EEVblog #891 - Siglent SSA3021X vs Rigol DSA815 Spectrum Analyser

[EEVblog]

Dave compares the new Siglent SSA3021A 2.1GHz spectrum analyser with similar priced Rigol DSA815.
Noise floor, clock and PLL phase noise and other performance aspects are measured and compared between the two models.
Bugs?, yup, got those too!

[PsychoBoy]

Hey Dave, you said that 2.1GHz version is software upgradable to 3.2GHz but how did you know that?
I did not find an information in datasheet if it is possible to do such upgrade.
However I don't own this puppy so I can't check if it's possible.
Could you check it in licensing menu?

In my opinion the overall look of Siglent SA is very nice. Please do teardown someday too!

[EEVblog]

Hey Dave, you said that 2.1GHz version is software upgradable to 3.2GHz but how did you know that?

I think I may have foolishly assumed this. I've been told it's the same hardware. Checking now...

In my opinion the overall look of Siglent SA is very nice. Please do teardown someday too!

That's the plan for tomorrow.

[nctnico]

IMHO things as 'looking cheap... cheap plastic...' are better left out because all test equipment nowadays has a cheap plastic housing.  Better stick to things with can be supported by facts.
I'm also quite surprised that Siglent followed Rigol as an example. That is kind of the deaf leading the blind. Siglent could have done much better by looking at R&S, Advantest or Keysight spectrum analysers!

[dr.diesel]

Siglent could have done much better by looking at R&S, Advantest or Keysight spectrum analysers!

   Yes Sir, I very much agree.

[rf-loop]

I can not follow what exatly happend around time 33:00 and bit forward.
You perhaps have in this position RBW selection manually forced ti fixed RBW. This used RBW is accepted with this span you have just used. After then you try full span and it do not accept it because your manually forced RBW is out of accepted value in full spand. (minimum RBW for full span is 1kHz. Minimum RBW for TG is 30kHz. There is also some other limits.) If you have set RBW manually and if value is not accepted example in full band and you try full band, it tell you what is wrong but it do not force it to accepted value. It do not overwrite your selection - it tell information  and user make decision if he want change his forced parameter. 
But somehow after this situatuon there come new warning about conflict setting with TG when TG is still off and it is clear bug.

[senseless]

I can not follow what exatly happend around time 33:00 and bit forward.

I am pretty sure it has to do with the FFT mode instead of sweep. It has to do sampling at least with 4.2 GSPS for full span.
After 'Auto Tune' magic the FFT mode was gone and full span was available.
Nevertheless it is a bug, because in the beginning sweep mode was set to auto and hence it should have changed from FFT to sweep automatically.

[Earendil]

Dave, I think you should mention SFDR too.

Siglent specifies residual response as -90 dBm (yes that's m, not c) typical. So the LO(?) leakage you're seeing at -132 dBm is not that bad compared to spec. Input related spurious is only -65 dBc.

Rigol 815's spec is -88 dBm for residual response and -60 dBc for input related spurious. So yeah Siglent has better noisefloor which is a distinct advantage but on paper Siglent doesn't have as significant performance advantage as suggested, imho.

Dave compares the new Siglent SSA3021A 2.1GHz spectrum analyser with similar priced Rigol DSA815.
Noise floor, clock and PLL phase noise and other performance aspects are measured and compared between the two models.
Bugs?, yup, got those too!

[tautech]

Hey Dave, you said that 2.1GHz version is software upgradable to 3.2GHz but how did you know that?
I did not find an information in datasheet if it is possible to do such upgrade.
However I don't own this puppy so I can't check if it's possible.
Could you check it in licensing menu?

First mentioned here:
https://www.eevblog.com/forum/testgear/siglent-ssa3000x-spectrum-analyzers/msg938261/#msg938261

Thread a few days later:
https://www.eevblog.com/forum/testgear/hack-of-sigllent-spectrum-analyzer-ssa3021x/

[Kire Pûdsje]

Just a remark about the delta markers. When you take a 'difference' in dB's between two values (ie a ratio) the unit disappears. Eg the 'difference' between 23 dBm and 13 dBm is 10 dB. Likewise the difference between -100 dBm/Hz and -120 dBm/Hz is 20 dB.

It seems the Rigol takes the ratio of the powers and then divides by the RBW. For example at 24:23 I read 0 dBm and -67 dBm. the ratio would be -67 dB. Dividing by the RBW results in -87 dB/Hz.

However in my view the only point where this would make sense is when the reference marker is at the peak (as in your case). For any other point except that special case the Siglent way (just dB) is more generic. For example if I would want to compare the noise floor at two points, supposing there is a -10 dB ratio between the two powers, I would rather read -10 dB, than -30 dB/Hz. This latter value would not make any sense to me.

[KenGaler]

Seems to me that it's not a coincidence that the base model doesn't quite make it up to the 2.4GHz ISM band.  You have to spend twice that to upgrade.

Ken

[EEVblog]

IMHO things as 'looking cheap... cheap plastic...' are better left out because all test equipment nowadays has a cheap plastic housing.  Better stick to things with can be supported by facts.
I'm also quite surprised that Siglent followed Rigol as an example. That is kind of the deaf leading the blind. Siglent could have done much better by looking at R&S, Advantest or Keysight spectrum analysers!

But those companies don't make a competing scope in the price/performance bracket. Siglent only care about taking sales away from the Rigol DSA815.

[EEVblog]

Seems to me that it's not a coincidence that the base model doesn't quite make it up to the 2.4GHz ISM band.  You have to spend twice that to upgrade.

Almost certainly not a coincidence!

[EEVblog]

Just a remark about the delta markers. When you take a 'difference' in dB's between two values (ie a ratio) the unit disappears. Eg the 'difference' between 23 dBm and 13 dBm is 10 dB. Likewise the difference between -100 dBm/Hz and -120 dBm/Hz is 20 dB.

