Best arbitrary waveform generator for around $500

[wraper]

I also can't remember that they ever came up with something remarkable.

Look at their new GDS-1000B series oscilloscopes (if not considering hacking, beats DS1000Z in almost every way including price, unfortunately no options for purchase available), GDS-2000A oscilloscopes.

Not sure the GDS1000B really beats the DS1000z in almost every way, it also doesn't in price:

http://www.newark.com/gw-instek/gds-1054b/oscilloscope-4-ch-50mhz-1gsps/dp/51Y5253
http://www.tequipment.net/Rigol/DS1054Z/Digital-Oscilloscopes/

Comparing apples vs apples, not newark prices? http://www.tequipment.net/Instek/GDS-1054B/Digital-Oscilloscopes/?v=0

The only positive I can see on the GDS1000B is that FFT uses up to 1Mpts compared to a few thousand points with the Rigols (all of them up to the DS6000) and the digital filter, and I haven't yet seen a GDS1000B in the wild but I'd guess FFT at full 1Mpts will be dog slow due to the large amount of data that needs processing (which the processors found in these low-cost bottom-of-the-barrel scopes can't perform in reasonable speed).

But yes, they were slow to update their product lines when Rigol DS2000 series came but their entry level scopes remained like DS1000E for many years.

Yes, but the DS1000E offered the right amount of features for its time and was priced right, both which isn't often the case with GW Instek gear.

FFT is not slow and actually becomes a usable feature in the scope. It has faster waveform update rates too.

[jhufford]

Wow, thanks for the info rf-loop! You're correct, I need to study, I don't know much about function gens, haven't used one since the circuits 101 days in college, and I've never played with an arb gen. But I'm trying to learn now as I finding myself needing a function gen, so thanks again for the info! it helped send me down the right path.

[BloodyCactus]

saying that, I have a Rigol 1032Z, I dont use the arbitrary much so cant comment on it, but its square wave beyond say 10mhz is crap. at its max supported 15mhz is pure crap, its just a sine wave! it has great memory depth for its arbitrary stuff but not really the bandwidth/speed to do precision stuff thats non sine

How exactly did you come to the conclusion that it is a sine wave? If your DG1032z produces really a sine wave when in square mode then it's defective, or you're doing something very wrong. At the moment my money is on the latter as my DG1062z (which has similar performance specs) produces a perfectly fine square wave within the limits of its specifications.

at 15mhz its closer to sine than square.

at 1mhz, beautiful!


at 15mhz, I dont consider it very good square wave, its just a sine wave with the bottom clipped.

[dadler]

What does it look like @ 1Vpp (assuming proper termination)?

Rigol specs: Square Rise/Fall Time Typical (1 Vpp) <10ns

FWIW: I tested the internal AWG on my DSOX3000T, and I see similar results (rise time is about 18ns). This was 5Vpp@10MHz as the internal AWG maxes out at 10MHz for square wave output.

[nbritton]

I'm in the market for a function gen too. And I'm considering the SDG1025 as well as the 2000 models. I'm willing to spend the $900 for the 2122X model, if it's worth it.

Why does the 2000 series have such a measly arb sample rate (75MSa/S), while sporting a 1.2GSa/S rate on the built-in waves.  But the 1000 series has 125MSa/S for both. The 5000 series has 500MSa/S for both. Ideas?

Thanks.

It appears that the SDG2000X has two arbitrary waveform modes, TrueArb mode (low jitter) which is 75 MSa/s and DDS mode (normal jitter) which is quoted as 300 MSa/s minimum. In the datasheet it also states that the device can do 20 MHz in arbitrary mode, whereas the SDG1000 is limited to 5 MHz. Additionally it appears that the modulation speeds are significantly faster. On the SDG2000X it quotes a max of 1 MHz, whereas on the SDG1000 it quotes a max of 20 KHz.

[Armxnian]

What does it look like @ 1Vpp (assuming proper termination)?

Rigol specs: Square Rise/Fall Time Typical (1 Vpp) <10ns

FWIW: I tested the internal AWG on my DSOX3000T, and I see similar results (rise time is about 18ns). This was 5Vpp@10MHz as the internal AWG maxes out at 10MHz for square wave output.

