Looking for reasonable DDS or AWG or function gererator, need opinions please

[Lightages]

I currently have an old analog 2Mhz function generator that has linear and log sweep functions. It is not very stable and the distortion adjustments drift quite a bit with temperature.

I am looking to improve on this and was looking at the latest hot topic on DDS AWG generators here, the Hantek HDG2002B or the old hot topic the Hantek DDS3X25. I don't see log sweep as part of the spec of the HDG2002B and I am not sure on the DDS 3X25 neither.

So I want something that does what these basically do but I do want log sweep. I would be happy with around 25MHz but of course more for the same money or a bit more is a good thing too. I also don't mind a USB only version as I have a computer on the bench.

So what is the current state of the DDS 3X25? Can it be hacked for higher frequency still? Does it have log sweep?
The HDG2002B? Log sweep?

Another possible option if anyone has experience with it:
http://www.ebay.com/itm/24Mhz-Dual-ch-DDS-Function-Arbitrary-Waveform-Signal-Generator-sweep-Software-/121496848485?pt=LH_DefaultDomain_0&hash=item1c49c6c865

Any other options in the $300 range that are known good buys?

[Lightages]

I guess there aren't as many people interested in this as multimeters and scopes 

[Syntax_Error]

I wish I could contribute, but all I can say is that I, too am interested in reading the answers to your questions, as there is some overlap with my own. I hope your thread takes off soon.

[idpromnut]

I can confirm that the HDG2002B only does linear sweeping (although there is an option that looks like you should be able to change it between linear and some other option(s) but I cannot change it's value in the firmware that I have).

[tautech]

Siglent have recently released new firmware to get the Log Sweep correct in their SDG1000 series.
Quite a few here have these AWG's and seem to be content with them. Models to 50 MHz.
Dual output, 50 ohm or 1 M, 7 ns risetime in pulse mode.
http://www.siglent.com/ENs/pdxx.aspx?id=145&T=2&tid=16

AFAIK nobody has spent the time to attempt to hack these to higher BW's.

[toli]

I was in the same spot as you are right now just a few months ago.

I've ended up purchasing a Siglent SDG1020 (20MHz model). I'm quite happy with it. It has dual output which is nice (but CH2 is limited in amplitude to 6Vpp or so), and the UI is easy enough to learn. I think overall its a solid performer, and it feels close enough to professional for my needs. It doesn't feel as cheap as it is. The latest fw did fix the log sweep option, so it can be used for Xfer function tracing much better now.

[Yago]

I shopped for a budget FG for ages, jees was not fun.
Ebay auctions were screwed over by the same people, who never won an auction... but they managed to get everything out of my budget.
The older HP models, were just that, very old, I didn't fancy getting something so complex and possibly unreliable.

I actually forgot about the Hantek, even had it highlighted in a folder on PC! doh!

Ended up with a used Fluke/Philips PM5138, only 10Mhz, but there are a couple of models in the range.
Can't give you specifics(it does have log sweep in modulation, not used it), other than it works as expected so far, it does appear grainy on the scope, perhaps low bit depth(10bit?) and that may be something to consider.
In good condition and built like a tank (cast alloy case I think), was £100.

[radioFlash]

I have the Hantek DDS 3X25. It seems to function adequately, but I hate that it must be controlled from a PC, and the software that it comes with works, but it's not particularly good. I've been looking as well for a FG, but one with physical controls.

[apelly]

Can't remember how much my Rigol DG was, but it was pretty cheap from memory, and pretty damn flexible.

I guess you've seen Shahriar's recent review of the Siglent SDG5162 by now: http://thesignalpath.com/blogs/2014/12/08/review-comparison-and-teardown-of-siglent-sdg5162-and-rigol-dg4162-arbitary-waveform-generators/

[Lightages]

Thanks for the links guys. Unfortunately those AWGs reviewed by Shahriar are out of my price range.

[TMM]

Also consider the Rigol DG1022, though value for money is not quite as good as the others. It can only modulate/sweep on CH1 iirc.

I guess you've seen Shahriar's recent review of the Siglent SDG5162 by now: http://thesignalpath.com/blogs/2014/12/08/review-comparison-and-teardown-of-siglent-sdg5162-and-rigol-dg4162-arbitary-waveform-generators/

Too bad they are about 4 times the OP's budget.

[LaurentR]

Also consider the Rigol DG1022, though value for money is not quite as good as the others. It can only modulate/sweep on CH1 iirc.

