Siglent SDG6000X series 200-500 MHz AWG's

[tv84]

It seems 6.01.01.35R5B1 is some beta version as some things are fixed but some are broken. 

tautech, don't call it beta just because of the bugs... If it is beta it should have followed the beta members path only.

I think it's an official version that should have been more thoroughly tested, according to previous members! Although I admit it's very difficult to emulate all the tests that the experts here submit their equipments to! 

Let's hope that Siglent is able to correct them in a timely manner.

[tautech]

It seems 6.01.01.35R5B1 is some beta version as some things are fixed but some are broken. 

tautech, don't call it beta just because of the bugs... If it is beta it should have followed the beta members path only.

I think it's an official version that should have been more thoroughly tested, according to previous members!

Yes and no, Siglent is getting lazy not renaming their FW file to drop the B, we have seen this before and it sometimes means the firmware is unfinished.
One time I got emails from DSO beta testers proclaiming surprise that some FW had been uploaded to the websites yet it was still in development. Normally a more secure method is used to share beta FW.

Let's hope that Siglent is able to correct them in a timely manner.

I have 3 factory confirmed SDG6kX issues in my inbox.

One of which is a new SCPI command to load a users IQ creation: :IQ:WAVE:USER "UserIQ_1.arb"  (for 6.01.01.35R5B1)
This should appear in the next Programming guide revision.

[tv84]

So, Siglent China has the FW for SDG6000X-E model:

http://www.siglent.com/upload_file/zip/firmware/Signal_generator/SDG6000X-E__6.01.01.33R2_CN.zip

It's Product ID: 10900

It's older brother is 10800.

The .ADS package also flashes a BOOT.bin...

[2N3055]

Fun fact:
Rigol just released new siggen DG2000. It is sort of replacement for DG1000Z series.

One thing they are very proud of is change of frequency without glitches.....

QUOTE :

"
When performing related tests such as switching power supply and inverter, it is necessary to use a function/arbitrary waveform generator to generate two frequency and phase coupled signals. The traditional DDS signal generator is prone to overshoot and signal interruption when the frequency changes. The DG2000 series signal generator update algorithm ensures no interruption during frequency switching and no overshoot, which can better ensure the safety and stability of the test.
"

So it can be done....

[tv84]

So it can be done....

What I don't understand is why are they so proud of it and don't implement the "update algorithm" in the others!!

[2N3055]

So it can be done....

What I don't understand is why are they so proud of it and don't implement the "update algorithm" in the others!!

Same here.. Probably needs rewrite of basic part of bit engine, and they don't want to do it on old platform.. Which is very shortsighted..

But if Rigol does it for all new AWG, it could be a problem for Siglent, for people buying new AWG right now.....

But  Siglent should do it for their current 1000x/2000x/6000x series... It would mean significant quality upgrade and a great brand building improvement ...  And not letting Rigol be better at this...

[TurboTom]

First, Siglent should get their homework done and sort the obvious problems that still exist with the current firmware before trying to add something new. Otherwise this will end up in a complete mess -- which it doesn't appear to be far from anyway right now.

[supperman]

I think I have a decent understanding of getting to a root password.. but does this not require an active Telnet port on the device? @tv84 I just received my SDG6000X and there seems to be no one listening at port 23 (connection refused - at least via LAN). Is this new (firmware version 35R5B1) or am I doing this wrong? I can ping the IP address fine and I can Telnet to port 5025 just fine. Even if I have root password.. how do I connect?

[supperman]

Never Mind. (Pic this time WITHOUT Meta Data - Ugh)

[kladit]

Can you say something about level accuracy at higher levels, say +10dBm?

Are the harmonics still down 60 dBc in the range up to 60MHz then?

I consider to buy an SGD6022X  and use it as  an dual tone generator for IP3 measurement and as an rf signal generator mainly.