Rise time seems to stay constant on the sdg1025. About 7ns for a square, which is pretty good. But the performance starts to fall off after 1MHz @1vpp. At 25MHz the square wave is almost indistinguishable from a sine wave. Amplitude doesn't affect performance that much.

The spec Siglent publishes is correct, but it's kind of meaningless if the square wave rated for 25MHz starts to transform after 1MHz.

[nbritton]

What about the Hantek HDG2000B series. The HDG2002B is a bargain at $300 and it can be hacked to 100 MHz.

[Groucho2005]

Rise time seems to stay constant on the sdg1025. About 7ns for a square, which is pretty good. But the performance starts to fall off after 1MHz @1vpp. At 25MHz the square wave is almost indistinguishable from a sine wave.

What scope did you use? Some SDG1025 screen shots taken with a reasonably fast scope:
https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg588973/?topicseen#msg588973

The 25 MHz signal is certainly no square but also far from a sinusoid.

[wraper]

No, it isn't. In my direct environment there are quite a few GW Instek devices, and while the general build quality is OK-ish (not great, still better than Atten), I've seen quite a few of QC issues (i.e. loose missing screws on a scope, loose PSU on another, and a good number of DOAs). The support was also less than stellar (actually pretty much the worst in T&M I've ever seen, and I've seen a lot!) and pretty much for most part really didn't give a shit. Rigol and Siglent are far from perfect and leave a lot to be desired, but in terms of QC where they are today is a much better place than where GW Instek has ever been in its existence.

I had to deal with Rigol warranty in EU. Can say it's almost non existent too. Then how about 2 Rigol DM3068 multimeters of almost the same date code and same PCB revision. One has two bodge wires on PCB and multiple ICs hand soldered (after a week developed strange AC range failure affecting precision), the other does not have all of that mess. Seems the first one had some issues at production and was heavily reworked. Huge quantity of flux residues were not cleaned too, and all of that over the places where guard rings are present. I have disassembled Instek gear and PCBs where very nice and clean.

[Wuerstchenhund]

Comparing apples vs apples, not newark prices? http://www.tequipment.net/Instek/GDS-1054B/Digital-Oscilloscopes/?v=0

Strange, it didn't show up when I searched, But I stand corrected.

FFT is not slow and actually becomes a usable feature in the scope. It has faster waveform update rates too.

As I said I haven't seen any GDS1000B in the wild (people here are pretty much cured from GW Instek) but I'd be really surprised if this $400 scope could handle 1Mpts of FFT data with its processor, something which puts even many more powerful scopes under visible strain.

[wraper]

As I said I haven't seen any GDS1000B in the wild (people here are pretty much cured from GW Instek) but I'd be really surprised if this $400 scope could handle 1Mpts of FFT data with its processor, something which puts even many more powerful scopes under visible strain.

On the video FFT on GSD-1000B is much faster than on DS-1000Z. Should be done in the FPGA.

[Wuerstchenhund]

at 15mhz its closer to sine than square.

No, it isn't. It's not a perfect square but it's far from being a sine wave (and even less os a pure sine, which doesn't occur in reality anyways).

BTW, my DG1062z shows the same, which is not surprising considering the similarities:

1MHz:




15MHz:



A quick view on the signal's frequency spectrum not only shows that this clearly isn't a sine wave but also explains why the 15MHz square wave looks like it does:



It's clearly visible that the 15MHz signal contains non-harmonic components, plus some of the expected harmonics are missing.

A check with a proper spectrum analyzer confirms the malformations increasing with the frequency increasing:

1MHz:


3MHz:


5MHz:


10MHz:


15MHz:


However, that's an imperfection that the DG1000z shares with many other AWGs in that price class, and if you believe that a GW Instek device performs any better then I'd say you're deluding yourself.

There simply are limits what one can expect when buying low end gear, and that's true for AWGs as well. Maybe the new Siglent SDG2000X does better in this regard, but I'm not holding my breath.

Edit: I forgot to mention, the scope used is a LeCroy WavePro 7300A (3Ghz 20GSa/s 48Mpts) connected via a 50ohms BNC cable (tested to 2.7GHz) directly into the scope. The spectrum Analyzer is a Rohde & Schwarz FSP7 (7Ghz), again using the same cable setup plus a Pasternack BNC-to-N adapter.

