EEVblog

Dave takes a look at the new Siglent SDG2122X / SDG2042X Arb Generator and compares it to the Rigol DG4162 and the older Siglent SDG5082.
1.2GS/s arb gen for US$499!
Including some performance measurements on the Tektronix MDO3000 spectrum analyser.
Trio Test & Measure: www.triotest.co...
Forum: www.eevblog.com...
EEVblog Main Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
Support the EEVblog through Patreon!
/ eevblog
EEVblog Amazon Store (Dave gets a cut):
astore.amazon.c...
T-Shirts: teespring.com/s...
Donations:
www.eevblog.com...
Projects:
www.eevblog.com...
Electronics Info Wiki:
www.eevblog.com...
💗 Likecoin - Coins for Likes: likecoin.pro/@...

[@voltlog]

The hobby community can only be happy. The way things go these days more and more hobbyists can afford equipping their home labs with nice instruments.

[@EEVblog]

+VoltLog It's crazy what you can get for the money these days. Wasn't that many years ago that a 2MHz sine/square/triangle function gen was the entry level standard.

[@iceberg789]

+VoltLog may be increase in sales making companies produce stuff at bulk and and lower costs. a lot of hobbyists buying decent equips these days.

[@voltlog]

there might be more factors to this, like for example R&D might be cheaper for these smaller companies, they don't design custom asics, they use off the shelve parts and they don't exactly use top of the line components in the hardware. Shaving a few dolars here and there coupled with a lower profit margin might be the reason the end equipment is cheaper.

[@kurchak]

@22:41 Just bought this model in 2025, and it looks like they listened to your recommendation! They ship it with a single BNC cable now 😄. Nice quality cable too. Thank you for telling them that!

[@steverobbins4872]

Dave, the "strange" bumps you saw on the spectrum analyzer are due to the loop BW of the PLL. A PLL can be thought of as a low pass filter for jitter and phase noise. In other words, a PLL will track phase noise that is within its loop BW, but can't track phase noise that is outside its BW. The ref osc (typically an oven-controlled crystal osc) will have extremely low phase noise, and the VCO that is part of the PLL will have significantly higher phase noise. So the output phase noise of the PLL crosses over from the ref osc curve to the VCO curve at the 3dB BW of the PLL. These crossover points typically look like the bumps you saw, rather that distinct spurs.

[@JAKOB1977]

Dave at his finest. objective and just let reality stand on its own two feet and let the chips fall as they please. love it.

[@JimGriffOne]

To get rid of sidebands, I would use a -10dB attenuator on the input of the SA to reduce input distortion and change the graph offset to +10dB. Of course you lose 10dB of noise floor, but that shouldn't really matter as much (except for military, mobile communications engineers, R&D, etc.) If you want to see what's at the noise floor, you usually wouldn't be analysing a signal several orders of magnitude greater at the same time anyway.

[@q3kq3k]

17:04 that's got to be my favourite Dave sound as of today :D

[@DeeegerD]

You can't compare 120 MHz between the two machines when one is at 9+ volts and the other is only at 4+ volts. Cut the Siglent voltage to 4v and I bet those sine waves clean up nicely as on the Rigol.

[@Wooksterish]

why is it that you can't compare the two channels if the voltages are different?

[@wadehsu2347]

TheRealMoomin it’s not comparing the two channels, it’s comparing the output of two different instruments. One Rigol and one Siglent. The problem of that comparison is that Siglent is driving twice as much output swing than Rigol at the expense of larger distortion. So if one really want to compare distortion accurately, the output drive level need to me the same.

[@tHaH4x0r]

@@Wooksterish A higher amplitude requires a faster rise time. For a certain signal A*cos(omega*t) you can find the slope with -A*omega*sin(omega*t). I.e. increasing the amplitude or the frequency causes an increase in maximum signal slope. As you can imagine, it is harder to drive reactive loads with a higher slope. So things like cable capacitance will ruin your day sooner! So yeah, Digger D is right, you should compare apples to apples between the two devices, and have the exact same waveform specs to compare, especially important at the higher frequencies.

[@hectorcordova5794]

On the Silent, you should press Waveforms/Sine/Harmonics and there are the harmonica functions.

[@raymundhofmann7661]

Would have been nice to see the spectrum of the distorted high freq. waveform.

[@benadams6332]

Nice review, thanks. This sure is a great value instrument. I like the smaller form factor much better than the huge scope shaped rigol.