It seems the Rigol takes the ratio of the powers and then divides by the RBW. For example at 24:23 I read 0 dBm and -67 dBm. the ratio would be -67 dB. Dividing by the RBW results in -87 dB/Hz.

However in my view the only point where this would make sense is when the reference marker is at the peak (as in your case). For any other point except that special case the Siglent way (just dB) is more generic. For example if I would want to compare the noise floor at two points, supposing there is a -10 dB ratio between the two powers, I would rather read -10 dB, than -30 dB/Hz. This latter value would not make any sense to me.

That makes sense. But if you put the marker in noise mode it should know this and give you the correct units of measurement you expect IMO.

[nctnico]

IMHO things as 'looking cheap... cheap plastic...' are better left out because all test equipment nowadays has a cheap plastic housing.  Better stick to things with can be supported by facts.
I'm also quite surprised that Siglent followed Rigol as an example. That is kind of the deaf leading the blind. Siglent could have done much better by looking at R&S, Advantest or Keysight spectrum analysers!

But those companies don't make a competing scope in the price/performance bracket. Siglent only care about taking sales away from the Rigol DSA815.

IMHO that is causing 2 problems for Siglent: Firstly: they can eat into R&S, Advantest or Keysight's lower end market if their spectrum analysers operate similar to those brands. Secondly: Rigol is a very poor example to copy from.

For example: I have an older Advantest spectrum analyser and the SSA3000 would definitely be a step up from that. So even if Siglent can't match the specifications of current models they can get a piece of the market which would otherwise be fullfiled with used equipment from R&S, Advantest or Keysight.

[tautech]

IMHO that is causing 2 problems for Siglent: Firstly: they can eat into R&S, Advantest or Keysight's lower end market if their spectrum analysers operate similar to those brands. Secondly: Rigol is a very poor example to copy from.

For example: I have an older Advantest spectrum analyser and the SSA3000 would definitely be a step up from that. So even if Siglent can't match the specifications of current models they can get a piece of the market which would otherwise be fullfiled with used equipment from R&S, Advantest or Keysight.

We like these types of problems ^^^. 

[EEVblog]

IMHO that is causing 2 problems for Siglent: Firstly: they can eat into R&S, Advantest or Keysight's lower end market if their spectrum analysers operate similar to those brands. Secondly: Rigol is a very poor example to copy from.

It doesn't matter, it's clever marketing.
If someone is looking to buy at this price point, the Rigol is the only other option. So if someone demo's both, it's easy for them to make operational feature comparisons.

[nctnico]

IMHO that is causing 2 problems for Siglent: Firstly: they can eat into R&S, Advantest or Keysight's lower end market if their spectrum analysers operate similar to those brands. Secondly: Rigol is a very poor example to copy from.

It doesn't matter, it's clever marketing.
If someone is looking to buy at this price point, the Rigol is the only other option. So if someone demo's both, it's easy for them to make operational feature comparisons.

If you insist on buying new but the way I see it the choice usually is between used A-brand (with warranty) or new B-brand if you are on a budget. And then the story is completely different. If I take my Advantest SA as an example (again): the Rigol and Siglent are in the same price range of what I paid for my Advantest SA including a 1 year warranty.

[nixfu]

Finally.....someone smacks Rigol in the face to challenge them to update the(long in the tooth) DS-815 and see they can do even better.

Rigol have been sitting with no competition for the DS815 for too long.   Maybe this will convince them to make something with even better $/performance than their now several years old model.

[CustomEngineerer]

At about 25:24, talking about the phase noise of the internal clock of the Rigol, "Likely an issue with the PLL response" pops up on the screen. I'd say thats pretty unlikely. I'm pretty sure Rigol understands how to do a PLL correctly. I would be shocked if they ever had a problem with a PLL in anything they've built. 

[egonotto]

Hello,

in about 10:33 Dave compare the noise floor. He said the same detector.
But it seams to me the in the Rigol is the Pos Peak and in the Siglent is a average detector. So the real difference of noise floor is lower than 20dB.

Best Regards
egonotto

[EEVblog]

If you insist on buying new but the way I see it the choice usually is between used A-brand (with warranty) or new B-brand if you are on a budget. And then the story is completely different. If I take my Advantest SA as an example (again): the Rigol and Siglent are in the same price range of what I paid for my Advantest SA including a 1 year warranty.

Used spectrum analyser in Australia cost a fortune. Shipping is a killer as you almost always have to import them.

[AlfBaz]

Speaking of prices in Australia, I was wondering about the prices you were talking about in the video. You mentioned 1700 (I presume US) with the TG for the 2.1GHz.

It appears TRIO are selling it for closer to 1800US without TG and EX GST. By the time you add the TG and gst it would be closer to 2.5k (which is still good, I might add)

Is there anywhere I can get it nearer the prices mentioned, as at that price I would still be tempted to shell out a bit extra and go with a 2nd hand "A brand"

[bitseeker]

I would be shocked if they ever had a problem with a PLL in anything they've built. 

Yeah, that only happens with Yaigol devices.

[EEVblog]

Speaking of prices in Australia, I was wondering about the prices you were talking about in the video. You mentioned 1700 (I presume US) with the TG for the 2.1GHz.
It appears TRIO are selling it for closer to 1800US without TG and EX GST. By the time you add the TG and gst it would be closer to 2.5k (which is still good, I might add)
Is there anywhere I can get it nearer the prices mentioned, as at that price I would still be tempted to shell out a bit extra and go with a 2nd hand "A brand"

Not in Australia officially, you'd have to import it yourself.
Also, you wouldn't get local warranty support.