Testing risetime and bandwidth with a Siglent SDS2102X Plus

[Yellie]

Hi,

I received a Siglent SDS2102X Plus scope a few days ago and wanted to the test the specs. First I connected the 200 MHz probe to the build in calibrating 1 Khz. squarewave and expected a risetime around 3.5 nsec but the display shows 3.4 usec. A different of 1000 times slower. The second check I did, after activating the option for the build in AWG, to display a 10 MHz. squarewave. I connected the 50 ohm AWG output to the 50 ohm impedance input by a 50 coax (Aircell 7) and BNC. A disappointing sinewave was the result on the screen.
The questions: 1 - Is it a misunderstanding that the risetime must be 3.5 nsec with a squarewave or is it a theoretical value? 2 - I updated my Siglent temporaly for this test to 350 MHz and in that case I expected that a 10 MHz squarewave must be shown as a real squarewave on the screen. Maybe the build in AWG cannot generate a real 10 Mhz squarewave or is the bandwidth of the scope still to small? 3 - Are there others with the same experience? I attached some pictures of my tests.

Thanks so far

[delvo]

The calibration signal is not really meant for that. It doesnt have a great rise/fall time. I compared with my SDS1104X-E which is upgraded to 200MHz. The test signals rise time is around 1us.
The "disappointing sine wave" you see is a result of the rise/fall time of the AWG. A 10MHz square wave has a period of 100ns. So it is positive for 50ns. The rise/fall time for the AWG is specified at <24ns. So it really cant look like a square wave at that speed anymore.

[Yellie]

Hi Delvo,

Thanks for the quick reply. I was a little bit confused and expected to see a nice 10 MHz squarewave on the 350 MHz bandwidth scope. So the main problem is the internal AWG of the scope? I don’t have an external high speed AWG but in that case my SDS2102X Plus (with 2352X Plus specs now…) has to show a nice 10 MHz squarewave? Are there others they did such a test? I’m very interested in that.

Best regards

[delvo]

The problem is really generating the square wave. I have a SDG2042X which has much better specs than the integrated AWG and its rise/fall time is not enough to get a nice looking 10MHz wave. Im sure it will look good when you get a good square wave source.

[Yellie]

That makes me feel confident and I’m glad to hear that. Thanks again.

[tggzzz]

Welcome to the forum. Good to see someone testing and thinking.

Basically an AWG is the wrong tool for that job.

There is a thread on the topic:https://www.eevblog.com/forum/testgear/show-us-your-square-wave/ You might like to speed-read it to find different approaches.

You should also understand the consequences of using "high" impedance 10Mohm *10 probes: the 15cm ground lead causes oscillations and they aren't 10Mohm at 100MHz. FFI, see the references at https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/

[Yellie]

Thanks for the advice, Ill will also have a look about testing squarewave items in this forum.
About my tests:  to prevent hf problems with probes therefore I used a 50 ohm coax and bnc’s. Mostly I will use the scoop in the future for lower frequencies and repair work. In that case this model Siglent is far enough a good choice I think. But in the background there is always the question, do I get real value for my money? ;-)

[Performa01]

I was a little bit confused and expected to see a nice 10 MHz squarewave on the 350 MHz bandwidth scope. So the main problem is the internal AWG of the scope? I don’t have an external high speed AWG but in that case my SDS2102X Plus (with 2352X Plus specs now…) has to show a nice 10 MHz squarewave? Are there others they did such a test? I’m very interested in that.

Here is a similar test: not a square wave with 50% duty cycle, but a pulse train at 10 MHz repetition frequency, 10 ns pulse width and 1 ns rise time (generated with a Siglent SDG7102A). Scope bandwidth is limited to 350 MHz because of full channel mode.

[Yellie]

Hi Performa01. That looks near a squarewave, so it must be possible.
When I try to do the same trick with the build in AWG in the Siglent SDS2102X Plus scoop it is near a sinus wave at 10 MHz. So it must be the AWG. A litlle bit disappointing about this feature but I bought the Siglent finally to use as a scoop and not as an AWG. So, it's good to know where the weak points are... 

[tggzzz]

Thanks for the advice, Ill will also have a look about testing squarewave items in this forum.
About my tests:  to prevent hf problems with probes therefore I used a 50 ohm coax and bnc’s.

That reply indicates you do not have a solid understanding of how coax cables and probes work, and how they affect the system you are testing.

Your first task should be to understand how the different classes of scope probe, how they work, why there are several classes and their relative advantages and disadvantages. Start with the references I gave.

Mostly I will use the scoop in the future for lower frequencies and repair work. In that case this model Siglent is far enough a good choice I think. But in the background there is always the question, do I get real value for my money? ;-)

Only you can decide that!

My suggestion: learn how to use the tool effectively. Only if it does not satisfy your needs should you consider another tool.

[gf]

Hi Performa01. That looks near a squarewave, so it must be possible.
...
So it must be the AWG. A litlle bit disappointing about this feature

It is possible. However, did you realize that Performa01's SDG7102A is 20x faster, and that it has a price tag of > 10,000 bucks?

What you get meets the specs given in the datasheet. Keep in mind that a 125 MSa/s AWG -- which is based on band-limited Nyquist-Shannon reconstruction -- can hardly do much better than 24ns rise/fall time w/o introducing significant overshoot, and jitter at certain frequencies and pulse widths. Why be disappointed if it performs close to the feasibility limit anyway?