[@MrCarlsonsLab]

LOL, Staircase rise and fall on the Rigol, (facepalm) I was hoping you would look at the rise and fall of the actual staircase itself :^) Good review Dave.

[@benadams6332]

+Mr Carlson's Lab It even looks like only 4-5 bits resolution on that staircase.

[@npntube6847]

+Alfred Stampe Not a matter of limited resolution but limited memory, I believe. The whole cycle is stored in the limited memory which leads to a low sampling rate and finally results in the stair case.

[@npntube6847]

+NPNtube Rigol: There must be a minimum number of stairs to have a defined ramp. I count a bit more than 32 stairs, maybe exact 32 stairs from 10% to 90%. 10Hz vs. 195µs ramp time vs 32 stairs lead to 16384 samples memory depth which matches the specification of the device. Siglent on the other hand specifies 8M samples... huge difference.

[@Washburn-rr5eh]

And she's buying the stair way, to heaven.

[@xDR1TeK]

+1 Siglent, neat ARB WaveGen

[@dryyourtears5275]

@10:56 that wide noisy thing around the carrier might be the phase noise of the local oscillator

[@drkastenbrot]

Im pretty sure i could do a nice dubstep bass growl on that device. Its quite similar to my favourite synth, xfer serum.

[@highfidelityinc]

Great video. The baseline spread at 80MHz -22dBm output of the SDG2000X (11:00minute in video) may be caused clock jitter or quantizing noise in the SDG2000X. Curious what others think.

[@svampebob007]

I'll just watch a quick video from Dave :)

[@peddersoldchap]

Why does the BK Precision 4060B series looks just like the Siglent featured in this video???

[@eastmanresearch3143]

Does anyone know of an op-amp (high voltage) which can be used with these waveform generators? I am working on generating automotive iso 7637-2 pulses for doing transient testing on equipment w/ the true-arb technology. Voltages will need to go up to +/- 150v

[@paigebartz1451]

Dave you should do a livestream where you just clear out every single package

[@ChrisSmith-wm7lz]

I'm a Newby to this. Could you do a video where you actually hook up the generator to a circuit and give an explanation of what's happening with the circuit.

[@drdiesel1]

Nice bench friendly short form factor. :thumbsup:

[@EEVblog]

+Andy Lawrence Yeah, small gear is nice. Gone are the days of huge gear hogging your whole bench.

[@tubical71]

+EEVblog that´s why i saved my money and bought a DSO, back when they cross the 500 bucks line...and i still need no newer one, as all i need is somewhere below 10MHz, mainly audio-frequencies...Thanx modern technology to let us buy these things, and also to let us buy old things relatievely cheap....:) And thanx to You, Dave, to put a light on these things!!!

[@IsaacC20]

Quick question about DSOs: If my uC runs at 80 MHz, do I need an oscilloscope capable of at least twice that bandwidth to see the signal?

[@alperenalperen2458]

Some people reccomend 5 times bandwidth for square waves. uC runs from square wave clock its bandwid is also depended on the rise time.

[@tHaH4x0r]

Your uC might run at 80mhz, but that is the internal clock. You will notice with most uCs the maximum output frequency you can achieve is about 2-3Mhz. What Alperen says is true, square waves look the way they do because they posses higher frequency content beyond their base frequency, and thus you need a higher bandwidth to display them properly. So a 15Mhz scope would more than suffice, which is practically any scope nowadays. About the doubling of bandwidth, I think you are confused by the Nyquist criterium. It dictates that your SAMPLING frequency should be at least twice the MEASURED frequency. This is a digital phenomena due to sampling. The 'bandwidth' on the oscilloscope though, is not from digitizing. It is instead an analog bandwidth. I.e. with a 100mhz scope you can measure at most a 100mhz pure sine wave (usually scopes do better than their quoted analog bandwidth though). On this note, scopes also often quote a sample rate, for example 1Gs/s. That means that their sample frequency is 1Ghz. This is the frequency that matters for the above nyquist criterium. But for any modern well designed scope, this sample rate is far above the analog bandwidth, even in worst case scenarios.

[@davidgaleski5303]

thank you

[@marcusvinicius-yo4ii]

I have a Koolertron 60MHz and looking to upgrade for a Siglent 2000x series and wondering if I should get the 2042x or the 2082x is worth the extra 120$? I already got a Siglent sds 1202x-e O'scope, spd3303x-e DC PS, and a siglent SDM 3055 DMM.

