Decent modern Signal/waveform Generator here in early 2024

[DaneLaw]

Are the fan noise an issue on the Siglents SDG2xxxX line, as it is a potent fellow with 1.2GSa/s, - way higher than both the SDG1000X (150MSsa) series and Rigols DG800 (125MSa) & 900 series (250MSa) as unsteady fan-noise, on Rigols latest sig gen are killing it for me & not what I had hoped & hard to love, not least after their "fanless" DG800/900 focus.
And likely anybody's guess when Siglent revamps these model segments.

Been watching a few reviews, and I can not even sense the fan on the SDG2042X Siglents, seems weirdly quiet even in videos where the user is sitting close to it...
Have noticed that many of the users on the forum purchased the SDG1032X or the bigger brother SDG2042x, makes sense, seems like a very solid build, from the teardowns and very beefy fundamentals in this price segment..

the SDG2042x still seems to go for around 600 euros, here in northern Europe, while abroad Amazon US it's a tad over 400 euro, been trying to see if the age late 2015/early 2016 signifies in any shortcomings, but then again its not like the wheel been reinvented since, many of the features are pretty fundamental, but still, some new adv features. do see daylight from time to time, and can meet a bottleneck in older hardware.
all the benefits are obvious., often more mature os, less bugs, a huge user back-catalog of info you can lean on and debate with etc..

Any comparable videos or articles, between Siglent SDG1000/2000 series and Rigol DG800/900, to give an insight into pros & cons..?

[tautech]

SDG1000X has a different output stage than the earlier 2000X models which provides faster rise time than its bigger brother.
Apart from that, lower BW and 14 vs 16bit they are much the same.

Otherwise we need go to SDG6000X to get a faster edge, higher BW etc.

[2N3055]

Are the fan noise an issue on the Siglents SDG2xxxX line, as it is a potent fellow with 1.2GSa/s, - way higher than both the SDG1000X (150MSsa) series and Rigols DG800 (125MSa) & 900 series (250MSa) as unsteady fan-noise, on Rigols latest sig gen are killing it for me & not what I had hoped & hard to love, not least after their "fanless" DG800/900 focus.
And likely anybody's guess when Siglent revamps these model segments.

Been watching a few reviews, and I can not even sense the fan on the SDG2042X Siglents, seems weirdly quiet even in videos where the user is sitting close to it...
Have noticed that many of the users on the forum purchased the SDG1032X or the bigger brother SDG2042x, makes sense, seems like a very solid build, from the teardowns and very beefy fundamentals in this price segment..

the SDG2042x still seems to go for around 600 euros, here in northern Europe, while abroad Amazon US it's a tad over 400 euro, been trying to see if the age late 2015/early 2016 signifies in any shortcomings, but then again its not like the wheel been reinvented since, many of the features are pretty fundamental, but still, some new adv features. do see daylight from time to time, and can meet a bottleneck in older hardware.
all the benefits are obvious., often more mature os, less bugs, a huge user back-catalog of info you can lean on and debate with etc..

Any comparable videos or articles, between Siglent SDG1000/2000 series and Rigol DG800/900, to give an insight into pros & cons..?

Just a note, make sure to call them 1000X and 2000X because there is older model 1000 without X suffix.

In a nutshell ?

SDG1000X has very good pulse creation specs. AWG memory is on smaller side (16k).
SDG2000X has better analog performance (sinewave distortion), better level specs, goes to 120 MHz. AWG memory is much longer.

For general purpose work 1000X is sufficient. For pulse work, it is better.
2000X has better stability and lower distortion. It is better for analog and AWG work.

[DaneLaw]

One thing to note is that Rigols are hackable but in a way that after "upgrade" they are not fully adjusted in whole new BW.
Rigol is not doing that. After "upgrade", "liberated" BW is not properly leveled. There are topics here on that...
Funny enough, older Rigol DG1000Z was fully calibrated and adjusted so that one worked properly even after "upgrade".

Which Rigol model are you referring to?
the current lineup? the "not fully adjusted", meaning a lack of linearity in the liberated range?? how sever?

[2N3055]

One thing to note is that Rigols are hackable but in a way that after "upgrade" they are not fully adjusted in whole new BW.
Rigol is not doing that. After "upgrade", "liberated" BW is not properly leveled. There are topics here on that...
Funny enough, older Rigol DG1000Z was fully calibrated and adjusted so that one worked properly even after "upgrade".

