Oscilloscope probe ground leads act as FM attennea

[dmulligan]

This is my first DSO so I am not sure if my expectations are at fault or if it's my new oscilloscope.  I really hope it's something I am doing wrong.  I hooked each of my probes using the alligator ground clip to the compensation signal output terminals, clicked auto and I see a lot more noise on the signal than I expect.


However I get a very smooth signal when I use the ground spring instead.


I have tried all 4 channels with the same results.  The calibration signal from my DHO804 looks smooth when viewing it from my 25MHz CRO using the alligator ground clip.  However the calibration signal from my CRO looks even worse when viewed on my DHO804 with the alligator clip but clean using the ground spring.

I also tried taking the DHO804 to another room and it was no better viewing it's own calibration/compensation signal.  I am nearly certain that AC circuit would be from the opposite phase as the one at my desk, certainly a different circuit down the the breaker.

I am using the ground lead provided by Rigol through my benchtop power supply.  In the other room I loosely put it into the ground plug.  I have also tested with the scope floating also with no difference.

What might I be doing wrong?

[Martin72]

However I get a very smooth signal when I use the ground spring instead.

Complete normal behaviour, no need to worry.
The lead with the alligator clip is acting like a "antenna" receiving high frequent signals.
You will see a difference between bandwithlimit on/off.

[dmulligan]

Complete normal behaviour, no need to worry.
The lead with the alligator clip is acting like a "antenna" receiving high frequent signals.
You will see a difference between bandwithlimit on/off.

I forgot to mention that I saw a lot less when setting the bandwidth limit to 20MHz.

This is expected behaviour even with about 250mV p-p noise riding on both the high and low parts?

[Fungus]

Mine looks like this:


It could be your environment. Try it somewhere else ... eg. at a friend's house.

Also try unplugging all the little wall warts that you have near the 'scope (ie. The Phone chargers, etc.) and turn off the lights.

[dmulligan]

Mine looks like this:

Here the difference is more because of the sample rate, 312 MSa/s vs. 2.5 MSa/s.

Peter

Excellent observation.  Sadly reducing the sample rate to 2.5MSa/s made no difference what so ever.  So far only the BW limit brings down the noise when using the ground lead.  Which makes it odd that reducing the sample rate even to 500K Sa/s didn't help.  I am going to upgrade to the latest firmware tomorrow to see if that will help.

David

[ebastler]

I don't think there is cause for concern. Since the noise is low when using the ground spring, the scope itself is obviously fine. Since all probes behave the same, there is no reason to assume that you have a dud there with a bad ground clip connection.

It seems pretty clear to me that you are capturing noise from some source in the environment -- which might include the scope itself or its power supply. A few things to try:

• Use an even longer ground lead as an experiment, to see whether the noise goes up further.
  
• Move the ground lead around, especially making the loop cross section smaller or changing its orientation, to see whether that reduces the noise.
  
• To investigate whether the loop is picking up noise from the scope itself or from its power supply, try moving your probe + ground lead closer or further away from these. (No need to connect to the 1 kHz test signal, just clip the ground clip to the probe tip.)
  
• Zoom in on the noise or do an FFT, to see whether there are pronounced frequencies. Those might give you a hint as to the potential source. Again, that's probably best done without the 1 kHz signal.
  

Edit: I don't think anyone has mentioned it, maybe since most of us have gotten so used to DSOs for a while: Compared to CROs, noise tends to be visually more dominant on DSO displays, even with the "digital phosphor" intensity grading which all modern DSOs provide and which aims to mimic a CRO display. You can play with brightness and persistence settings, but will always see the overlay of many noisy sweeps more clearly than on a CRO. On the bright side, you will also see rare events more clearly! -- Also, when comparing to your 25 MHz CRO, setting the 20 MHz bandwidth limit on the DHO804 makes for a fairer comparison.

[David Hess]

Mine looks like this:

Here the difference is more because of the sample rate, 312 MSa/s vs. 2.5 MSa/s.

A change in sample rate by itself should not affect the noise.  Aliasing will cause the noise to fold over in the frequency domain and the standard deviation of the signal will be unaltered.

