Dave takes a look at the new Siglent SDG2122X / SDG2042X Arb Generator and compares it to the Rigol DG4162 and the older Siglent SDG5082. Including some performance measurements on the Tektronix MDO3000 spectrum analyser.
Thanks Dave, nice overview and well done.
Nice to see it measures up well against competition.
If the license on the $499 40MHz version is hackable, it'll sell like hot cakes.
Thanks Dave for the review!
Wondering if 40MHz gen will also have distortion at the top frequency ranges. Anyone to measure this?
Is it this thread for questions to the Siglent CEO who plans to visit you in December?
Here is one: Is he aware that the green LEDs on the buttons are very dim and if they are going to do anything about it?
Maybe giving him this questions before his visit would be a good idea, so he can consult with staff and prepare.... just sayin....
It was funny when Dave was unable to synchronize the phase of the two 10Hz square wave signals because the delay was limited to 100ms.
On the Rigol set the pulse frequency to something appropriately high to allow you to set a low lead edge/rise time, then enable a 1 cycle burst and use the burst period to set the frequency you desire.
You accidentally whole EEVblog #805
CEO question: rust, WTF mate?
European prices from the Siglent website are as follows:
SDG2042X 40 MHz 475 Euro
SDG2082X 80 MHz 589 Euro
SDG2122X 120 MHz 855 Euro
Tempting, I wonder how Rigol will respond to these prices?
European prices from the Siglent website are as follows:
SDG2042X 40 MHz 475 Euro
SDG2082X 80 MHz 589 Euro
SDG2122X 120 MHz 855 Euro
Tempting, I wonder how Rigol will respond to these prices?
These are ex. VAT. The 2042X for example would be 565,25 Euro (Batronix) for a private customer.
I just had a play along myself with the Arb function gen built in to the MDO3000. It doesn't quite have the features of these dedicated boxes, but the performance is pretty decent:
Using the same 10MHz 500mVpp input signal:
The maximum output of the Tektronix scope is 50MHz sine, but it still looks pretty clean compared to the Siglent at it's maximum.
On the Rigol set the pulse frequency to something appropriately high to allow you to set a low lead edge/rise time, then enable a 1 cycle burst and use the burst period to set the frequency you desire.
I wonder if this is what Siglent are doing internally to implement their feature. Regardless, what you describe isn't exactly user friendly -- seems like a bit of a garden path to have to go down just to get an occasional pulse.
On the Rigol set the pulse frequency to something appropriately high to allow you to set a low lead edge/rise time, then enable a 1 cycle burst and use the burst period to set the frequency you desire.
I wonder if this is what Siglent are doing internally to implement their feature. Regardless, what you describe isn't exactly user friendly -- seems like a bit of a garden path to have to go down just to get an occasional pulse.
With the Rigol basically what most of the inbuilt functions (triangle, pulse, harmonic, etc) are doing is allowing you to create a 16K (16384) sample waveform by setting the various parameters relevant to the type of waveform, and set the sampling rate. The output just becomes the 16K samples repeated over and over at the desired sample rate. When you enable burst all it is doing is creating dead time between outputting each 16K waveform. Obviously you now lose a fair bit control over where the falling edge is placed since it has to be contained within the same 16K samples as the rising edge. e.g. If you set the 'Pulse frequency' to 10MHz to achieve a fast rise time and then use the Burst function to set the period t=1ms to achieve 1kHz, the falling edge can't be more than 1/10MHz = 100ns after the rising edge.
Strangely, on the lower frequencies the rigol does not allow you to set the rise/fall time low enough to achieve a rise/fall over 1 sample which is why you get the 'stair steps'. You could get around this by creating a custom arb waveform if you really needed to. On the Square function the Rigol does always transition in 1 sample and the DAC always runs at maximum bandwidth so you achieve the fastest rise/fall time at every frequency.
What the Siglent seems to have on the pulse function is the ability to 'pause' at the high and low parts of the waveform so it can always run the edges at maximum sample rate, possibly dedicating as many samples to the edges as possible. I'd be interested to know how it handles edges >8.4ns. E.g. can it handle 12ns correctly, or does it only do 8.4ns increments or possibly even larger increments? That will tell a lot more about how it is achieving that functionality.
You accidentally whole EEVblog #805
CEO question: rust, WTF mate?
