the LED light strips on my bench, and that created nasty interference spikes everywhere...)

Finding the source of interference signals, and the path how they couple into your measurements, is not an easy task. Small changes can have a big impact, and there may be influences which you are unaware of and/or which are outside of your control -- e.g. some device which only runs part of the time and which causes interference. Be diligent and systematic, and take good notes: short description what you did and changed, note what was connected to the scope, photos of shielding or probe arrangements, screenshots of scope data. You will track it down eventually!

[Fungus]

I happily would try to investigate this, if you tell me how i could track the sourc of it.

As noted in the video I posted. Try switching any LED lights off, unplug everything around it.

[eTobey]

I have done this. Im thinking about talking it out ona field with my camper ;-). It has a 220V system.

[Fungus]

Try it at a friend's house, or at work...

[jim_griff]

Pic 1: Probe sat by scope with clamp connected.
Pic 2: Probe sat by scope with clamp disconnected.
Pic 3: Probe hanging next to the Rigol PSU brick.

The noise is definitely coming from the Rigol PSU brick. The waveform shape is different, and the lower frequency ringing is likely due to parasitics in the power cable going to the scope.

[eTobey]

It would really help, if you would scale it, to see its frequency. Without this, it could be another source too, as stated by the others here.