[2N3055]

Maybe you have one with the noisy PSU?
There were some reports about that in early specimens.
That is why Rigol changed to better quality PSU.

But mostly that is because new scope has 100MHz + BW and now you can see high frequency old scope didn't show.
Also, CRT scopes show fast signals with less brightness (pretty much invisible) if slow part of the signal is set to normal brightness.
Digital scopes show more detail, because they show faintest parts of the signal still in a way that is visible. You can see difference in brightness, but if there was a signal you will see it.

And then there is probing, once you have the scope that clearly shows more than 100 MHz, you realize that short piece of wire with a clip is antenna and inductive pickup....

[Fungus]

Mine looks like this:

Here the difference is more because of the sample rate, 312 MSa/s vs. 2.5 MSa/s.

Peter

Good point. I just pressed AUTO to make that screenshot, I didn't fiddle with the memory.

Here's mine with ground clip and the same memory settings:

[Fungus]

And then there is probing, once you have the scope that clearly shows more than 100 MHz, you realize that short piece of wire with a clip is antenna and inductive pickup....

Yep. It could easily be something in the room.

[Martin72]

I just tried this out here as well.
At first glance it looks good, but if you dissolve it further, you can see a ripple on the "roof".
If you take advantage of the good properties of the rigol and resolve more vertically, you can clearly see a ripple of about 30khz.
In my case, the ripple remains, regardless of whether groundspring or alligator clip or not connected to ground at all (the probe already has ground).
This is a disturbance that probably comes from the power supply.
But this is really nothing bad...

[ebastler]

If you take advantage of the good properties of the rigol and resolve more vertically, you can clearly see a ripple of about 30khz.
This is a disturbance that probably comes from the power supply.
But this is really nothing bad...

Thanks for giving the "zoom in to look for characteristic frequencies" suggestion a try!

While your cursor positioning for the frequency measurement is debatable (I would estimate ~40 kHz),  the switching supply clearly seems to be the source here.

[dmulligan]

Mine looks like this:

So that was an interesting exercise.  In order to match your conditions I had to upgrade to firmware 00.01.01 as pressing Auto gave me a different memory depth and sample rate.  I also saw a couple more differences in your screenshot pointing to you using new firmware.  Sadly I get no change in behaviour with new firmware. 

Do you have BW Limit enabled for that channel?  Does it change if you set your probe attenuation to 10x since you are running your probe at 10x?

[Martin72]

While your cursor positioning for the frequency measurement is debatable

Argh, new glasses please...

OK, but I don´t repeat this today, also the following not...
Following an idea, I went back to the study and did the following...
Siglent HD and the rigol switched on, first measured the siglent ref. generator with the siglent, then with the rigol.
Then measured the rigol ref. generator with the siglent.
The case is clear, the disturbance comes from the generator, which catches something...
But as I said, this is nothing bad as small as it is, just do not think that would now be a flaw.

[ebastler]

Sadly I get no change in behaviour with new firmware. 

Do you have the switching power supply right next to the scope? If so, does anything change if you move it as far away as the cable allows? (Not too likely because the noise is probably largely wire-bound, not radiated. But worth a try, I think.)

Also, which brand of power supply was shipped with your scope? It seems that Rigol shipped a no-name, fixed 12V supply with early units, but had problems with it. They transistioned to LiteOn or Lenovo supplies -- proper USB-C supplies capable of negotiating the voltage, which must cost them more, and probably are better quality and less noisy too.

[ebastler]

The case is clear, the disturbance comes from the generator, which catches something...

That's even better news! Noise on the 1 kHz reference signal really is irrelevant. As long as the scope input is unaffected and can measure real-world signals with low noise, everything is fine and dandy. 

[dmulligan]

I have a LiteOn PSU.  If the noise is coming from the PSU, wouldn't I get the same noise using the ground clip as with the alligator lead?  The PSU is on the floor and my scope is on my desk.  I will move it to yet another room soon.

I knew to expect extra noise on a DSO especially a higher BW scope, this is more than I expected.  The frequency and amplitude of the noise looks random.

