Oscilloscope trace comparison

[ryansoundlab]

I just acquired a Tek MSO22 and did a comparison of screen trace quality with my two other main scopes: Keysight MSO3012T and a Rigol HDO4204  (yes, I'm a scope addict).  Clear traces are important to my audio work.  The scopes were set to equivalent conditions (BW limit=20MHz, same vertical and horizontal scales, 10:1 probes, no averaging, etc) and simultaneously fed the same sine and square wave signals on channels 1 and 2.  The photos below show the screens after running a single shot trace.  (Photos intentionally large to show detail.)
[ Attachment Invalid Or Does Not Exist ]




No question that I prefer the Tek trace as super clean and stable.  Keysight is close behind while the Rigol is almost unusable because the noise on the trace causes erratic triggering.  The Tek is 70MHz, Keysight 100MHz, and the Rigol 200MHz, but I thought BW limiting would "equalize" that aspect of it.  Also note that the Tek and Keysight are inherently 8-bit machines while the Rigol is 12 bit.  Could be a factor.  What other settings could explain the significant difference with the Rigol?

[2N3055]

For comparison here is Siglent SDS2504xHD  with full 500MHz BW.
And SDS2504xHD  is also acoustically so silent people think it does not have fan.
Since you already own a Siglent AWG, SDS2000xHD should have been your choice, not that Rigol.

Anyway, beside that, what you show is very bad. I would still check probes and grounding. It should not be that bad...

[2N3055]

And with 20MHz BW limit. Trace is pretty much a single pixel wide, visibly better than all your scopes..

[KungFuJosh]

With the budget of the HDO4204 he should get the SDS3000X HD.

[Aldo22]

What other settings could explain the significant difference with the Rigol?

Is that 600 mVpp at 1 kHz?
I only have a really cheap scope, but it doesn't look that bad on mine.
I have to turn off BW, turn on Peak Detect, and scale the signal down to about 100 mVpp (50 mV/div / x10) just to see that much noise.
Something's wrong with your setup.

[2N3055]

With the budget of the HDO4204 he should get the SDS3000X HD.

Sure. My point is that even SDS800xHD is better than any of the scopes shown.
Still I think something is wrong here.
While I think SDS3000xHD is superior scope to HDO4204 in quite a few parameters, HDO4204 should not have that much noise.

[nctnico]

I just acquired a Tek MSO22 and did a comparison of screen trace quality with my two other main scopes: Keysight MSO3012T and a Rigol HDO4204  (yes, I'm a scope addict).  Clear traces are important to my audio work.  The scopes were set to equivalent conditions (BW limit=20MHz, same vertical and horizontal scales, 10:1 probes, no averaging, etc) and simultaneously fed the same sine and square wave signals on channels 1 and 2.  The photos below show the screens after running a single shot trace.  (Photos intentionally large to show detail.)

No question that I prefer the Tek trace as super clean and stable.  Keysight is close behind while the Rigol is almost unusable because the noise on the trace causes erratic triggering.  The Tek is 70MHz, Keysight 100MHz, and the Rigol 200MHz, but I thought BW limiting would "equalize" that aspect of it.  Also note that the Tek and Keysight are inherently 8-bit machines while the Rigol is 12 bit.  Could be a factor.  What other settings could explain the significant difference with the Rigol?

Noise from the ADC and input circuitry is likely the culprit on the Rigol. In my experience digital Tektronix scopes are quite good where it comes to having low noise and thus clean & crisp traces.

[Martin72]

What other settings could explain the significant difference with the Rigol?

Hi,

To be able to compare anything, the conditions must be as similar as possible.
That means the same sample rate, memory, and above all:
The same probes.
I wouldn't connect all the scopes to a single signal source at the same time, and certainly not the way they're set up in the picture.
To rule out the possibility that there's something wrong with the Rigol, I would feed the signal from the generator directly using a good BNC cable.
What you're seeing right now is far too noisy, even by Rigol standards.

[KungFuJosh]

What you're seeing right now is far too noisy, even by Rigol standards.

 

[Aldo22]

Just for comparison. The DSO2000 is one of the cheapest benchtop oscilloscopes, and some people complain that it’s noisy.
It’s connected to the internal signal generator via PP-150 probes (10x).
I can’t imagine that the Rigol is noisier than the Hantek.
Something’s not right here.

[ryansoundlab]

Thanks for all the advice.  I'll give the Rigol setup another shot with a direct BNC cable and try other channels, too, just in case it's in the first ADC.  I think I went for the Rigol because they were having a good sale!  Next time, I'll do the Siglent.  If I can do better, I'll post tomorrow.

[nctnico]

What other settings could explain the significant difference with the Rigol?

Hi,

To be able to compare anything, the conditions must be as similar as possible.
That means the same sample rate, memory, and above all:
The same probes.
I wouldn't connect all the scopes to a single signal source at the same time, and certainly not the way they're set up in the picture.
To rule out the possibility that there's something wrong with the Rigol, I would feed the signal from the generator directly using a good BNC cable.
What you're seeing right now is far too noisy, even by Rigol standards.

Even if it turns out the Rigol shows less noise when connecting it differently, there is something going on with ground loops and injected noise somewhere. This is something to keep in mind as it could point towards potential problems when the Rigol scope is part of a bigger test setup.

[Martin72]

there is something going on with ground loops and injected noise somewhere.