[TurboTom]

Just hooked the SDG6022X (+) via a 20dB attenuator to my SA. Kept the maximum mixer level below -20dBm since then I can be quite sure that the distortion of the SA is neglible (below the noise). The last time I measured absolute level accuracy of the SDG with the power meter, it was quite accurate (better than +- 0.2db), this time I didn't pull the power meter from the basement. Maximum single tone level @60MHz of the SDG is just short of +18dBm (17.9 something), in output combine mode to generate a "clean" two-tone signal (59.5MHz and 60.5MHz), the maximum level of each tone is slightly less than +12dBm (which is to be expected if a single tone can be generated at about 18dBm). Harmonics at +18dBm are closer to -50dBc than to -60dBc, at 0dBm the figures get better than -60dBc. I'm not sure if I can trust my SA's TOI measurement but the figures look approximately correct. I attached some screenshots. Hope this answers your questions 

Cheers,
Thomas

[kladit]

Thomas, thank you very much for your effort, very helpful.

Either the built in combiner provides not enough isolation between the two generators
or the then used single built-in amplifers IP3 is not good enough at this higher levels.

I hope not using the internal combiner but an external one (Wilkinson type) will provide
the necessary isolation.

[Performa01]

Either the built in combiner provides not enough isolation between the two generators or the then used single built-in amplifers IP3 is not good enough at this higher levels.

There is no internal combiner. Wave Combine is done in the digital domain. Consequently, we see the inevitable intermodulation distortion of the output amplifier. This is not too bad, but certainly not nearly good enough for TOI measurements of high-level components/circuits. 

I hope not using the internal combiner but an external one (Wilkinson type) will provide the necessary isolation.

Sure it will.

I've tested this with an external resistive wideband power combiner. Together with additional 10dB attenuators for each output and another 20dB attenuator for the SA input this makes for a total of 36dB attenuation and 26dB isolation between the generator outputs. This is sufficient for the 3rd order intermodulation products to be some 89dBc at +16dBm generator level:


SDG6000X_IMD_+16dBm_ext_26dB+2x10dB

EDIT: replaced the old screenshot with one from a recent test where the SA mixer level is lower and we now actually see the intermodulation coming from the generator itself.

[TurboTom]

I also already thought about an external combiner. Unfortunately, right now I've only got a three-stage cascaded Wilkinson style combiner/splitter (made of Chinesium), designed for a frequency range of 400~2500MHz. I tried it anyway at 60MHz and to my surprise, it performed better than a straight SMA-T  . Yet, at 450MHz, the performance of the external combiner lacked in general, probably since the two input ports aren't isolated well enough from each other and the (negative) feedback of the SDG's output buffers almost isn't working anymore (due to phase shift and considerable drop of open-loop gain) so the output stage is turned into a really good mixer...

FYI, here are the two screenshots. At 60MHz, the SDG6022X looks promising...

Cheers,
Thomas

[kladit]

Performa01 and Thomas,

thank you very much for your efforts.

I have just ordered an SDG6022x.

Merry Christmas, 73,

Klaus

[Performa01]

My previous screenshot came from an old test which was obviously flawed. What we saw there was the SA mixer intermodulation because of too high input level.

So I've repeated the test and will correct my previous posting.

Anyway, here's an additional scenario, with the resistive power combiner and only 6dB isolation between channels, but an additional external 30dB attenuator for the SA. Generator levels were now +16dBm, so the total external attenuation is 36dB.

It can be seen, that even just 6dB isolation is sufficient to keep intermodulation distortions below 88dBc at 60MHz.


SDG6000X_IMD_+16dBm_ext_26dB

EDIT: Measurement repeated with correct -20dBm input level.

[Performa01]

I've initially goofed up again and the SA input level was too high at -10dBm for the test in my last posting. Now corrected for -20dBm and voila, the intermodulation products have almost vanished.

This means: at least at low frequencies like 60MHz, (lack of) isolation between the outputs is not an issue and does not cause significant intermodulation distortions.

[Performa01]

As I've demonstrated in my previous postings, intermodulation distortion of the outputs is insignificant even with only 6dB isolation between them - with the reservation that this is only valid for low frequencies like 60MHz.

Some might be interested in the output intermodualtion performance at high frequencies, so I repeated my test at 480MHz.

Once again we have just 6dB isolation coming from the resistive power combiner. An additional external 14dB attenuator provides a total of 20dB attenuation. Generator levels are 0dBm as this is about the limit for the SDG6052X at high frequencies.


SDG6000X_IMD_480MHz_0dBm_ext_20dB

Now things look certainly different and the 3rd order intermodulation products are about 56dBc.

For meaningful results of  two-tone tests at high frequencies, good isolation between the outputs is mandatory.

[kladit]

The SDG6022X arrived.

