Rohde & Schwarz SML Series Signal Generators - SML01 SML02 SML03 SMV03

[KungFuJosh]

I think it was the cable after all. I think there might be an issue with the shielding on the cable I was using.

I made a shorter hyperflex 5 cable, and the voltage fluctuation is gone, and the consistency of the color grading is much more stable.

[kmoonwalker]

 

the higher the freq -> more problems with everything which was supposedly good - now you have a scope that you can see literally everything instantly is something is wrong/off - congrats

[Martin72]

How's this look? Did I set it up correctly?

SML03:



SDG2122 for comparison:



Thanks,
Josh

I recreated this earlier.
Source SDG6052X, Scope SDS3104X HD, Magnova BMO350.
Coefficients for siglent Hamming and Hann, for Magnova Hamming.

[kmoonwalker]

Hi, do you have similar noise floor without averaging on SDS3000?

Both look way better than on Josh setup so I wonder if his hoise floor is impacted externally or this is the difference between 2000 and 3000 series

[tautech]

Hi, do you have similar noise floor without averaging on SDS3000?

Both look way better than on Josh setup so I wonder if his hoise floor is impacted externally or this is the difference between 2000 and 3000 series

Maybe they might chose to display the Siglent FFT menus for you to examine the many ways FFT can be configured.....

[kmoonwalker]

Hi, do you have similar noise floor without averaging on SDS3000?

Both look way better than on Josh setup so I wonder if his hoise floor is impacted externally or this is the difference between 2000 and 3000 series

Maybe they might chose to display the Siglent FFT menus for you to examine the many ways FFT can be configured.....

as you can see on pictures posted by Martin - two first are from Josh and later ones from Martin settings looks same for Siglent just Martin used averaging - anyway they have at least 10 dBm noise floor difference which makes me wonder if there isnt really something that Josh could improve on his setup

or 3000 series really has such better dynamic range for fft

[Martin72]

Maybe they might chose to display the Siglent FFT menus for you to examine the many ways FFT can be configured.....

Except for the averaging, I followed Josh's settings, which he shared in the post I quoted(picture).

[egonotto]

Hello,

Martin uses 65536 samples while Josh only uses 16384 samples.

Best regards
egonotto

[tautech]

Hi, do you have similar noise floor without averaging on SDS3000?

Both look way better than on Josh setup so I wonder if his hoise floor is impacted externally or this is the difference between 2000 and 3000 series

Maybe they might chose to display the Siglent FFT menus for you to examine the many ways FFT can be configured.....

as you can see on pictures posted by Martin - two first are from Josh and later ones from Martin settings looks same for Siglent just Martin used averaging - anyway they have at least 10 dBm noise floor difference which makes me wonder if there isnt really something that Josh could improve on his setup

or 3000 series really has such better dynamic range for fft

A Peak Hold setting might account for that.

Without a Menu showing how might anyone know ? 

[Martin72]

@egonotto:
My scope has twice the sample rate with the same time base.
I followed Josh's post, where he took the measurement at 2 µs/div, not the one I´ve quoted before, sorry.
Then he has 32k points and I have 65k.
https://www.eevblog.com/forum/testgear/rohde-schwarz-sml-series-signal-generators-sml01-sml02-sml03-smv03/msg5923934/#msg5923934

[kmoonwalker]

yep but samples amount should not impact the noise floor level - just better representation

I think we need to wait for Josh to share a pic with full scale alike Martin shared cause he cutted bottom part on pics after cable repair.
Also there the fall towards noise floor looks closer to Martins current results so maybe that fixed it for him, too


I am curious also due to fact I am considering to purchase some scope and such comparison (expecially with Magnova added) is very nice to see

[Martin72]

A Peak Hold setting might account for that.

Without a Menu showing how might anyone know ? 

I mentioned earlier that Josh had also posted a screenshot of his FFT settings.