Although the frequency looks to be random, perhaps it isn't.  If I am using FFT correctly, the only peak that sticks out above harmonics of the original signal is 63.0 kHz.

[David Hess]

I knew to expect extra noise on a DSO especially a higher BW scope, this is more than I expected.  The frequency and amplitude of the noise looks random.

It looks pretty normal to me for a 100 MHz bandwidth when using the ground clip and the built in calibration signal.  My 300 MHz DSO would look even noisier.  If I use a probe tip to BNC adapter to connect directly to a BNC, then the calibration signal is clean. (1)

(1) Some old Tektronix oscilloscopes used a BNC for the calibration signal, so this can be done on the oscilloscope itself but a function generator output works just as well for calibration.

[Martin72]

this is more than I expected. 

Once again, and as far as I'm concerned for the last time, you're worrying about this unnecessarily.

[dmulligan]

My #1 concern was that I don't have a dud scope.  I didn't feel any better about that having seen Fungus' and Martin72's relatively clean waveforms didn't make me feel any better. 

I tried moving the scope around to different parts of my house and my 3rd location produced much better results.  So the problem is environmental.  I am not really surprised as I am 100M from a transmission line and there is so much computer and electronics gear around my desk. 

Thank you all for helping alleviate my fears.  Now I need to figure out what I can do to use my ground alligator leads at least sometimes and, more importantly, learn how to use this little thing.

[Fungus]

Mine looks like this:

Do you have BW Limit enabled for that channel?  Does it change if you set your probe attenuation to 10x since you are running your probe at 10x?

Oh, I forgot my probe was in 1x mode. I don't have bandwidth limit enabled though.

In 10x mode I see pretty much the same thing.

[Fungus]

My #1 concern was that I don't have a dud scope.

The only thing that counts is what you see when there's no probe connected to the 'scope.

This is mine with probe connected and the probe's clip connected to the probe tip:


This is mine with no probe connected:

[CosteC]

Do you have grounding lead connected? DHO800/900 series with USB-C PSU introduced new failure mode - lack of grounding and "unnoticed loss of grounding"

What is possible far easier in DHO800/900 is to add ferrite core on PE/grounding lead which may help in some cases.

[ebastler]

I am using the ground lead provided by Rigol through my benchtop power supply.  In the other room I loosely put it into the ground plug.  I have also tested with the scope floating also with no difference.

[iMo]

..
What is possible far easier in DHO800/900 is to add ferrite core on PE/grounding lead which may help in some cases.

That will not help you in any cases, imho. The ferrite bead put on a grounding wire will introduce serial impedance, like 10-100 ohm based on frequency.. You want the opposite - to have as small as possible impedance towards the ground..

[CosteC]

..
What is possible far easier in DHO800/900 is to add ferrite core on PE/grounding lead which may help in some cases.

That will not help you in any cases, imho. The ferrite bead put on a grounding wire will introduce serial impedance, like 10-100 ohm based on frequency.. You want the opposite - to have as small as possible impedance towards the ground..

In general you are right, yet in this case environment seems very noisy, maybe PE is noisy too. Putting ferrite on cable is easy experiment. Probably it shall be on emitting device, yet it may be celling light, so somewhat difficult.

[dmulligan]

I understand now that there is nothing wrong with my DHO804, now I am trying to locate the source of the noise.  It feels inconsistent.  I tried shutting everything off in this room, which meant unplugging everything except my scope.  No significant difference on the original nor a nearby wall outlet.  An outlet in the next room seems better but an extension cord back to the original location seemed maybe a little better but hard to tell.  I clearly need to find a way to measure this better so that I can more immediately see if it's improving.  It doesn't help that I am still learning how to use this scope and haven't had much time to spend on it.

I tried FFT again, and this time I found something different.  I set the trigger point to just a little left of the screen and changed the horizontal scale so that I was just looking at the top part of the pulses.  Note: I definitely don't know how to use FFT properly, however this time I found a collection of peaks which, unless I did something very wrong, seems to be the sources of my noise, at least today.  Also I've moved the scope about 2 meters from where I had it on my desk due to something else I was working on and in this orientation/position the noise is less than half of what it was before, only ~150 mV.  Here were the results.