[Wuerstchenhund]

As I said I haven't seen any GDS1000B in the wild (people here are pretty much cured from GW Instek) but I'd be really surprised if this $400 scope could handle 1Mpts of FFT data with its processor, something which puts even many more powerful scopes under visible strain.

On the video FFT on GSD-1000B is much faster than on DS-1000Z. Should be done in the FPGA.hHSBqsvJM[/url]

Yes, on a promotional video made by GW Instek, with no visibility of the test parameters and scope settings. No thanks, I'd wait until I see some independent confirmation before I believe them.

Plus it seems GW Instek still shoots it's videos on an old VHS camcorder, but hey at least it's in color and not black & white. 

On the other side, the care GW Instek shows to their online representation does very well reflect the care they show towards their customers.

[wraper]

Yes, on a promotional video made by GW Instek, with no visibility of the test parameters and scope settings. No thanks, I'd wait until I see some independent confirmation before I believe them.

Plus it seems GW Instek still shoots it's videos on an old VHS camcorder, but hey at least it's in color and not black & white. 

On the other side, the care GW Instek shows to their online representation does very well reflect the care they show towards their customers.

You were concerned about FFT being very slow. IMO that video is completely enough to see that it is not the case.

[Wuerstchenhund]

You were concerned about FFT being very slow.

No, I was concerned about FFT over 1Mpts being slow.

IMO that video is completely enough to see that it is not the case.

No, it isn't. All it does is show the scope doing some FFT over an unknown number of points (and any scope is fast at FFT if the number of points is sufficiently low). And thanks to the crap low resolution video it's even impossible to see what the actual settings are.

As a demonstration of the claimed capabilities of the GDS1000B the video is pretty much worthless.

[Armxnian]

Rise time seems to stay constant on the sdg1025. About 7ns for a square, which is pretty good. But the performance starts to fall off after 1MHz @1vpp. At 25MHz the square wave is almost indistinguishable from a sine wave.

What scope did you use? Some SDG1025 screen shots taken with a reasonably fast scope:
https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg588973/?topicseen#msg588973

The 25 MHz signal is certainly no square but also far from a sinusoid.

25MHz Sine:


25MHz Square:


What it should look more like, 1MHz Square:

[nbritton]

It appears you're using the DS1054Z. You do realize you would need at least 225 MHz of bandwidth to faithfully reproduce a 25 MHz square wave on your scope. You would need to capture at least the 9th harmonic, which you can't do on the DS1054Z.

Fundamental: 25 MHz
3rd Harmonic: 75 MHz
5th Harmonic: 125 MHz
7th Harmonic: 175 MHz
9th Harmonic: 225 MHz

[Armxnian]

At 25MHz, the 9th harmonic doesn't even exist. That's the point we're trying to make . At that frequency, it's transforming into a sine wave. As Wuerstchenhund showed, the spectral content of the square wave at the high frequency does indeed contain characteristics of a square wave, but not as much as you would like, at least on the sdg1025.

[nctnico]

Well a square wave has an infinite number of harmonics so there is no signal generator to create the perfect square wave and there is no oscilloscope to view it. What is the point of all this?

[Groucho2005]

Rise time seems to stay constant on the sdg1025. About 7ns for a square, which is pretty good. But the performance starts to fall off after 1MHz @1vpp. At 25MHz the square wave is almost indistinguishable from a sine wave.

What scope did you use? Some SDG1025 screen shots taken with a reasonably fast scope:
https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg588973/?topicseen#msg588973

The 25 MHz signal is certainly no square but also far from a sinusoid.

25MHz Sine:


25MHz Square:


What it should look more like, 1MHz Square:

Without knowing the bandwidth of the scope and how you measured this (probe or a properly terminated 50 Ohm cable) it's impossible to comment.

[Armxnian]

Without knowing the bandwidth of the scope and how you measured this (probe or a properly terminated 50 Ohm cable) it's impossible to comment.

Output is terminated into 50ohms on a 100MHz scope. But it doesn't matter, because I checked beforehand with a 1.5GHz SA and the content after the 3rd harmonic barely resembles a square wave.

After 8MHz, which you can easily view on a 100MHz scope, you can see that the square starts to transform into a sine. Even after 1MHz you can see curves instead of sharply rising and falling edges. The scope does not lie. Sub 1MHz the output is great, not so much after. All of the other functions perform great, but the square suffers. IMO, if you're going to claim a "square" wave output at 25MHz, it better look something like it does in my 1MHz screen capture.

