Best arbitrary waveform generator for around $500

[dadler]

If you want fast edges, you have to pay for them. There aren't really any free lunches in AWGs.

Square wave on my 33522B meets the rise time specs of 8.4ns (fixed, based on the implementation of the generator) all the way to 30MHz. But this is an expensive generator, and 8.4ns isn't even that fast.

For the money, these low-end generators from Siglent and Rigol do offer good performance. If you need** fast edges, buy a higher end generator.

**I can't imagine anyone buying an entry level generator "needs" super-fast edges.

[Wuerstchenhund]

Well, the SDG1000 is a toy, as a reliable tool it's really crap. The only thing it's really good at is to create a jittery signal, which is nice for playing around with the jitter analysis tools of my scope, but that's about it.

The money I pad for the SDG1020 was not money well spent. So even if I had 'bricked' the damn thing this would not have been a huge loss (and I'm sure even then it could be revived by removing the flash and flashing it externally). So there was nothing bold in what I did.

Why are you giving the SDG1000 praise in this thread and bashing in the other thread?

You're pulling stuff out of context. For a start, the posting you quoted is from 2013, and at that time some SDG1000 (including mine) were suffering from excessive jitter. Siglent later published instructions for a hardware fix, and that cured the problem. This most certainly isn't an issue for any SDG1000 bought today (or even in 2014).

The other thing you would probably only understand if you read all my postings regarding the SDG1000 and AWGs in general is that my requirements even for personal T&M gear are probably much more stringent than for the average hobbyist, and bluntly spoken that simply means that stuff in the low end class is unlikely to be anything more than a toy for me. I bought the SDG1020 mostly out of curiosity of what you can get for the few hundred Pounds the thing did cost at the time, which is why when I wrote the posting in 2013 I didn't really care if my experiment with the firmware would brick my SDG1020 or not. Well, it didn't, and I later fixed the jitter problem, and my old SDG1020 now serves with a happy beginner.

However, I'm also well aware that my own requirements aren't necessarily the same ones other people have, plus I know pretty well what you can get for your money. This also means that just because the SDG1000 didn't satisfy my requirements (which it was unlikely to do anyways, even the more expensive Rigol DG1062z didn't do that!) doesn't mean it won't do for others. Considering what else is out there in this bottom-of-the-barrel price range, the SDG1000 is actually a pretty good offering.

So yeah, I wouldn't buy a Siglent SDG1000 (or a Rigol DG1000z) again, but that doesn't mean they're also a bad purchase for everyone else!

[Groucho2005]

If you need** fast edges, buy a higher end generator.

Or a dedicated pulse generator. Although these are getting rare.

[nctnico]

So yeah, I wouldn't buy a Siglent SDG1000 again, but that doesn't mean they're also a bad purchase for everyone else!

I'm quite happy with mine but my typical use isn't very demanding on a function generator. I didn't even bother to fix the jitter problem and I don't take the AWG function serious. If I ever find myself in a situation where I need something better I simply buy something better.

[nbritton]

I'm mulling over the idea of going with the Hantek HDG2002B. This is a ~$300 5 MHz AWG that can be hacked to 100 MHz. Its datasheet says it's 16 bit with a 250 MSa/s and 64M memory depth. From what I understand its square wave output is similar to the units we have talked about in this thread, which is to say it's mediocre. However, its sine wave output appears good all the way up to 100 MHz. If I need high frequency square waves I'm thinking I could build a square wave generator circuit that is clocked using the sine wave output from the AWG.

Also it apparently has 16 digital output channels, but I'm perplexed as to how these work... is it just a square wave output?

Here is the datasheet for it, are the other specs comparable to the other units we discussed?: http://hantek.com/en/ProductDetail_12_149.html

https://www.eevblog.com/forum/testgear/hantek-hdg2002b-awg-5mhz-or-100mhz-let's-see!/

[nbritton]

Which AWGs can I program with mathematical functions? i.g. algebra, trigonometry, and calculus.

Edit: Also the HDG2002B runs Linux, does anyone know if I can drive the signal gen directly from a shell prompt or python script?