If I haven't made any glaring errors, my problem seems to be coming from a few locations with line of site through my window from 5 to 8 KM away.  All (most) of the FM and TV transmitters in my city are arranged along a long hill across the river valley from me, all visible from my window.  Could this really be the source of my noise or am I have I really messed up how I am using FFT?

[Martin72]

Hm, look like the full FM band 87.5Mhz-108Mhz...
Won´t believe that this is the cause, but you could try it like I did:

https://www.eevblog.com/forum/testgear/new-dho804-does-this-compensation-signal-look-noisy/msg5134284/#msg5134284

[tautech]

~90-100 MHz is FM broadcast.
Careful probing technique is required with low level signals.

[Fungus]

Do you live near a local FM radio station?

[Martin72]

This is a disturbance that probably comes from the power supply.

No:

https://www.eevblog.com/forum/testgear/rigols-new-dho800-oscilloscope-unbox-teardown/msg5139402/#msg5139402

Must come from the "inside"...

[ebastler]

Do you live near a local FM radio station?

[...] my problem seems to be coming from a few locations with line of site through my window from 5 to 8 KM away.  All (most) of the FM and TV transmitters in my city are arranged along a long hill across the river valley from me, all visible from my window.

[dmulligan]

Hm, look like the full FM band 87.5Mhz-108Mhz...
Won´t believe that this is the cause, but you could try it like I did:

https://www.eevblog.com/forum/testgear/new-dho804-does-this-compensation-signal-look-noisy/msg5134284/#msg5134284

Thank you for making me look back at the way you zoomed in.  My noise is much higher frequency than yours.  The first three captures are as similar to yours as I could make them.  I didn't set the vertical cursors until the last two captures.

[Martin72]

Modulated signal....Your scope is receiving...

[dmulligan]

~90-100 MHz is FM broadcast.
Careful probing technique is required with low level signals.

The signal in question is the 3v calibration signal where I am measuring 300-500mV of noise on it, depending on conditions I have yet to narrow down.  Not quite true, 2 meters to my left and the scope turned to a different orientation I was seeing only 150mV of similar looking noise.  I am trying to determine if the noise is something I can avoid or if I live in a noisy area and all signals require careful probing technique.

[tautech]

I can get 100's on mV doing the same under LED lighting....this is only the additional sensitivity a DSO can provide.
Trick is to know what it is and where from so to not focus on it and only on the part of the signal that matters.
20 MHz BW limit is very useful at times.

[ebastler]

As Martin said, the frequency range almost exactly matches the FM band (at least as defined in this part of the world, 87.5 to 108 MHz). And in your later post the signal does look modulated, although it looks more like AM modulation -- is that being used in this frequency range in Canada?

If you can visit a friend out of town, further away from the transmitters, it's something I would try to get a better understanding and confirm the radio interference theory.

Also -- does the noise appear specifically when you measure the calibration signal? What happens when you just ground the probe tip (clipping the probe's little ground lead to it)? What happens if you do the same, but grounding via e.g. a 100 kOhm resistor? If it should just be the 1 kHz reference output that catches the radio interference, that would be odd but would not affect your use of the scope.

[dmulligan]

I can get 100's on mV doing the same under LED lighting....this is only the additional sensitivity a DSO can provide.
Trick is to know what it is and where from so to not focus on it and only on the part of the signal that matters.
20 MHz BW limit is very useful at times.

Agreed.  I am trying to determine the sources of the noise I am seeing in my work area.  I was amused at and wanted to share the large commercial FM broadcast interference I am seeing, that or asking for advice about how I am measuring incorrectly.  Sadly I am having trouble narrowing down just what it is if it isn't the FM broadcasts.  It's not in this room and I've been waiting for a time that I can shut down the entire house and bring it back up one breaker at a time.

[Fungus]

If you can visit a friend out of town, further away from the transmitters, it's something I would try to get a better understanding and confirm the radio interference theory.

