Agilent Technical Forums : DSOX3000 FFT Spurs at 125 MHz harmonics ...

Guest

DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 17, 2012 7:46 PM

3000x , fft , spur

Reply

	fft_spur2.txt (382 bytes)
	fft_spur2.png (24.2 KB)
	fft_spur3.txt (382 bytes)
	fft_spur3.png (24.4 KB)

DSOX3024A I see spurs at 125MHz, 250, 500 etc on the FFT function. This occurs on all 4 channels, scope input impedance set at either 50 ohms or 1 Meg. Input BNC open or terminated either with a scope probe or 50 ohm terminator. It is about 20 dB above the noise level. It is also present at various span and center frequency settings. It does not appear to be external interference. Has anyone else noticed this?

I have added the screen shots and the scope setup data.

Edited by: KG4ARN on Apr 18, 2012 4:27 AM

Edited by: KG4ARN on Apr 18, 2012 4:33 AM

Guest

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 18, 2012 4:26 AM

in response to: Guest

Reply

	fft_spur2.txt (382 bytes)
	fft_spur2.png (24.2 KB)

Here is the screen shot and the scope setup data.

algoss

Posts: 370
Registered: 11/03/06

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 18, 2012 6:49 AM

in response to: Guest

Reply

These are just spurs from the samplers. The inherent (small amount of) jitter in the sampling clock means that there are small harmonics added to the captured data. When you look at it on an FFT they appear to pop out. You will see this on any scope.

Based on what I see, they are very small. In the spur2 picture, for example, they appear to be at ~-60dBm, which is ~225uV rms. At 5mV/division, this would be invisible. I can't tell from your picture or setup what the vertical setting of the scope is, but it must be pretty small.

Al

Guest

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 18, 2012 10:37 AM

in response to: algoss

Reply

Thanks for the reply Al. The vertical setting on Channel 2 is 5.00V/ The spurs are not small they are 20dB above the noise level. Their amplitude above the noise appears to be unaffected by the channel vertical attenuator setting.

Later this evening (Eastern Time), I can post some better tracings showing the spurs and how they appear with a signal on the scope.

I do not see this on my other digital scope (TEK TDS2022B) FFT function. I would like to know if you are seeing this on your scope or if anyone with an X3000 series can reproduce this or this is something peculiar to my scope.

Ed

algoss

Posts: 370
Registered: 11/03/06

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 18, 2012 11:12 AM

in response to: Guest

Reply

Ed,

I'm rolling out the door right now, so I don't have time for a full answer, but:

I'm not seeing those spurs on my MSO-X 3054A. I see some very small ones, but they come and go
I'm running V 2.00 firmware, what are you running?
Please turn the channel on, so the settings can be seen, and add 2 measurements, Vpk-pk, and Vrms(AC)

Al

Guest

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 18, 2012 5:12 PM

in response to: algoss

Reply

	sin_19_5.png (31.9 KB)
	sin_19_4.png (26.0 KB)
	sin_19_3.png (31.5 KB)
	sin_19_2.png (39.0 KB)
	sin_19_1.png (32.3 KB)

Here are the screen shots with and without input signals from the WaveGen. 19 MHz Sin wave was used.
The firmware is 2.10 (Shipped with the scope)

The spur at 125 MHz is fairly steady in amplitude, the 250 MHz varies a lot, and 500 MHz is in between with regard to variation.
The spurs do not seem to be affected by the channel attenuation or the level of the input signal.
In the image sin_19_4 the WavGen output is turned off.

algoss

Posts: 370
Registered: 11/03/06

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 19, 2012 6:06 AM

in response to: Guest

Reply

Please go back to the settings you had in the original post, with no inputs connected to anything. I'd like to see the noise level on the unconnected channel, which was the original question.

Turn on the channel you are interested in. Set the vertical appropriately. Add the FFT. Add Vpk-pk and VRms on the channel. Make sure they show actual values.

While you're at it, add some markers to the FFT. Set Y1 to the approximate peak (I know that it varies) of the 125 MHz spike. Set Y2 to the approximate peak of the 375MHz spike.

BTW, which model of 3000 is this?