Edit 2: Do you know of a signal gen that can be driven directly from a shell script?

[nbritton]

Why can't you just wrap a USB interface around a 18-bit DAC and call it done? I'm not understanding why these AWGs are so expensive.

[Wuerstchenhund]

I'm mulling over the idea of going with the Hantek HDG2002B. This is a ~$300 5 MHz AWG that can be hacked to 100 MHz. Its datasheet says it's 16 bit with a 250 MSa/s and 64M memory depth. From what I understand its square wave output is similar to the units we have talked about in this thread, which is to say it's mediocre. However, its sine wave output appears good all the way up to 100 MHz.

It also seems to be highly unreliable, plus you don't even know how good its specs are above the guranteed 5MHz. Just because it can be made to output a something that looks like a 100MHz sine doesn't necessarily mean it's a true sine wave with low jitter. But I guess it depends on how much reliability you want and if you can afford to throw away $300.

If I need high frequency square waves I'm thinking I could build a square wave generator circuit that is clocked using the sine wave output from the AWG.

Don't underestimate the amount of effort required to generate a good square wave that isn't crap at with these low end AWGs.

Which AWGs can I program with mathematical functions? i.g. algebra, trigonometry, and calculus.

Any of them, just not on the device directly. An AWG reproduces a waveform it receives in binary (or proprietary) form, which aside from a number of built-in predefined arbitrary waveforms is usually created outside the instrument, i.e. on a PC running a waveform editor (which with cheap AWGs is also crap), or generated from MathLab or Excel.

Edit 2: Do you know of a signal gen that can be driven directly from a shell script?

Shell on what? On an external PC? Yes, you should be able do that with any AWG that comes with a remote control interface (LAN, USB, GPIB) and which supports some standard communication protocol like SCPI.

Why can't you just wrap a USB interface around a 18-bit DAC and call it done?

Sure you can, if you don't care about jitter or other distortions.

I'm not understanding why these AWGs are so expensive.

They're expensive because generating valid waveforms is a pretty complicated thing, even more so for square waves. Just converting some bitstream into an analog signal isn't enough.

[wojt]

I'm mulling over the idea of going with the Hantek HDG2002B. This is a ~$300 5 MHz AWG that can be hacked to 100 MHz. Its datasheet says it's 16 bit with a 250 MSa/s and 64M memory depth. From what I understand its square wave output is similar to the units we have talked about in this thread, which is to say it's mediocre. However, its sine wave output appears good all the way up to 100 MHz.

It also seems to be highly unreliable, plus you don't even know how good its specs are above the guranteed 5MHz. Just because it can be made to output a something that looks like a 100MHz sine doesn't necessarily mean it's a true sine wave with low jitter. But I guess it depends on how much reliability you want and if you can afford to throw away $300.

People who hacked it to 100 MHz have also prepared calibration for this range. I have done it on my HDG and it works as good as I can measure and need.

[nctnico]

That HDG2002B looks mighty nice when hacked to 100MHz. Can both channels do 20Vpp in the low frequency range?

Edit: never mind. Seems the firmware is crap. 

[Howardlong]

That HDG2002B looks mighty nice when hacked to 100MHz. Can both channels do 20Vpp in the low frequency range?

Edit: never mind. Seems the firmware is crap. 

It is. Although if you're into aftermarket adjustments, there's plenty to tinker with. As well as liberating it, I added an external serial port to mine, ethernet magnetics and RJ45, and a high speed frequency divider for use as a counter into the GHz. The hardware's not too bad, but you can't easily polish the turd that is the software.

[jhufford]

Is it possible to make a relatively cheap circuit that can take a signal input and re-output it with really fast edges? In other words, if the edges aren't sharp enough, can you run it through another circuit to sharpen them up? My initial guess is this would work, but might make for some jitter due to imperfect triggering of the edge shaper. So maybe lower freq square waves (jitter is low % of fundamental) but with super sharp edges...  Anyone?

[nctnico]

Something like that is called a comparator. The low end ones have an open collector output, the fast ones have a push-pull output. Another option is to use an LVDS to CMOS receiver.