I second that. Not nearly the incredible bargain a DS1054Z is, but a reasonable performer for the price. It's a bit dated, but that also means it can be found occasionally used on EBay for <$300, otherwise, they seem to sell really well on TEquipment (best selling function gen - $356 after discount), so you wouldn't be the only owner out there :-)

[MartyD]

Siglent have recently released new firmware to get the Log Sweep correct in their SDG1000 series.
Quite a few here have these AWG's and seem to be content with them. Models to 50 MHz.
Dual output, 50 ohm or 1 M, 7 ns risetime in pulse mode.
http://www.siglent.com/ENs/pdxx.aspx?id=145&T=2&tid=16

AFAIK nobody has spent the time to attempt to hack these to higher BW's.

In opposite to Rigol there seem to be hardware differences between the various Siglent 1000 series generators. In other peoples teardowns I could spot minor differences on the main boards, also the units >20MHz have a cooling fan. So I'm not sure they can simply be hacked to a higher bandwidth without performance impacts in the higher frequency ranges.

However, I think they offer the best bang for the buck at the moment in the entry level segment. It looks and feels like a quality product and I have been through quite a few entry level generators before I settled with the SDG. Now that the firmware has matured there's no more argument against it. If you want/need higher stability, just drop in a $25 TCXO from ebay.

If you want "hackability" you're probably better off with a Rigol product. Though I'm not too keen on that...

Marty

[Howardlong]

Be aware that the HDG2002B is incredibly buggy, especially for anything more complex that your standard non-complex waveforms: in general as a test instrument it can't really be trusted, I frequently double check the output signal with a scope to make sure it's doing what it should be. It seems they spent a reasonable amount of time on the hardware but the software team went to the pub for lunch and never came back.

[Lightages]

Another question: Does the Rigol DG1022 have the option to just take recorded waveforms from the DS1052E or DS1054Z and play them back? Is it easier to do so Rigol to Rigol than from the Rigol scope to Siglent AWG?

[ivan747]

Another question: Does the Rigol DG1022 have the option to just take recorded waveforms from the DS1052E or DS1054Z and play them back? Is it easier to do so Rigol to Rigol than from the Rigol scope to Siglent AWG?

As an owner of a DS1102E and a SDG1010, it's nightmarish 
My workflow is the following:
-Export a CSV file from the Rigol scope, not full memory depth because it takes ages and cannot be taken advantage of by the AWG
-Edit the waveform with a DECENT waveform software, Tektronix ArbExpress (this shouldn't be done, the license forbids it, but I always have my Tek scope when opening arb express, you know    -I'm looking for alternatives here). Editing, in my case, is clipping the waveform, editing its amplitude, sampling rate (ArbExpress can compress a waveform using frequency-limited interpolation) etc.
-Export from ArbExpress. The CSV format is the format used on Tektronix scopes (the CSV used for AWGs is very different and confuses Siglent's EasyWave).
-Import that into EasyWave
-Export into AWG

The main problem is shuffling around file format, as you see.

*I am not entirely sure it's going to be better from rigol to rigol*. Why? Well, in theory (I haven't checked, sorry), the formats are compatible, but, you still need to clip the waveforms, change the sample rate and such, don't you? As far as I know, this is not practical to do inside the Rigol generator (and I'm sure of that because I have seen how low resolution that grayscale display is). So you still need the computer. You save maybe 2 or 3 steps out of 7.

As a brand, I trust Rigol more than Siglent, but the Siglent is giving me twice the sampling memory and 4 times the resolution (12bit vs 14 bit).

If I was doing arbitrary waveforms all the time, then I would personally get the Rigol to keep me sane, but since I was already stretching my budget by quite a bit looking for a decent analog or DDS FG, and I wasn't originally looking for an AWG, I thought "well, I'm getting the AWG and the color LCD, that's a pretty nice bonus".

I'm still looking for a way to open the CSV (or whatever format) directly on the SDG1010 via a USB drive. If anyone can help me out, I would be thankful.

By the way, here's my review for the Siglent, if you're interested:
http://www.tequipment.net/Siglent/SDG1010/Fully-usable-AWG-at-a-very-low-price/c629c3f1890b4/

[Lightages]

Thanks for the information on using Rigol waveforms with the Siglent.

[rf-loop]

I'm still looking for a way to open the CSV (or whatever format) directly on the SDG1010 via a USB drive. If anyone can help me out, I would be thankful.