Btw, don't use an AWG if you want square waves with fast rise/fall time. Rather use a dedicated pulse generator. For instance, the Si5351 clock generator chip achieves  about 1ns rise time, and inexpensive Si5351 boards are available for 30 or 40 bucks. Or get e.g. a Leo Bodnar pulser if you need even faster edges (significantly shorter than 1ns).

[Martin72]

(significantly shorter than 1ns)

A little bit - 40ps...
In some thread someone had determined the bandwidth of a SDS2000X plus by frequency sweep and FFT, FFT mode was peak if I remember correctly.
I just do not know where that was and how it works exactly.

[gf]

I think it was Performa01, too. Sweep slowly with a leveled RF generator and let the scope record the result with FFT with peak hold. FFT window was rather short, just covering a small fraction of the sweep time, and peak hold accumulates many FFT windows.

[2N3055]

I think it was Performa01, too. Sweep slowly with a leveled RF generator and let the scope record the result with FFT with peak hold. FFT window was rather short, just covering a small fraction of the sweep time, and peak hold accumulates many FFT windows.

You still need a 500MHz sinusoidal signal source with very good amplitude flatness...

But there is no need to "verify BW". Scope is just fine. I agree 100% with tggzzz that OP should start learning as much as he/she can how to use the scope, probes, caveats, trials and tribulations..

But in the background there is always the question, do I get real value for my money? ;-)

You bought best scope in it's class. Most of the problem with it is that it has thick book you need to learn to even start to use fraction of it's features.

My best suggestion would be to just take a tipical project you are interested in and start experimenting and learning.  There is only one thing you need to be carefull: make sure to not overload scope in 50 Ω input mode. I would suggest to start your learning figuring out why it is so......

P.S. tggzzz has very nice web site with many good  articles in regards to scopes and probes. Make sure to read it..

[BrokenYugo]

 A good fast edge pulse source is fairly easy to build, fun day project, usually looks like a really fast logic gate slowly oscillating, or a relaxation oscillator based around avalanching a BJT. Many threads on this topic, but start with appendix B of AN72.

http://cds.linear.com/docs/en/application-note/an72f.pdf

[Martin72]

You still need a 500MHz sinusoidal signal source with very good amplitude flatness...

OK, I plead guilty....
I brought this into play because I would like to measure the bandwidth of my scope this way and I now have a good source for it.

[gf]

(significantly shorter than 1ns)

In some thread someone had determined the bandwidth of a SDS2000X plus by frequency sweep and FFT, FFT mode was peak if I remember correctly. I just do not know where that was and how it works exactly.

I found the message with the explanation again:
https://www.eevblog.com/forum/testgear/bandwidth-limit-on-siglent-sds2000x-plus-oscilloscope/msg5006257/#msg5006257

[tggzzz]

A good fast edge pulse source is fairly easy to build, fun day project, usually looks like a really fast logic gate slowly oscillating, or a relaxation oscillator based around avalanching a BJT. Many threads on this topic, but start with appendix B of AN72.

http://cds.linear.com/docs/en/application-note/an72f.pdf

Pulse generators are fine after you have characterised the pulse's amplitude with a faster scope. What you see on the display is the convolution of the input signal with the scope's response.

IMNSHO it is preferable to use a step generator, since

• they directly reflect a normal use case, the shape of a "digital" waveform
• are easier to build so they are "right by design"
• require less characterisation

Leo Bodnar's is one such, but even modern jellybean 74LVC1G* gates can be persuaded to give 2.5V 250ps into 50ohms; see https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg1902941/#msg1902941 . Naturally great care has to be given to layout and decoupling - don't even think of using solderless breadboards, but manhattan might be OK for scopes up to a few hundred MHz.

[tautech]

I received a Siglent SDS2102X Plus scope a few days ago and wanted to the test the specs.

Why ?
Risetime is stated in the datasheet at 3.5ns on P10:
https://int.siglent.com/u_file/download/22_09_19/SDS2000X%20Plus_Datasheet_EN01C.pdf

-3dB BW for these 100 MHz models is tested and confirmed with 3 waveform sources at ~185 MHz:
https://www.eevblog.com/forum/testgear/siglent-sds2000x-plus-coming/msg2958288/#msg2958288

[gf]

I received a Siglent SDS2102X Plus scope a few days ago and wanted to the test the specs.

Why? Risetime is stated in the datasheet

It's not like you don't trust the datasheet. The journey is the destination - you just need to have measured it yourself once (and understand what you are doing, and why you do it the way you do it) Consider it an exercise.

[tggzzz]

I received a Siglent SDS2102X Plus scope a few days ago and wanted to the test the specs.

Why? Risetime is stated in the datasheet

It's not like you don't trust the datasheet. The journey is the destination - you just need to have measured it yourself once (and understand what you are doing, and why you do it the way you do it) Consider it an exercise.

Precisely!

It is about testing the person's understanding of the dcooe, how to use it, measurement techniques and pitfalls.

I applaud the OP for doing it, think they will learn a lot, and will emerge a better engineer

[mwb1100]

Not to mention that whenever you get a new toy you just want to see what it can do.

[tautech]

Not to mention that whenever you get a new toy you just want to see what it can do.

That presumes you know how to use it and find its limits, not everyone does.
There's a good reason why I shelled out on a RF gen and Leo Bodnar 30ps pulser.

[Martin72]

"Tell me, and I will forget.
 Show it to me, and I may keep it.
 Let me do it and I will be able to."

[BillyO]

A little bit - 40ps...

Well, you don't have to go that far.  Especially if you want to employ a little math.

For testing a rise time near 3.5ns, 500ps would be fine and introduce an error of just over 1% without doing any math.

You would introduce much more error by hooking everything up incorrectly.