[@marcusvinicius-yo4ii]

nvm, decided to splurge on the SDG2122x XD

[@lmiddleman]

That Rigol has commas between every three digits after the decimal point in the frequency? Is that common anywhere in the world? Major annoyance if that can't be disabled.

[@jope7137]

+lmiddleman It's called "digit grouping" and it is not uncommon, because it makes reading big numbers more easy. You can disable it on the Rigol (and also exchange dot and comma as decimal and group separators). Interesting Wikipedia article: en.wikipedia.org/wiki/Decimal_mark

[@ethanpoole3443]

+Jo Pe It is a shame that we all could not have cone to a common agreement on how to specify decimals and digit groupings in a single universal format for everyone. It makes internationalization very confusing when 3,100.099 and 3.100,099 mean both the same value and simultaneously a very different value, either of which may be the correct value.

[@MrJackZhan]

Hi Dave, what do you recommend for a budget-friendly Arb Gen? I have a Tek 2445 scope and I am looking for a proper Arb Gen. Thank you!

[@lmiddleman]

Doesn't it make sense that the delay is limited to 100ms when the frequency is 10 Hz? As in one full period?

[@userPrehistoricman]

+lmiddleman It would make more sense for the maximum delay to be two periods.

[@lmiddleman]

I was assuming it was plus or minus 100ms.

[@userPrehistoricman]

lmiddleman Negative delay? So this is how they make time machines...

[@joe72205]

+lmiddleman I think Dave misread, 100.000,000 is 100K in "European" style decimal separators

[@schloumf]

hello EEVblog, i know you have not time to answer personal electronic questions. So can you drive me to a book or a website that would instruct me how to mix a high tension 50 hertz, with the 1,5 volts and any Hertz from that come out of a Siglent function/arbitrary wave form generator ? -- an exemple: a microwave oven transformer 50 hertz that would give, after a circuit, the same voltage with any of the hertz coming out of the Siglent. Thanks very much EEVBlog !

[@FFcossag]

Is Sigilent Green people?

[@vinayutubeaccount]

No 4K?

[@brandonpais2532]

3:30 Whats the "blackman" function?

[@KarlBaron]

+Brandon Pais en.wikipedia.org/wiki/Window_function#Blackman_windows

[@worm6942]

What did you do at uni?

[@DocFirewire]

Ya, the Rigol is bigger, so it must be better! But Daves #1 Argument against the Siglent seems to be the term "weird" - All in all, clearly a fanboy babbling along! We are talking cheap chinese gear competiting against each other, so calm down please. Let me show you something with a risetime of 6 picoseconds if you care about high performance. I remember that gear with this specs was well worth a fully loaded E-class Benz 10 yrs ago, lotsa $$$. This chinese stuff does the basic jobs in any respect. Speaking of $$$, I now go pet my Tek 2230 downstairs... 😁

[@ElmerFuddGun]

+EEVblog WOW... that Tek MDO wasn't responding for 18 seconds! That's just crazy bad designing. I mean did Microsoft design it or something? I would have likely shut it off thinking it had hung BUT it looks to have a soft power button too and wouldn't respond to that either! For the price of that I would have sent it back as unacceptable!

[@GeorgeTsiros]

15:25 195.312,5 us is 195 milliseconds, not microseconds. The comma is the decimal point.

[@Diggnuts]

"won't be comprehensive"..... Zooms out......

[@gudenau]

Under 300 club. :-D

[@maxsnts]

Ins't that 195 thousand us?

[@milamiglia2437]

They need to get rid of the buttons and use mouse or touch interface.

[@lookatthisvidsandfun]

Please don't! Don't let the UI cancer swap into electronic test equipment!

[@lookatthisvidsandfun]

I've found some very strange UI bugs, and also the glitches in the signal could be pretty destructive! kzbin.info/www/bejne/nJyUqIWni5hkiMk

[@gahlanaventis]

plz Dave do a favor for non English speakers collect English substitles for your videos and upload all to us

[@EEVblog]

+Gahlan Aventis It is either incredibly time consuming, or very expensive to do this for long technical videos.

[@droelfdroelfify]

+EEVblog Plus, Dave is really easy to understand even for non-native speakers. I happen to be an expert on this subject being a non-native speaker myself.

[@joopterwijn]

Haha, then you have to clean up you lab! (Ceo visit).

[@PatrickOuthier]

+Joop Terwijn Hopefully Dave remembers to roll out the red ESD mat.