To sdg6022x-isolation.jpg :

The two outputs were combined by a Mini-Circuits ZFSC-2-6 Wilkinson type combiner with 6dB attenuators at each
input port and a 3dB attenuator at the output port which is connected to a SSA3021X analyzer.

The internal attenuation of the analyzer was adjusted so that a 1dB increase of attenuation gave a 3 dBm
decrease of the IM3 products and to the attenuation at the level where this stopped , the analyzer  was assumed to
be linear now and one more dB attenuation was added as safety margin.

The IM3 levels jumped between -79.3 dBM an -82 dBm by noise. (see later)
For easy calulations -80 dBm were set.

IP3 of the analyzer = 80 dBm/2 + 3 dBm = 43 dBm

The generators own IP3 is equal or below this IP3 with the isolation used.

To sdg6022x-isolation3.jpg :

The output level of the generator was decreased by 6dBm to avoid a possible saturation of the
combiners ferrite. So the analyzer no has -3dBm intput level.
The frequency was set to the lower IM3 product, the span was set to 100 Hz and the average to 20 times.

IP3 now 95.4/2 - -3 = 44.7 dBm

This seemed to be the limit of the anlayzer, but not so, the noise level came from the SDG6022X.

To sdg6022x-noise.jpg :

Nothing was changed at the analyzer but the two outputs of the generator were switched off.
Now the noise level goes down at -111 dBm and below.

So the isolation/IP3 of the generator cannot be measured because of its own noise level, that's the bad news.

And the SSA3022x is able  to measure IM3 products of clean signals well below  -100dBm, that's the good news.

So for IP3 measurement the SDG6022X  can replace an rf-generator because of its high output level even with all
the damping needed for isolation.

But the noise level is disappointing. Or I have got a noisy one?

Any similiar measurements and critics are welcome.

--
Klaus

[kladit]

A jpg was missed for some reason.

[TurboTom]

The SDG6022X arrived.

...

Wow, that was fast! Batronix? 

The apparent noise level may be related to the generator's phase noise since your two carriers are only 20kHz apart. I tested with a carrier spacing of 200kHz. If you look at my measurements a few posts above, you see how much rounded the noise "floor" between the two peaks is. If you arrange them at 10% of that spacing, the noise level will increase a lot.

I'll try to replicate your test later this evening (without the combiner -- no time yet to make one).

Cheers,
Thomas

[Performa01]

No, the generator is fine. Its general noise level is low and the phase noise is better than 110dBc/Hz already at 1kHz distance from carrier, see my measurements, reply #163 in this thread:
https://www.eevblog.com/forum/testgear/siglent-sdg6000-series-awg_s/msg2624667/#msg2624667

Maybe it's the phase noise of the SA. I cannot test this, since I don't have one.

But I can test the SDG6052X:


SDG6000X_IMD_7MHz_12dBm_ext_28+2x20dB_RBW10Hz

Generator levels +12dBm, 20dB inline attenuators for each generator output, 6dB resistive wideband power combiner, followed by 22dB (step) attenuator. Resulting input level for the analyzer is -36dBm.

Noise floor is -150dBm/Hz in the middle between the two carriers, which are spaced only 20kHz apart. If I use 1Hz RBW, then the noise floor isn't visible anymore, despite the 120dB display range that I've set for these tests. The noise floor changes by less than 1dB if I switch both generator outputs off.

Intermodulation distortion is still not better than some 89dBc, so there is some distortion component with a linear relation to the input signal level. This appears to be simply the limit of my analyzer.

EDIT: some corrections/clarifications, especially on the input attenuator setup.

[genghisnico13]

But the noise level is disappointing. Or I have got a noisy one?

edit: ignore comment, I didn't see the span in the screenshot, my bad.
I'm not an expert but I believe the problem is how you have your spectrum analyzer set up, you have more than 30dB of attenuation, lower it or disable it completely.

[TurboTom]

@Performa01 is correct! The relatively high noise floor is related to the SA's phase noise, not the AWG's.