[nbritton]

Without knowing the bandwidth of the scope and how you measured this (probe or a properly terminated 50 Ohm cable) it's impossible to comment.

Judging by the screenshot he is using a Rigol DS1000Z. I see another critical problem with his measurement, on the screenshot you can see he is using a 1X probe. IIRC the Rigol probe derates in 1X mode to around 7 MHz of usable bandwidth. Furthermore, the DS1000Z doesn't have 50 Ohm termination, you would have to put the SDG1025 into Hi-Z mode and then plug the scope's 10X probe directly into the SDG1025's BNC connector.

1000Z Bandwidth:
Frequency      Amplitude
  10 MHz          0.0 dBm
100 MHz         -1.5 dBm
150 MHz         -3.0 dBm
393 MHz        -10.0 dBm
447 MHz        -20.0 dBm

The system bandwidth (scope + probe) on the DS100Z is calculated at 83.2 MHz, in the dBm measurements above I don't know if they were taken using the scope's probe or using a BNC cable, it makes a difference.

1/sqrt((1/scope bandwidth)^2+(1/probe bandwidth)^2) = system bandwidth
1/sqrt((1/100 MHz)^2+(1/150 MHz)^2) = 83.2050294338 MHz

Square Wave:
Fundamental: 25 MHz
3rd Harmonic: 75 MHz
5th Harmonic: 125 MHz
7th Harmonic: 175 MHz
9th Harmonic: 225 MHz
11th Harmonic: 275 MHz
13th Harmonic: 325 MHz

[Armxnian]

Judging by the screenshot he is using a Rigol DS1000Z. I see another critical problem with his measurement, on the screenshot you can see he is using a 1X probe. IIRC the Rigol probe derates in 1X mode to around 7 MHz of usable bandwidth. Furthermore, the DS1000Z doesn't have 50 Ohm termination, you would have to put the SDG1025 into Hi-Z mode and then plug the scope's 10X probe directly into the SDG1025's BNC connector.

I see that reading comprehension is a challenge for you.

I set the scope to 1x to show that I was using a 1Vpp signal. I'm not using a probe, I'm using a BNC cable that goes into a T-connector with one side terminated into 50ohms. You don't need to view 13 harmonics to judge the performance of a wave.

I was trying to help you make a purchase. If square wave performance was important to you I would have recommended to perhaps try something other than the SDG1025. But it seems you're more interested in finding errors in my simple setup. Good luck.

[Wuerstchenhund]

Well a square wave has an infinite number of harmonics so there is no signal generator to create the perfect square wave and there is no oscilloscope to view it.

That is absolutely true.

What is the point of all this?

I guess the point is that at higher frequencies (i.e. >10MHz) the square wave produced by the Rigol DG1000z and Siglent SDG1000 is pretty crap, but this is true for most (all?) low end AWGs and a limitation that comes with the low purchase price.

If square wave performance was important to you I would have recommended to perhaps try something other than the SDG1025.

If square wave performance is important then you have to invest a lot more money than the several hundred bucks for a low end AWG and you're probably looking into serious mid-range AWGs from Keysight or Tektronix which will set you back several grand.

As the old saying goes, you get what you pay for.

[nbritton]

Judging by the screenshot he is using a Rigol DS1000Z. I see another critical problem with his measurement, on the screenshot you can see he is using a 1X probe. IIRC the Rigol probe derates in 1X mode to around 7 MHz of usable bandwidth. Furthermore, the DS1000Z doesn't have 50 Ohm termination, you would have to put the SDG1025 into Hi-Z mode and then plug the scope's 10X probe directly into the SDG1025's BNC connector.

I see that reading comprehension is a challenge for you.

I set the scope to 1x to show that I was using a 1Vpp signal. I'm not using a probe, I'm using a BNC cable that goes into a T-connector with one side terminated into 50ohms. You don't need to view 13 harmonics to judge the performance of a wave.

I was trying to help you make a purchase. If square wave performance was important to you I would have recommended to perhaps try something other than the SDG1025. But it seems you're more interested in finding errors in my simple setup. Good luck.

Where in your original post did you state the conditions of your test? You just posted a screenshot with no words, I can't read minds. Thank you for taking the time to test this, I was simply trying to make sure the test was setup properly.