By the way, here's my review for the Siglent, if you're interested:
http://www.tequipment.net/Siglent/SDG1010/Fully-usable-AWG-at-a-very-low-price/c629c3f1890b4/

In rewiev: AM modulation only 1kHz.
This is not true. Internal AM modulation freq range is 2mHz to 20kHz (but of course user need also know basic fundamentals about modulations.)

CSV (Comma Separated Values) from USB stick to SDG. Of course it can. It is made for this.

You can write acceptable CSV even with windows notepad.

This is simply example.
(also it is good to note that in practice offset and vertical  value etc is not so much meaning (it automatically scale it but when you turn it on, these values are "default" values. Front panel you set freq, level and offset (as for all arbs))

Attached: Sine1.csv.txt
Just remove .txt save it to USB and read it with SDG1000 and use it.
If you want other waveform, just edit its data area xpos (arb memory "address") range is 0 - 16383 and nothing else.

If you manually want edit it, take care that not add any hidden control etc characters to file due to text editor.

Siglent scope CSV is compatible. Other brand waveform CSV may need modification before it can read in SDG.  (copy paste 16384 vertical values to "value" column)

I do not have now any SDG1000 what I can use for test but if I remember right it also automatically scale csv values to arb memory so that max and min values match full level in arb memory.


NOTE: in file "header" part, there is "data lenth,16384"
(it is not my mistake!  Siglent produce it.  I can not now test (not available SDG1000 for testing) what happend if repair this error in file so best keep it as it is)

[ivan747]

Well, thanks for the info. I'm still figuring out a reliable way to clip the waveforms (change their sampling rate too) and maybe adjust their amplitude to use the full resolution of the DAC.

[rf-loop]

I did some short test with oscilloscope and SDG5082.
Result was surprise. Due to very limited time visit in lab I do not have enough data at all for proof anything. Also I do not have SDG1000 what I can use for testing.

Surprise start after I made mistake and load wrong .CSV to  SDG5082 arb memory.
SDG5000 CH1 arb memory is exactly same as SDG1000 (16k)
My accidentally loaded CSV was made with oscilloscope and with 280kpoints waveform where also vertical was  not at all full display.

SDG5000 read it without any err message. Then I store it to CH1 16k arb and agen, not any err message.

Ok, then I start run this and look with scope what is coming out. Surprise (Also at this time I can see that this was not wavefform what I think I load to SDG5000, instead it was this 280kpoints CSV): It was vertically rescaled up (p-p to match full scale) but more surprise. It was also horizontally scaled to match this memory and waveform was horizontally just same what is was on the scope screen before save it as CSV.  It was not cutted, it was scaled to 16k memory.

Due to this surprise I did other. I take sinewave and oscilloscope adjusted so that display (wfm lenght) just one cycle and now 14k memory. Then save it as CSV to USB.   Then USB to SDG5000 and there to CH1 16k memory. It read it just without err message. Result same. Run it back to scope and vertical was agen up scaled and horizontally output was just same full cycle of sine.

Only what I now think is that this "feeature" need really test and document better. Also I hope I can do same with SDG1000 later.

Also, when I look these oscilloscope produced CSV files, there header information is different.
Previously this attachment .CSV was made with EasyWave software.

I have not seen any good documents how SDG1000 and SDG5000 really handle .CSV waveform files.
If it really can aautomatically adjust lenght and height of waveform it may be useful feature, specially if it is documented how and what it really do in SDG1000 and in SDG5000.

Add: 16k waveform Arb. .CSV data points can be what ever from 2 points to full 16k (16384). What ever amount of points SDG5000 play it as one Arb cycle. (btw, if Arb cycle have 10 cycle of sine then 10MHz Arb freq setting give 100MHz out and so on).  With today some more tests it is clear that SDG1000 and SDG5000  .CSV file things need investigate more or get detailed information from Siglent.

Later I will try with Rigol DS1000Z waveform .CSV if it is possible

[dom0]

Ended up with a used Fluke/Philips PM5138, only 10Mhz, but there are a couple of models in the range.
Can't give you specifics(it does have log sweep in modulation, not used it), other than it works as expected so far, it does appear grainy on the scope, perhaps low bit depth(10bit?) and that may be something to consider.
In good condition and built like a tank (cast alloy case I think), was £100.

The PM5138 has quite a few sister models. The lowest endish thing has 2 MHz max and the highest one goes to 20 or 25. These are good generators generally, nice to use. 10 bit is correct, I think.

Shameless self-quote from another thread:

I wouldn't focus on the very-well known function gens on the used market, since they are popular they are more expensive.