Get a battery and go for a drive...

[tautech]

I can get 100's on mV doing the same under LED lighting....this is only the additional sensitivity a DSO can provide.
Trick is to know what it is and where from so to not focus on it and only on the part of the signal that matters.
20 MHz BW limit is very useful at times.

Agreed.  I am trying to determine the sources of the noise I am seeing in my work area.  I was amused at and wanted to share the large commercial FM broadcast interference I am seeing, that or asking for advice about how I am measuring incorrectly.  Sadly I am having trouble narrowing down just what it is if it isn't the FM broadcasts.  It's not in this room and I've been waiting for a time that I can shut down the entire house and bring it back up one breaker at a time.

Your FFT clearly shows peaks at each FM station.
Engage Peak markers and compare results to published local FM frequencies.

We had a customer that was convinced his new SDS1104X-E was broken but after assurance it wasn't he eventually found a unknown wallwart behind a curtain across the room. Some of these things just spew EMI.

[dmulligan]

As Martin said, the frequency range almost exactly matches the FM band (at least as defined in this part of the world, 87.5 to 108 MHz). And in your later post the signal does look modulated, although it looks more like AM modulation -- is that being used in this frequency range in Canada?

If you can visit a friend out of town, further away from the transmitters, it's something I would try to get a better understanding and confirm the radio interference theory.

Also -- does the noise appear specifically when you measure the calibration signal? What happens when you just ground the probe tip (clipping the probe's little ground lead to it)? What happens if you do the same, but grounding via e.g. a 100 kOhm resistor? If it should just be the 1 kHz reference output that catches the radio interference, that would be odd but would not affect your use of the scope.

Yes, I said that it looked like FM band interference in the first message I posted an FFT capture.  No we don't have AM at 87.5 to 108 MHz in Canada.  We have the same FM band as the US. 

With the probe tip grounded using the aligator clip and lead, I do indeed have the same noise.  It changes drastically as I move the probe around.  I even found a place on my desk, with the cable looped a bit with only 45mV of noise.  The noise is about 75-125 mV with a 100k resistor depending on where I wave it around.  It's mostly 75-80 unless I let go of it.

If I use the ground spring to the calibration out and ground the signal is nice and clean.

[Fungus]

I tried using the worst cable I could find and it was all good.


If I touch the clip with my finger I get ripple on the signal.

[tautech]

If I use the ground spring to the calibration out and ground the signal is nice and clean.

This ^ 

Now you know how to work around the probe reference (gnd) lead making a nice RF loop, you have all you need to go forward .....but don't forget about that nice little RF loop as non-contact frequency measurement can at times be useful.

[ebastler]

I tried using the worst cable I could find and it was all good.

Any change if you also connect the black GND wire to the scope's GND clip (next to the reference output)? We want a ground loop here!

Edit: Well, normally we don't want it. But for testing...

[Fungus]

I tried using the worst cable I could find and it was all good.

Any change if you also connect the black GND wire to the scope's GND clip (next to the reference output)? We want a ground loop here!

Nope.

The GND is connected the BNCs internally so no chance of a loop. 

[ebastler]

Nope.
The GND is connected the BNCs internally so no chance of a loop. 

Well, there is a GND connection between BNC and Ref GND clip inside the scope, and one outside the scope (if you make the cable connection). That's a GND loop, right?

But when you say "Nope", you mean you tried and it changes nothing? Then that loop is not where the scope picks up noise -- or you live in an electrically quiet area.

[Fungus]

But when you say "Nope", you mean you tried and it changes nothing?

Yep. It all looks great either way.

or you live in an electrically quiet area.

Looks that way.

[dmulligan]

..
What is possible far easier in DHO800/900 is to add ferrite core on PE/grounding lead which may help in some cases.

That will not help you in any cases, imho. The ferrite bead put on a grounding wire will introduce serial impedance, like 10-100 ohm based on frequency.. You want the opposite - to have as small as possible impedance towards the ground..

In general you are right, yet in this case environment seems very noisy, maybe PE is noisy too. Putting ferrite on cable is easy experiment. Probably it shall be on emitting device, yet it may be celling light, so somewhat difficult.