I replicated the tests with the 7.03/7.05MHz pair, as a combiner using 2*10dB attenuators on the two "female legs" of an SMA T and a 20dB attenuator form the "male leg" into the SA, providing a total attenuation (due to impedance mismatch within the T) of approx. 33dB, but very little feedback into the outputs of the generator. I added the external attenuation in the "Amplitude offset" menu of the SA so displayed levels refer to the input levels of the combiner / output of the AWG. I increased the internal attenuation of the AWG up to a point where intermodulation products disappeared in the noise floor and maybe a dB or two more. The RBW that has to eb selected in order to get a low enough noise level to display the highly attenuated signals with still a sufficiently large dynamic range, requires some serious sweep times, especially if the signal should be averaged for a better accuracy.

On both my SAs I've got the measurement functions activated and used the TOI program since it directly finds the peaks, generates the levels table and calculates the TOI point. Almost forgot to mention: For this test, I synchronized the instruments (AWG and SA) to my external reference (DIY unit, based on an Efratom LPRO101 rubidium oscillator). But I think, using the internal time bases wouldn't make much difference.

On the DSA815, the phase noise is specified to be better than -80dBc/Hz @ 10kHz (I measured about -90dBc/Hz) and better than -100dBc/Hz @ 100kHz (-110dBc/Hz measured). The phase noise vs. frequency has the typical inverse exponential shape, so nothing out of the ordinary here:



The analog test with the SSA3000X provides basically the same result, yet, due to the higher distortion in the front-end, lower input levels are necessary and thus a lower RBW has to be selected to reach a similar dynamic range of the signal. @Kladit used 1Hz, I couldn't be bothered to wait so long for the sweeps and used 3Hz  (with the Rigol DSA, I set up the test, went doing something else and came back after an odd hour or so...). The results are very similar, yet the noise floor appears more flat, almost inversely curved.



This gets even more obvious when viewing the signal over a higher span (changed RBW to 10Hz to speed things up, hence higher noise floor):



This is related to the funny shape of the phase noise of the SSA3000X with the valleys on both sides of the peak, and then rising about 10dB again about 50kHz off the peak (due to the superposition of the two tones here, the "dick and balls"  shape of that noise curve isn't as pronounced). Yet, at the selected frequencies in this example, the two tones both float within the central valley of the noise floor.

Even a small modification to the frequencies (setting them a little bit further apart -- 50kHz, 7.00MHz and 7.05MHz this time) changes the shape of the phase noise floor to a central hill between the (now) individual valleys of the two tones:



All of this is a clear indication that the phase noise is generated in the SAs and that with this kind of equipment, it's not possible to accurately measure the phase noise of the SDG6000X since it's considerably lower than that of any of the two mentioned spectrum analyzers.

Merry Christmas to all! 

[TurboTom]

Couldn't keep off it: In my "RF nonsense" box, I found a JCA01-2367 amplifier module that I salvaged from a Superconductor "Superfilter" teardown. It's specified for a frequency range of 824~849MHz but I found it to work over a much wider range (3dB bandwidth of approx. 400~950MHz when driven at -10dBm -- attention, I dialed in the TG level wrong so the vertical scale is 10dB less than shown, basically the scale - as it is now - can be interpreted as the gain of the amplifier):



I wasn't able to figure out the 1db compression point since the driving level available with the instruments I used, didn't permit to go beyond 0dBm drive level at the relevant frequencies. So at approx. +26dBm output power (almost 400mW!), I was more or less able to verify the 0.1dB compression specified in the datasheet:



So, since the 500MHz maximum output frequency of the SDG6000X lies well within the usable range of the amplifier module, I thought why not do a "real world" test of an TOI measurement. With the two 10dB attenuators before the summing joint, I could drive the amplifier at approximately -13dBm which should be good enough for a test. First job was to check the generator for its internal intermodulation (by feedback via the attenuators into its outputs):



Here, the intermodulation products are shown to be less than -75dBc (lower) and -77dBc (upper) -- note that Siglent confused the lower/upper designation in the TOI measurement table 

So then came the measurement of the amplifier which turned out quite nicely:



At a measured TOI point of ~42dBm, it pretty well matches the manufacturer's specs of 41.9dBm. Also the intermodulation products now are about 10dB above the reference measurement (at the same SA mixer level) which indicates that the measurement is valid and the intermodulation is happening in the amplifier and not the rest of the signal chain.

This shows that TOI measurements of RF gear are clearly possible with the SDG6000X and the other available instruments. Yet, an external combiner is mandatory, the AWG's internal summing function won't do the job since the TOI of the single output driver of the generator would swamp the measurement setup with its own IM products.