Which Rigol model are you referring to?
the current lineup? the "not fully adjusted", meaning a lack of linearity in the liberated range?? how sever?

New DG800/900 series.. I mean, new, it has been few years now, since 2018...
Not lack of linearity, but output levels in "liberated" ranges drops off. By a lot..
There are numerous other problems too, software being buggy...
There is a hack topic and it is all there.

Shame. I like old DS1000Z, for what it is, and have kept it, despite having SDG6000X and SDG7000A available.

In meantime Rigol released new DGxxxPro series of AWG with yet another brand new OS and interface.
New system to again debug from the scratch, without ever committing enough resources to fully polish and debug old ones...

[Mechatrommer]

SDG1000X is a piece of entry level "proper"  lab equipment. Square wave edges on UTG962 are 16 ns, and 4ns for SDG1000X.

is that matters? rise time? synch output of UTG900 is 963ps (~1ns)... other facts, GUI doesnt go to 0 when twist knob from 1MHz, it will go cleverly down to 900kHz, fwiw...
https://www.eevblog.com/forum/testgear/uni-t-utg932utg962-200msas-function-arbitrary-waveform-generator-220394/msg2927402/#msg2927402

[cncjerry]

Only issue with the Siglent 1032x, which is upgraded in my case, is the frequency resolution isn't as expected.  I need a 10,000,001hz signal and even with a common reference, it is measured as 10,000,000.988hz.  I think I read that the resolution is .023hz.  I have other synthesizers, some over 30yrs old, that get a lot closer with resolution to .001hz

[nctnico]

Only issue with the Siglent 1032x, which is upgraded in my case, is the frequency resolution isn't as expected.  I need a 10,000,001hz signal and even with a common reference, it is measured as 10,000,000.988hz.  I think I read that the resolution is .023hz.  I have other synthesizers, some over 30yrs old, that get a lot closer with resolution to .001hz

That is the 'problem' with these low cost generators. I have tested various units from several brands but they all have the same problem; the DDS resolution is not very high so the accuracy of the frequency setting is not very high despite the units touting a setting resolution of 1uHz. Unfortunately the number of significant digits (and thus the accuracy) for the frequency is often not specified in the datasheet. If you need precise timing performance, you need to look at generators from Keysight or Tektronix (AFG31000 series for example).

[2N3055]

Only issue with the Siglent 1032x, which is upgraded in my case, is the frequency resolution isn't as expected.  I need a 10,000,001hz signal and even with a common reference, it is measured as 10,000,000.988hz.  I think I read that the resolution is .023hz.  I have other synthesizers, some over 30yrs old, that get a lot closer with resolution to .001hz

That is because it is not a synthesizer, it is NCO.
And none of modern AWG will be synthesizers in classical sense with PLL dividers and such.
If there are some, you bet it will be obviously stated in a datasheet.

Data you stated would mean frequency setting resolution of 23 ppb....
Is that not good enough?
For 340 €?

[DaneLaw]

One thing to note is that Rigols are hackable but in a way that after "upgrade" they are not fully adjusted in whole new BW.
Rigol is not doing that. After "upgrade", "liberated" BW is not properly leveled. There are topics here on that...
Funny enough, older Rigol DG1000Z was fully calibrated and adjusted so that one worked properly even after "upgrade".

Which Rigol model are you referring to?
the current lineup? the "not fully adjusted", meaning a lack of linearity in the liberated range?? how sever?

New DG800/900 series.. I mean, new, it has been few years now, since 2018...
Not lack of linearity, but output levels in "liberated" ranges drops off. By a lot..
There are numerous other problems too, software being buggy...
There is a hack topic and it is all there.

Shame. I like old DS1000Z, for what it is, and have kept it, despite having SDG6000X and SDG7000A available.

In meantime Rigol released new DGxxxPro series of AWG with yet another brand new OS and interface.
New system to again debug from the scratch, without ever committing enough resources to fully polish and debug old ones...

I see, yes current lineup DG800 & DG900 which are these fanless units from Rigol with a quite distinct OS-theme that doesn't look anywhere like their new "Pro" models interface

I am familiar with the new Rigol DGxxxPro-line and its certain fundamental issues here the last few weeks in use, which makes me wanna look for alternatives and gain an overview' when it comes to a modern decent signal gen in early 2024.. - and some insight into which way people would go in 2024 if they were in the market for one.
Mostly 3 other models seem obvious SDG1032X and SDG2042X from Siglent and the entry Rigol DG811 as it looks like that  Rigol entry-unit are liberateable to higher revisions, though have to look into what extent the output drops, but all of them are AC-based and for good reason, hence stationary use.

