Function Generator purchase

[Achilles]

Well, I have a Siglent SDG1025. I preferred that one instead of the Rigol because you load the Arb waveform via usb-stick and it's just a CSV table. So you can edit it in wordpad, excel or whatever.
The software itself isn't that great, but I just used it once for testing and generator a blank CSV (for the headers and so on) which I use now to build a waveform. I have a Mac with virtual windows and after a bit of drivers-searching and installing the Generator is found by the software and can also be loaded as NI-VISA device in Labview. Price was 280Euro shipped.

That's the good parts....but there were some here who had problems with jitter and the service is....well......chinese, but not as bad as my experience with hanteks service.

Isn't the hantek a USB device? I would really prefer a stand-alone model. You'll never know if you want to switch to linux or whatever one day or if you'll get good drivers in time after a new Windows release. Also, you always need to have a computer around when you work with it. If it's somehow possible I would suggest to pay a bit extra for a stand-alone generator.
If you don't want/need arbitrary functions you should buy a nice old philips, hp or whatever design. They are usually a lot cheaper and will work just fine.

[smackaay]

I don't know what the big deal is. I bought a Rigol DG2021. The unit works nicely and the software works great on my win7 machine.

[clifford]

If you're able to write very simple programs, in almost any language, then waveforms can be easily be generated or converted to a format that the generator can read.

A waveform (RDF file) is after all just 4096 16-bit values, ranging from 0 to 16383 (0 to 1023 for channel 2).

I see it as a non-problem. Others think it's the end of the world

I fully agree: it is a non-problem. just a short additional note:

The values always range from 0 to 16383, for the 2nd channel the 4 least significant bits are ignored. So you can use the same RDF file for both channels.

(As it's always 4096 values and for the 2nd channel only every 4th value is used.)

The values are stored in little-endian byte ordering.

If you look at files generated by the UltraWave software and the DG1022 itself, then you will see that they are using the two MSBs in each 16 bit word for some additional flags. I have not found out what is stored there. But it seems to be save to ignore the two MSBs when reading RDF files and just set them to zero when writing RDF files.

As in 99% of all my cases my arbitrary waveforms come from some kind of software anyway, I simply write out such RDF files directly and do not bother about the quality of the PC software provided by RIGOL.

[clifford]

If you look at files generated by the UltraWave software and the DG1022 itself, then you will see that they are using the two MSBs in each 16 bit word for some additional flags. I have not found out what is stored there. But it seems to be save to ignore the two MSBs when reading RDF files and just set them to zero when writing RDF files.

Update: The MSB marks the user-defined points. A point where the MSB is cleared is a point that is the result of an interpolation between user-defined points. This bit is only relevant when you want to edit a previously stored waveform. So it probably would be cleaner to always set this bit than to always clear it when generating your own waveform files.

Only one more bit to go: I still have no idea what the second most significant bit encodes.