Look for not so well known models.  For example, HP 3312a, HP 3324A, Philips/Fluke generators (e.g. pm5138, pm5139, pm5136), there are also always some Wavetek generators around, but I found it hard to get solid specs on them and skipped them. That's a very long list.

Patience pays, I bought my fg (50 MHz Philips) for c. 150 $ (excluding shipping etc.), but I had to wait a few months for a good offer... OTOH it might be hard to get spare parts for "unpopular" generators. But then again, spare parts are usually rather expensive... especially for HP, Tek, ... devices.

When doing audio stuff it might be worth to look for an additional sine oscillator or some fg with lower distortion numbers than general-purpose fun-gens, since those usually have rather high distortion (most synthesized fgs from the 80s-90s are in the 0.1-0.5 % ballpark). Building a good, solid, low distortion sine oscillator is btw. a nice project to do. I recommend it.

Be careful with the HP 3325A as only the sine and square go to an appreciable frequency.  The triangles and ramps only go up to something like 11 kHz.

This is a common limitation of all early synthesized generators I am aware of. Generally, always fetch and read the specs for gear if you are about to spend an appreciable amount of money.

There is also the 8112A/8116A series. But they're not overly good generators IMHO. They're based on some very tricky analog oscillators, controlled by DACs, and have rather poor stability and precision. In this regard they are not better than analog generators.

There is also the 8165A, a very early synthesizer. I don't know anything more about it.

[Lightages]

Thanks for all the replies. It looks like if I want to play it safe and get something modern, the Siglent SDG1025 would be a good buy for me. Higher frequency output on square and other waveforms than the Rigol DG and good build quality.

If I want to take a chance and take one for the team to see what is like there is this still:
http://www.ebay.com/itm/24Mhz-Dual-ch-DDS-Function-Arbitrary-Waveform-Signal-Generator-sweep-Software-/121478483178?pt=LH_DefaultDomain_0&hash=item1c48ae8cea&autorefresh=true
It could be OK, or real bad.

I would also still consider the Hantek DDS3X25 but there seems to be no recent activity on it here so I am not sure of its status.

[rf-loop]

Thanks for all the replies. It looks like if I want to play it safe and get something modern, the Siglent SDG1025 would be a good buy for me. Higher frequency output on square and other waveforms than the Rigol DG and good build quality.

Siglent frequency limits by different models are mainly in Sinewave.  In square wave function it still also internally works producing sinewave what is now not coming out. (limited in 1050 model to 25MHz, in other models same max as sine). It is used internally to produce square wave using comparator method (very common method  in many old and also new function generators). It have advantages and disadvantages.  With higher frequencies it give less time jitter.

Pulse, what can also use mostly as square wave, is produced  by DAC's as also other waveforms.

When sinewave is limited to 10, 20, 25MHs (1025 model 25MHz) there still can produce 50MHz (and over) sinewave (quality is not as perfect what it is with sinewave function). It can simply do with math equation using Arb. (Equation draw with EasyWave)
10 cycles sine with arb. Setting Arb then for 5MHz and output is 50MHz.  here 5 cycles:
10*(sin(5*x))     here ten is for 20V pp range (-10 to +10V) (property setting 20Vpp)

Or something other, example this: 10*(sin(2*(sin(2*(sin(3*x))))))
(yes it is (simplified version) how square wave looks when we reduce amount of harmonics.  There is 3 cycles )  Just open EasyWave and write equation (draw --> Equation draw) and it draw it for you and save file (.csv) to USB stick and run with SDG.

Or of course also some complex things.

SDG1000 models sinewave quality is also quite good in this price class. Here can find some tests with spectrum analyzer.

It need also note that today product versions do not have old time square wave glitch issue.

Newest FW give also:

- Channels can lock so that slave channel follow when master channel frequency is adjusted (ser select which one is master). There can also set offset frequency between channels.

- Ouput impedance (better say Load impedance setting, because output itself is always 50ohm)
Old version have selections Hi-Z and 50ohm (for display right level dependent if load is 50ohm or Hi-Z. 
Now there can set also other loads.  Example if signal is going to 600ohm input, just set output for  600ohm and agen it display right signal level. (load impedance can set wide range. If some amplifier input is 10k then just set 10k...etc )

(LOG sweep up/down is finally also repaired, now it works right. (old works right only for 1 decade start stop range.)

[ivan747]

RF-loop. I will be experimenting with this on the SDG1000

[dom0]

(load impedance can set wide range. If some amplifier input is 10k then just set 10k...etc )

I doubt the amplitude accuracy of most generators is high enough that such a high impedance load matters...