Al

Guest

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 19, 2012 6:16 PM

in response to: algoss

Reply

I asked Agilent and they said that you are correct Al. They are expected interleave spurs from the AD converters (as you said they were) and they are about 50 dB down from the top of the vertical range.

(edited 24APR2012) I originally used the word "specification" which is incorrect. Agilent states that there are no (warranted or typical) specifications for this phenomenon.

I asked the engineer for permission to post his response to me, but he is not out of the office until Tuesday. I'll post it when I get permission. Thanks again for your help with this. Ed

Edited by: KG4ARN on Apr 24, 2012 5:35 AM

Willco788

Posts: 99
Registered: 05/28/11

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 20, 2012 2:47 PM

in response to: Guest

Reply

I'm seeing the same thing on my 3024 but when I drop the vertical per/div to 2mv/ the harmonics are gone.

algoss

Posts: 370
Registered: 11/03/06

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 20, 2012 4:28 PM

in response to: Willco788

Reply

So, it's all figured out. The spurs are caused by sampler "jitter", an inherent issue in digital scopes.

The better (higher price, higher performance) the scope, the smaller the spurs
It is possible to 'mask' this with a little math. There are scopes that don't show them at all. Some of them just use some notch filtering in the FFT to get rid of them. I saw one scope that didn't show them on screen, but if you extracted the data and ran your own FFT, they 'magically' reappeared.
The issue is not how much they sit above the noise floor, but more importantly how much below the signal they are. The concept of SFDR (Spurious Free Dynamic Range) comes into play here.
The level above the noise floor will vary, based on attenuation. As Willco788 saw, they may 'disappear' at low attenuation levels because the noise floor comes up enough to hide them.
All of those spikes, and the noise in between, get integrated together and show up as RMS noise in the time domain. A given scope could have less total noise, but show some real spikes in the frequency. Scope designers try to avoid that, but it happens.

Have a great weekend!
Al

Guest

Re: DSOX3000 FFT Spurs at 125 MHz harmonics
Posted: Apr 24, 2012 5:53 AM

in response to: Guest

Reply

	image003.png (33.7 KB)
	image002.png (36.0 KB)

Here is the response from the Agilent Engineer. Posted here with permission.

Please also note that I have edited my previous post to remove the word "specification" as this phenomenon is not an Agilent warranted or typical specified parameter.

Ed

*********************************************************************************************************************************************************************************

Hello Ed,

Please allow me to introduce myself. I’m an applications engineer with Agilent Technologies. I support oscilloscopes and other digital solutions. I received your email.

First of all, thanks for contacting us directly.

In fact, this is normal. This is called “interleave spur.” In general, “fast” ADCs are built up of slower ADCs by interleaving them together. The benefit is faster ADCs, of course. The down side is that you end up with these spurs. Other technologies do exist, but the max sampling rate is currently lower, and the cost is higher.

I think you have seen that the peak level on the spurs (and FFT noise floor) change with the vertical setting on the channels. This is also normal. For each setting, however, what you will find is that there is about 50 dB of Spurious Free Dynamic Range (SFDR). To be clear, if you put in a signal that takes up most of the screen for a given vertical setting, the tallest spur will be about 50 dB down. The rest for the FFT noise floor will be even lower (~10 dB). Here are two screenshots that illustrate this:

Both screenshot same FFT settings and time scale; the markers are tracking the 1 MHz input and the first spur at 125 MHz. we want to look at the delta-Y results.

Input signal: 30 mV p-p sine wave at 1 MHz; vertical scale set to 5 mV/div (image002)

Input signal: 10V p-p sine wave at 1 MHz; vertical scale set to 2 V/div (image003)

To get more SFDR, a fundamentally different instrument is needed, such as a swept tuned spectrum analyzer. A oscilloscope, On the other hand, is a general purpose instrument… a jack of all (many) trades, master of some (timing measurements).

Hope that helps! Please let me know if you have further questions.

Regards,

Kevin Smith
Applications Engineer - Ask me about USB 3.0 compliance!
Test and Measurement Contact Center

Thread: DSOX3000 FFT Spurs at 125 MHz harmonics