After watching both Defpom SDG1032X video that tautec loan'ed him and Afrotechmods solid walkthru of the flagship SDG6052X and Dave's SDG2120X video where he compared it to the older SDG5082 and also the bigger Rigol DG4xxx that blackdog displayed risetime against (thx).. I'm starting to get my heading and the pros & cons between the models.

[nctnico]

Only issue with the Siglent 1032x, which is upgraded in my case, is the frequency resolution isn't as expected.  I need a 10,000,001hz signal and even with a common reference, it is measured as 10,000,000.988hz.  I think I read that the resolution is .023hz.  I have other synthesizers, some over 30yrs old, that get a lot closer with resolution to .001hz

That is because it is not a synthesizer, it is NCO.
And none of modern AWG will be synthesizers in classical sense with PLL dividers and such.
If there are some, you bet it will be obviously stated in a datasheet.

Data you stated would mean frequency setting resolution of 23 ppb....

That is only one sample point. Without knowing the guaranteed significant digits for the frequency setting, you don't know how big the error is worst case. In the end the DDS has a finite number of bits; the error between wanted and actual frequency depends on that and this error can be much bigger than 23ppb. It is like a divider based baudrate generator. Some frequencies can be created perfectly, other frequencies have a significant error.

[2N3055]

Only issue with the Siglent 1032x, which is upgraded in my case, is the frequency resolution isn't as expected.  I need a 10,000,001hz signal and even with a common reference, it is measured as 10,000,000.988hz.  I think I read that the resolution is .023hz.  I have other synthesizers, some over 30yrs old, that get a lot closer with resolution to .001hz

That is because it is not a synthesizer, it is NCO.
And none of modern AWG will be synthesizers in classical sense with PLL dividers and such.
If there are some, you bet it will be obviously stated in a datasheet.

Data you stated would mean frequency setting resolution of 23 ppb....

That is only one sample point. Without knowing the guaranteed significant digits for the frequency setting, you don't know how big the error is worst case. In the end the DDS has a finite number of bits; the error between wanted and actual frequency depends on that and this error can be much bigger than 23ppb. It is like a divider based baudrate generator. Some frequencies can be created perfectly, other frequencies have a significant error.

Actually, just quickly tried..

I can set 10ppb of set frequency at 10 Mhz and 100Mhz. Dead accurately. That is 10 µHz resolution at 10 Mhz.
I typed in 10.000 000 43 MHz and got exactly that. I cannot go better than that at this moment, 10 digits is limit of counter i have here.

Later I will check in more detail...

For Tektronix you have I see specification of: Frequency accuracy ±10^-6 of setting (all except ARB)

It is obviously more complicated than simple number.

But my point is : those are set frequency resolutions in ppb.... Why is that an issue on 340 € AWG with +- 25 ppm initial frequency accuracy...

[DaneLaw]

How accurate are these 10Mhz reference in/out sources on entry test-gear, like modern DDS-based signal generators.

Got a reading that looks like +40Hz by taking the 10Mhz in/out up the Rigol unit's own counter., is that more or less what to expect on entry units in the sub 300 price cat or more the counters limitasions'

[nctnico]

How accurate are these 10Mhz reference in/out sources on entry test-gear, like modern DDS-based signal generators.

Got a reading that looks like +40Hz by taking the 10Mhz in/out up the Rigol unit's own counter., is that more or less what to expect on entry units in the sub 300 price cat or more the counters limitasions'

It could be the counter, it could be the generator. A quick test for generators with two or more channels is to have it output 2 frequencies at exact multiples but far apart. Like 1Hz and 10MHz. Then trigger an oscilloscope on the lower frequency. Most of the sub 1000 euro generators I tested show a drift on the higher frequency indicating there is a frequency offset between them. Also locking to a 10MHz source properly is something you shouldn't expect to work flawlessly (as in no frequency offset).

It doesn't mean low cost function generators are bad but you have to be aware of their limitations where it comes to frequency accuracy (which -again- is typically not specified on sub 1000 euro generators).

[2N3055]

How accurate are these 10Mhz reference in/out sources on entry test-gear, like modern DDS-based signal generators.