I found a snap on ferrite bead on an unused cable.  27mm long and 9mm ID (has a plastic cable holder).  I tried looping the ground cable and the USB-C cable or straight through with the AC cable with no noticeable difference.  I even tried the probe ground lead for fun which made a significant difference for the worse.

Oh well. I know how to work around the noise by using the ground spring which is good enough for now.  I still hope to find out how to avoid it all together.  Maybe I'll have to move in with Fungus since he lives in a exceptionally EMI quiet area.

[CosteC]

Ferrite on ground lead - as expected
Any luck with turning OFF some stuff in room/home? LED lighting can be very noisy and LED strips are naturally long, good antennas.
Another classical issue source are variable frequency drives. Now in washing machines, fridges, air conditioners and other stuff with electric motors.
EV cars are also horrible. I would be surprised if EV chargers would be different. Many kW SMPS always emits.

[dmulligan]

Ferrite on ground lead - as expected
Any luck with turning OFF some stuff in room/home? LED lighting can be very noisy and LED strips are naturally long, good antennas.
Another classical issue source are variable frequency drives. Now in washing machines, fridges, air conditioners and other stuff with electric motors.
EV cars are also horrible. I would be surprised if EV chargers would be different. Many kW SMPS always emits.

A few days ago I reported that I turned everything off and unplugged all the things in this room with no significant change.  I am waiting for an opportunity to shut off all breakers in the house and turn them back on one at a time.  There are so many devices in my home, it could be coming from anywhere.  Is this why people wear tinfoil hats?

[Martin72]

Is this why people wear tinfoil hats?

 

Do you have a metal box ?
You could put the scope in there and then see if the irradiations become less.
But I am afraid that the probe will still be the "antenna".

[dmulligan]

Is this why people wear tinfoil hats?

 

Do you have a metal box ?
You could put the scope in there and then see if the irradiations become less.
But I am afraid that the probe will still be the "antenna".

Tinfoil hats and metal boxes are just two different styles of faraday cages.  I am thinking about wrapping a cardboard box in foil, grounding the foil then using the web interface to look at the results.

[ledtester]

[quote author=dmulligan link=topic=398355.msg5143530#msg5143530
A few days ago I reported that I turned everything off and unplugged all the things in this room with no significant change.  I am waiting for an opportunity to shut off all breakers in the house and turn them back on one at a time.  There are so many devices in my home, it could be coming from anywhere.  Is this why people wear tinfoil hats?
[/quote]

Just don't go too crazy trying to find it...

Coppola - The Conversation - ending
https://youtu.be/SEo7FGkmRx0

[dmulligan]

I finally found the opportunity to shut down the power to the rest of the house.  I didn't set the scope up in the same location due to travel time to my normal work area.  I regret not doing that as the last step.  I was only seeing 35mV of noise with my scope plugged in beside the breaker which wasn't good for testing for lower noise.  I tried a few more outlets around the basement with similar luck.  Eventually I decided to use an extension from the outlet where I saw the lowest noise from and located my the DHO804 at the top of the stairs.  I was only seeing 80-100mV of noise but at least the FFT showed the same peaks across the FM band.  Shutting down all other circuits yielded no change.

Learnings:
1) Noise source doesn't seem to be local.
2) Noise improves the lower down I get.  The other way to think of that is the noise improves the further I get the scope away from line of sight to the towers.
3) (Separate from this test) Twisting the ground leads around the probe lowers the noise, often significantly.

I would really like the convenience of both working upstairs and using ground leads at least sometimes.  I wonder what other mitigation methods I can use.  I have some more split ferrite beads of unknown value coming from Amazon today that I can try on my bench power supply cable, ground, scope AC and scope DC cables.  I've read that a grounded antistatic mat has helped with some types of probe noise, I don't know if that could help mine.  I suppose that I could simulate that with some foil or a cookie sheet and a ground wire.