I understand the rust, that's actually the one thing I don't fault them on. Sea air is highly corrosive. Short of plating the steel in a corrosive resistant metal such as nickel the chassis is going to rust. Normally things like machine tools get get coated in a layer of waxy grease prior to shipment, but they obviously can't grease up electronics equipment. They build the units down to a price, so I guess that means the enclosures don't get anti-corrosive nickel plating. Iron redox reactions are a real pain in the ass to deal with.
You accidentally whole EEVblog #805
CEO question: rust, WTF mate?
I understand the rust, that's actually the one thing I don't fault them on. Sea air is highly corrosive. Short of plating the steel in a corrosive resistant metal such as nickel the chassis is going to rust. Normally things like machine tools get get coated in a layer of waxy grease prior to shipment, but they obviously can't grease up electronics equipment. They build the units down to a price, so I guess that means the enclosures don't get anti-corrosive nickel plating. Iron redox reactions are a real pain in the ass to deal with.
Precisely. If you want your money to be put towards making the chassis (which you can't see) prettier rather than putting that same money towards more capable electronics, then you have the Rigol SigGen available to you. I know which I'd prefer.
The rust would bother me, how much does it cost to put it in a plastic bag with some desiccant?
edit-I didn't know they had the bag/desiccant, it sounds like they need to keep the cases out of a humid environment instead or prep them to handle it.
The rust would bother me, how much does it cost to put it in a plastic bag with some desiccant?
That's exactly how Siglent products are shipped.
I've never seen rust on Siglent products other than those posted here and they all seem to be on the cut edges of the plated chassis.
I have a Tek TDS2012B that has rust on the cut edges of the chassis too.
The first I heard of rust was it seems from poorly plated sheet steel from Siglents chassis suppliers.
Siglent has replied in various threads that this has been rectified.
How difficult would it be to either zinc dip or paint a chassis after construction but before electronic assembly? The other option of course is to make the chassis out of aluminum.
How difficult would it be to either zinc dip or paint a chassis after construction but before electronic assembly? The other option of course is to make the chassis out of aluminum.
There has been much todo of Siglent rust and it is likely they do that now, I'm sorry I don't know.
In saying that one also must understand the process of zinc cathodic protection, it seems not many EE's do.
https://en.wikipedia.org/wiki/Cathodic_protection
I'm wondering how the high frequency measurements at the end where done. Without the scope set to 50 Ohm it is likely the input capacitance messes things up and the measurement is not very relevant at these frequencies.
BTW If Siglent's CEO visits give him a 'Programming for dummies' book and some Hammerite
BTW If Siglent's CEO visits give him a 'Programming for dummies' book and some Hammerite
LOL I see that little rust spot really upset many users here.
The serious question about this would be something like:
"Do they know on what point the housings have been stored or handled in such a moist or harsh environment that some started developing rust spots"
"Have they taken measures to prevent this from happening in the future"
Many manufacturerers have outsourced the production of these parts and buy them from specialized manufacturers.
I use to call them "sheet metal benders"
(of course german: Blechbieger)
Here is one of them:
http://www.psw-umformtechnik.de/index.php?article_id=5&clang=1A visit and inspection of the production facilities is usually a good thing to do when you contract one of those.
However that is all guessing, depending on the production volume it's sometimes cheaper to make these things in house.
Hello Dave,
You mentioned that the Rigol has more built-in arbitrary waveforms than the Siglent.
But even the Siglent has quite a wide range of arbitrary waveforms, like mathemical, etc.
Maybe it's a good idea to highlight the specific use of some of these arbitrary waveforms in an actual use case. Specific circuit or application which needs to be tested, and for which you need to have a specific arbitrary waveform on the input... I am a bit blank on this area, and would like to have a better understanding of the specific use of these pre-built arbitrary waveforms. If possible a coverage of several categories, and each time a specific use case illustrated on a specific circuit or application.
Believe that this will be very helpful to the bigger audience.
I have only used a signal generator in school for measuring out filters, but never used any of these built-in arbitrary waveforms. Not available at my time AFAIK
The Siglent X series seems to be much cleaner. But it would be interesting to take a look at the harmonics from the Marconi generator Dave used as a comparison as well. Usually RF generators are not that clean when it comes to harmonics. Having the 2nd or 3rd harmonic at -30dB or worse is not an exception.
I'd be interested to know how it handles edges >8.4ns. E.g. can it handle 12ns correctly, or does it only do 8.4ns increments or possibly even larger increments? That will tell a lot more about how it is achieving that functionality.
I have attached screen-shots from 15ns risetime dwonto 9ns rise time in 1ns steps using the SDG2042X. So you can see it does 1ns increments at least. I have not checked shorter increments yet...