Got a reading that looks like +40Hz by taking the 10Mhz in/out up the Rigol unit's own counter., is that more or less what to expect on entry units in the sub 300 price cat or more the counters limitasions'

It could be the counter, it could be the generator. A quick test for generators with two or more channels is to have it output 2 frequencies at exact multiples but far apart. Like 1Hz and 10MHz. Then trigger an oscilloscope on the lower frequency. Most of the sub 1000 euro generators I tested show a drift on the higher frequency indicating there is a frequency offset between them. Also locking to a 10MHz source properly is something you shouldn't expect to work flawlessly (as in no frequency offset).

It doesn't mean low cost function generators are bad but you have to be aware of their limitations where it comes to frequency accuracy (which -again- is typically not specified on sub 1000 euro generators).

Synthesized frequency resolution is usually specified on most of them to the point of being better than frequency stability of timebase (initial accuracy, drift, tempco) which will be largest contributor to frequency accuracy..

But, he is connecting AWG internal 10MHz ref out to AWG internal counter, that is running from the same reference I presume... It should count exactly 10.000000 or maybe one last count up down... It is self referenced.

[nctnico]

How accurate are these 10Mhz reference in/out sources on entry test-gear, like modern DDS-based signal generators.

Got a reading that looks like +40Hz by taking the 10Mhz in/out up the Rigol unit's own counter., is that more or less what to expect on entry units in the sub 300 price cat or more the counters limitasions'

It could be the counter, it could be the generator. A quick test for generators with two or more channels is to have it output 2 frequencies at exact multiples but far apart. Like 1Hz and 10MHz. Then trigger an oscilloscope on the lower frequency. Most of the sub 1000 euro generators I tested show a drift on the higher frequency indicating there is a frequency offset between them. Also locking to a 10MHz source properly is something you shouldn't expect to work flawlessly (as in no frequency offset).

It doesn't mean low cost function generators are bad but you have to be aware of their limitations where it comes to frequency accuracy (which -again- is typically not specified on sub 1000 euro generators).

Synthesized frequency resolution is usually specified on most of them to the point of being better than frequency stability of timebase (initial accuracy, drift, tempco) which will be largest contributor to frequency accuracy..

But, he is connecting AWG internal 10MHz ref out to AWG internal counter, that is running from the same reference I presume... It should count exactly 10.000000 or maybe one last count up down... It is self referenced.

That is the point I'm trying to get across here: On many sub 1000 euro (ballpark figure) generators you won't get exactly 10MHz. Or exactly 1Hz even when referenced from an external source. And once again, resolution is not accuracy! You may be able to set a frequency in 1uHz units but the generator doesn't have the internal divider/multiplier resolution to achieve such an accurate setting. The test I outlined by setting two frequencies is easy to do without needing external references.

[rhb]

FWIW  I got burned by an early FY6600 and bought a used 33622A from Keysight's ebay store.  It is awesome, but came at a very steep price.  However, no regrets despite the $3500 price.  At $9k new it's a bit much.  At .03% THD (measured at 0.028% by my 339A) and < 1 ps jitter (verified against a GPSDO) it's my benchmark signal source for HF.

Despite swearing I'd never buy another F***Tech product I broke down and bought an FY6900.  In head to head testing the most serious deficit is the FY6900 reference XO.  With 1 ppb OXCOs for <$25, that's easy to fix.  On instruments with external 10 MHz inputs, you just need the OXCO in a suitable enclosure.  Ironically, my most expensive HPAK purchase is the only one without an OXCO.

The waveform distortion outside the 0.5001-5.0000 V range may be more difficult to fix.  I think an op amp swap cures that, but I may not bother as most of the time I want an AWG it's in that voltage range.  I have plenty of step attenuators to go lower.  And no use case for higher.  Though if I did a 33 dBm 1-930 MHz Chinese amplifier ought to do under 100 MHz.  The two I bought are flat to <1 dB according to my 8560A.

Both the 33622A and FY6900 UIs annoy me, but in different ways.  The motivation for the F***Elec is putting together the lowest cost full RF bench for HF ham bands simply to see what's possible.  Not being an EE, my access to suitable test kit was non-existent until a few years ago.  Now it's a mid '90s >$700k list suite.  But well over 30x what I think one can do today with reasonable effort.  The nanoVNA & $20 LCR/transistor tester and progeny have had  major impacts.

I think nctnico is spot on that it's more complicated, i.e. more factors to consider, than the datasheets indicate.

Have Fun!
Reg

[2N3055]

How accurate are these 10Mhz reference in/out sources on entry test-gear, like modern DDS-based signal generators.