Are shorter or lower inductance ground leads available?  LeCroy makes a "HF Compensated Ground Lead", I wonder if I can find them for Rigol probes or if I can find affordable probes with some.  I wonder what sort of sorcery makes them special and how effective they are.  From the probe datasheet "A passive network in the lead itself reduces the loading effect of any resonance."  At 35USD I really hope they are special and at that price I don't think I'll be buying one to test if it even fits my probes let alone helps with my problem.

[ebastler]

Are shorter or lower inductance ground leads available? 

Do the Rigol probes come with ground springs which you attach right to the front of the probe (after pulling off the hook attachment)? If so, have you tried what difference these make? They are not applicable in every situation, since a ground connection needs to be available very close to the measurement point -- but should work on the 1 kHz reference output, and will give you another data point.

EDIT: Ah, sorry -- just saw your older post a page up where you already mention that you successfully tried these.

[dmulligan]

Are shorter or lower inductance ground leads available? 

Do the Rigol probes come with ground springs which you attach right to the front of the probe (after pulling off the hook attachment)? If so, have you tried what difference these make? They are not applicable in every situation, since a ground connection needs to be available very close to the measurement point -- but should work on the 1 kHz reference output, and will give you another data point.

Yes, the problem goes away using the ground springs.  I mentioned that a few pages back.  It's less convenient but it works.  The inconvenience gets worse the more channels I need to use, especially as I go over 2.  I'm going to 3d print some probe holders but as I am clumsy I suspect they won't help as much as probe clips would.

I don't think I've mentioned before that I see the same noise on any circuit I probe as long as I am using ground leads.  Sadly my probe noise makes I2C decoding fail to work most of the time.

My options are now 1 or more of the following:
1) Use ground springs
2) 20MHz BW filter
3) Use probe at 1x

[ebastler]

I don't recall all the steps you have already taken on this journey. Did you try some (partial) shielding already? Maybe try a grounded metal plate oriented vertically, relatively close to your device under test, in the direction of the incoming radio waves? Or a grounded plate (with an insulation layer on top) as your workbench surface?

[macboy]

Based on a quick eyeball of the frequency peaks and the description of the transmitters on a hill across the river, I'd guess you live in the NCR? I do as well, but much further south and so further away from most of the transmitters. My lab is also in the basement, which helps. I have never experienced an issue with noise as bad as yours.

The FM peaks are certainly a smoking gun. There is no question why you pick them up; any conductor will pick up RF radiation to some extent. The spring is very short compared to the ground lead, so it picks up less. Being so close to multi-kW transmitters is certainly a mixed blessing; great for picking up clear radio or TV stations, but bad otherwise. Try holding the ground lead parallel to the probe rather than letting it form a big loop. You could try adding a ferrite on the scope probe lead very near the probe (not on the ground lead!!, but on the coax from probe to scope). As before, this may help or hinder, but is easy to try.

Reducing the sampling speed does not eliminate high frequency noise, as you have found. The scope does not apply a low-pass (anti-aliasing) filter to the signal, and the sample-and-hold prior to the ADC is very fast. It will capture the instantaneous voltage of the waveform at the sample time regardless of the sampling period. This means all HF noise is still there, just aliased to a different frequency.

As for other mitigations, you could try the scope's built in filters: low pass or band stop. In either case, the sampling frequency will need to be high enough that the noise at 87 - 107 MHz appears at that frequency and is not aliased to a lower frequency. So sample rate will generally need to be very high. Filters are effective with single-shot or repetitive captures, as long as you have lots of samples to spare (high sampling rate). Obviously, it will reduce HF content of the signal in addition to noise.

Since the FM radio noise is not coherent with the input signal, using averaging can help dramatically, even with a small number of samples like 4 or 8. This requires a repetitive signal of course, but when that is what you are observing, I recommend averaging at a low rate (4/8) in general. Averaging helps to enhance resolution and clean up noise so you can see the real shape of the signal under the unavoidable noise, and it does not reduce HF signal content like a filter (or eres mode) does.

Finally, if you have a quality USB-C PD charger (like from a laptop or chromebook), try powering the scope from it, to see if some of the noise is actually conducted from the power supply.