[kmoonwalker]

from what I saw on Martins pictures from both Magnova and 3000 series I gotta say both are tempting still I need to settle for one :/

[tautech]

I'll put up some examples of 50 MHz FFT with a few scopes in the next day or so, SDG6000X and SSG3000X will be the sources....gotta big mess in the engineering shop to clean up first from yesterday's Ag equipment modifications......
And it's perfect weather to check zero of my new to me shooter......
Never enough time....

[KungFuJosh]

Martin's time base and span were different than mine. In this one I matched his settings (+4 avg), and it might look similar to you now. 😉

[KungFuJosh]

And since sampling was brought up, here's a comparison with the same cable and setup, but connected to the SDS5054X. It has a higher sampling rate than both of the HD scopes, but a slightly higher noise floor also.

[KungFuJosh]

I just made another cable, this time I did a much better job on the shielding braid than I did on the other one. It improved stability even further.



Thanks,
Josh

[kmoonwalker]

 

now this is a proper noise floor - congrats

just one question - is this taken with SDS2000x hd series or your 5000x scope?

[KungFuJosh]

now this is a proper noise floor - congrats

just one question - is this taken with SDS2000x hd series or your 5000x scope?

Thanks!

This is with the 2000X HD. The easy giveaway is the GSa/s rate is maxed at 2 on the 2k and 5 on the 5k. That and the 5k is slightly noisier (see reply 90 above). 😉

[kmoonwalker]

now this is a proper noise floor - congrats

just one question - is this taken with SDS2000x hd series or your 5000x scope?

Thanks!

This is with the 2000X HD. The easy giveaway is the GSa/s rate is maxed at 2 on the 2k and 5 on the 5k. That and the 5k is slightly noisier. 😉

I do not have Siglent specs so well in my head and on some scopes there are settings to tamper with so better to ask and be on safe side

Means that noise floor is pretty good on 2000 and 3000 series and that is valueable info for me - Many thanks!

[KungFuJosh]

Means that noise floor is pretty good on 2000 and 3000 series and that is valueable info for me - Many thanks!

Noise floor is practically the same between those two. In theory, the 2000 might be slightly better because of the higher range of the 3000 series. I've not seen anything that convinced me one way or the other tho. Both are excellent.

[tautech]

Means that noise floor is pretty good on 2000 and 3000 series and that is valuable info for me - Many thanks!

You ain't seen my contribution yet ! 

Started building a series of screenshots before a planned dinner with lifelong friends interrupted proceedings and now with 3 too many under the belt will need to wait until tomorrow to complete..... 

[egonotto]

Hello,

surprisingly, the noise of the SDS3034X HD in the 1 V/div range is slightly better than that of the SDS2104X HD.
With the SDS3034X HD with a bandwidth of approximately 420 MHz, the noise is at 4 mVrms.
With the SDS2104X HD with a bandwidth of approximately 180 MHz, the noise is slightly above 5 mVrms.

Best regards
egonotto

[tautech]

I'll put up some examples of 50 MHz FFT with a few scopes in the next day or so, SDG6000X and SSG3000X will be the source.

50 MHz 0dBm FFT tests with 50 MHz span using SSG3021X only...too much dicking around with cabling otherwise. 
Siglent 6 GHz rated 600mm N type cable with N-BNC adapter used for all tests, except SDS800X HD when a Tek 50 Ohm feedthrough was also used to mimic the 50 Ohm input 800X HD models do not have.

All screenshots used a Max Hold setting and max FFT pts (2-8 MPts) available in each model provided and last we played with some ERES and a narrower Span with SDS6204A.

Take note of the Span used, RBW settings, reported frequencies and amplitudes in each screenshot.
Attachment filenames represent the model in use.

We start with a FFT from SVA1032X and work to SDS6204A.

[KungFuJosh]

Our time bases were at 2us/div. Here's 1ms:




Same setup, switched to 2us:




There's no comparison if we don't keep our setups consistent.