Got a reading that looks like +40Hz by taking the 10Mhz in/out up the Rigol unit's own counter., is that more or less what to expect on entry units in the sub 300 price cat or more the counters limitasions'

It could be the counter, it could be the generator. A quick test for generators with two or more channels is to have it output 2 frequencies at exact multiples but far apart. Like 1Hz and 10MHz. Then trigger an oscilloscope on the lower frequency. Most of the sub 1000 euro generators I tested show a drift on the higher frequency indicating there is a frequency offset between them. Also locking to a 10MHz source properly is something you shouldn't expect to work flawlessly (as in no frequency offset).

It doesn't mean low cost function generators are bad but you have to be aware of their limitations where it comes to frequency accuracy (which -again- is typically not specified on sub 1000 euro generators).

Synthesized frequency resolution is usually specified on most of them to the point of being better than frequency stability of timebase (initial accuracy, drift, tempco) which will be largest contributor to frequency accuracy..

But, he is connecting AWG internal 10MHz ref out to AWG internal counter, that is running from the same reference I presume... It should count exactly 10.000000 or maybe one last count up down... It is self referenced.

That is the point I'm trying to get across here: On many sub 1000 euro (ballpark figure) generators you won't get exactly 10MHz. Or exactly 1Hz even when referenced from an external source. And once again, resolution is not accuracy! You may be able to set a frequency in 1uHz units but the generator doesn't have the internal divider/multiplier resolution to achieve such an accurate setting. The test I outlined by setting two frequencies is easy to do without needing external references.

My Rigol DG1000Z is not here at the moment.

But I know that SDG6000X  creates exactly 133.3333334 MHz when I set to it, for instance.
I took counter, set its reference to external and slaved it to 10MHz out from AWG.
That takes AWG and counter timebase out of equation, so we are measuring how accurate is AWG frequency synthesis accuracy and resolution.
Best counter I have is 10 digits and it was showing set frequency on AWG accurate to 10 digits.

10 digits accurate synthesis is 1ppb (0.001ppm) accuracy and resolution.
How many digits you expect to be there? 12, 15 ?  parts per trillion? parts per quadrillion?

It is an general purpose AWG. It is not rubidium frequency standard...

[Detlev]

I set 10,000,001Hz here on the SDG6022 and that's what I get at the output (see photo).

Both devices are synchronized to a Leo GPSDO

The frequency counter has a gate time of 1s and average 100 measurements. The statistics are also interesting here.

[Detlev]

My Rigol DG1000Z is not here at the moment.

But I know that SDG6000X  creates exactly 133.3333334 MHz when I set to it, for instance.
I took counter, set its reference to external and slaved it to 10MHz out from AWG.
That takes AWG and counter timebase out of equation, so we are measuring how accurate is AWG frequency synthesis accuracy and resolution.
Best counter I have is 10 digits and it was showing set frequency on AWG accurate to 10 digits.

10 digits accurate synthesis is 1ppb (0.001ppm) accuracy and resolution.
How many digits you expect to be there? 12, 15 ?  parts per trillion? parts per quadrillion?

It is an general purpose AWG. It is not rubidium frequency standard...

Yes 😎

[Detlev]

here again with 10.000 000 001MHz... count the zeros 🤣

I don't think it's bad for a hobby device 👍

[RoGeorge]

count the zeros 🤣

Yes, that's annoying, close to unusable.  There is still space left in the Siglent screen, so why they didn't add thousand separators?  Didn't they notice how unreadable that is?

How did they count the zeroes during developing, by measuring the text with a carpenters ruler? 

[Detlev]

count the zeros 🤣

Yes, that's annoying, close to unusable.  There is still space left in the Siglent screen, so why they didn't add thousand separators?  Didn't they notice how unreadable that is?

How did they count the zeroes during developing, by measuring the text with a carpenters ruler? 

Yes, it's not nice that you can enter more decimal places than the SDG can display afterwards

[Smokey]

I noticed that on the older SDG1010.  It will let you, from the front panel, enter more digits than it shows for values on the screen.
Like setting the duty cycle, it will only show 10.1% but it will let you enter 10.15%.  You can watch the signal change by that amount on the scope and see that it's actually doing it.

[NE666]

I am familiar with the new Rigol DGxxxPro-line and its certain fundamental issues here the last few weeks in use, which makes me wanna look for alternatives

Is there any chance that you might elaborate and share your findings with us?  Perhaps in the existing DG800/900Pro thread if that is more on topic?

Thanks in advance.