The setup shown in the picture does suggest that, yes.
In any case, this is not how you should handle things later on in a “real” measurement situation.

Edit:

I tested an HDO4024 back then (2023), and the signals were never that noisy.

[KungFuJosh]

Thanks for all the advice.  I'll give the Rigol setup another shot with a direct BNC cable and try other channels, too, just in case it's in the first ADC.  I think I went for the Rigol because they were having a good sale!  Next time, I'll do the Siglent.  If I can do better, I'll post tomorrow.

If it's still within the return period, it's something to think about.

[Picuino]

[2N3055]

That is not it.. That is something different..

[Aldo22]

That is not it.. That is something different..

The video demonstrates which settings result in more or less noise and why this isn't a problem.
When I set up my scope to see as much noise as possible:
- BW Limit off
- Peak Detect
- 8Mpts / 250MSa/S
- Probe (10x) instead of direct connection

Then the traces become thicker and a bit noisier. At 600 mVpp, however, this isn’t dramatic yet.
But yes, with the Rigol, there must be more issues with the setup.

[PlainName]

Are we not all assuming that the Tek is the standard here, and the others are worse since they show a noiser signal? But what if the Tek was cheating by smoothing things to make it look good, and the others are showing actual noise? We know some Chinese vendors do that kind off thing (cf. weighing scales) and just imagine Tek wouldn't dream of it...

Edit: typo 

[nctnico]

Are we not all assuming that the Tek is the standard here, and the others are worse since they should a noiser signal? But what if the Tek was cheating by smoothing things to make it look good, and the others are showing actual noise?

No, that would mean some form of bandwidth filtering which isn't happening for sure. Tektronix has decades of experience making good ADCs and analog frontends for their scopes.

[Martin72]

I was messing around with it a bit earlier...
First, a 1 kHz sine wave directly via BNC, then through a 1:10 probe from Testec.
I tried various settings.
Scope:Siglent SDS3104X HD.

[cncjerry]

I dont understand why the various sellers cant get the Rigol model number correct.  Do a search on DHO4204 and HDO4204 and unless there are two models offered, I find the same scope with the label above the screen printed as HDO4204.

It would be interesting to swap the probes in the test or just send the HDO4204 back as there is something wrong with it or the probe. The triggering or channel alignment looks off as well. Even a Rigol 984 wide open has nowhere near that much noise with probes just lying haphazard on my desk.  I think I have 13 scopes and I remember having a problem with those cheap probes with my first cheap scope (Owon because of the battery) where the wire breaks inside from bending when using the clamp supporting the back of the probe in the palm of you hand.  This bends the wire and it broke almost immediately.  Owon sent me 4 more probes.

I also never understood why Keysight needs 6x 'i' to spell InfiniiVision.

[KungFuJosh]

I dont understand why the various sellers cant get the Rigol model number correct.  Do a search on DHO4204 and HDO4204 and unless there are two models offered, I find the same scope with the label above the screen printed as HDO4204.

They are the same. They had to change the model number to avoid a lawsuit.

[ryansoundlab]

OK, so I did more testing on the Rigol.  Hooked up only one connection from signal generator to the scope to avoid ground interaction.  First three images were done with the same signal level as previously (200mv rms):
(1) Rigol switchable probe set for x10  (this image compares as the same with the previously posted image for the same condition - reference check)
(2) Tek probe (x10 fixed)
(3) "Junkbox" switchable probe set for x10

Following five images were done at 20mv rms to allow for x1 attenuation:
(4) Direct cable from the signal generator to the Rigol input
(5) Signal through Rigol probe set for x1
(6) signal through "junkbox" probe set for x1
(7) signal through Rigol probe set for x1 shown on Tek scope
( Direct cable from the signal generator on Tek scope

My impressions from images (1) - (3) (x10 measurements):
Rigol probe is inferior to even the junkbox probe.  Both are inferior to Tek.

From images (4) - (6) (x1 measurements on the Rigol scope):
The Rigol scope itself can produce a very good trace with the direct input (signal generator at 600 ohm output setting).  So ADC's must be OK?
Things deteriorate fast with the Rigol probe set for x1.  Lots of noise.
Even the Junkbox probe is better at x1 but still not great noise pickup.

From images (7) - ( (x1 measurement on the Tek scope):
For comparison to the Tek scope,  with the Rigol probe set for x1, compare (7) with (5) above. Did not have a x1 Tek probe to compare with, unfortunately, but
definitely a deteriorated signal compared to straight wire which looks absolutely beautiful on the Tek ( and about the same on the Rigol (4), too! 

So it seems Rigol probes are definitely inferior.  But I'm surprised at the noise picked up by the probes in the x1 setting.  Could be my LED lights.  I know my lab also suffers from power line hum pickup.  Just the limitations of my space.

(1) Switchable Rigol probe on x10:


(2) Tek probe x10 fixed:


(3) Switchable junkbox probe on x10:


(4) Direct cable to scope - x1:


(5) Switchable Rigol probe on x1:


(6) Switchable junkbox probe on x1:


(7) Tek scope with switchable Rigol probe on x1:


( Tek scope with direct cable input x1:

[Martin72]

Throw those Rigol probes in the trash and get yourself some decent ones...

[le_yum]

for science, also try switching off all instruments not involved, and lighting too (specially if led)