Siglent SSA3000X and SSA3000X-Plus Spectrum Analyzers

[tautech]

A few pages into this thread is when Dave got a SSA to review and teardown, he got himself in a pickle with the UI until he got his head around it. The lesson he learnt was to NOT let fingers work faster than brain. < Double note !!

SSA3032X is my first SA and I expected a very steep learning curve even for basic usage however with the excellent examples that rf-loop has posted here it was not so hard at all. Start with simple exercises until familiarity with the instrument is gained, try to display parameters of the SAME signal in the way rf-loop has to understand this instruments capabilities.
Then maybe you will be ready to progress onto real work.

[rhb]

 I have been very methodically going through all the features with an antenna connected to the input to learn what they do using the FM broadcast band as input.  I'll get to other signal sources in time.  First things first.

After rereading the entire manual it became clear that it's a language issue.  To an English speaker "Peak Peak" is synonymous with "Maximum Peak".  In fact, that button selects the minimum and maximum points on the trace and then gives the frequency difference between the two points.  As the minimum changed the behavior was random.  "Min Max" would be a more appropriate label in English.

Having it hang and alter the static IP address from what I had assigned is not something that inspires confidence in a complex product.  I had assigned 192.168.0.10, but after it rebooted it was set to 192.168.0.1.   Naturally pinging 192.168.0.10 no longer worked.  I had set the gateway address as 0.0.0.0 because there is no external WAN link.  After the hang and reboot it was set to 192.168.0.2.  I did not do anything that should have caused that.  I have since set it back to 0.0.0.0.

[rf-loop]

After rereading the entire manual it became clear that it's a language issue.  To an English speaker "Peak Peak" is synonymous with "Maximum Peak".  In fact, that button selects the minimum and maximum points on the trace and then gives the frequency difference between the two points.  As the minimum changed the behavior was random.  "Min Max" would be a more appropriate label in English.

If we think example oscilloscope or example signal generator. What is oscilloscope Peak-Peak measuring?  How about example function generator or signal generator if there is level setting, say example Voltage level. What is 1Vpp. It is 1V Peak to Peak. Indepenmdent of if peaks are 0 and 1V or -0.5V to + 0.5V, and so on.

How ever we name this function it is not difficult to remember that SSA Marker Peak menu  [Peak Peak]  can also read as min Peak - max Peak.

This function is one-shot function. Once you push [Peak Peak] button it find these min and max peaks and set Delta marker for these (marker number is selectable by user if need and also other Delta markers are in use). 
After then this marker (ref and delta) frequencies are fixed (until you push again [Peak Peak]) and these markers follow level in these frequencies. Display (top) show Delta Marker  frequency and level differences between markers. User can select what Delta marker pair is used for Peak-Peak. Just select wanted marker number in Marker menu before go to Peak menu.

Here is one simple example.

Tracking generator output (signal level -20dBm) goes to external circuit input and external circuit ouput connected to SA input.

[rhb]

Thank you for an excellent example of the proper use of the function. Perhaps it could be included in the manual along with the circuit.  It rather looks as if it is the series and parallel resonances of a quartz crystal.

I notice that with the input terminated with 50 ohms and the unit at the factory preset, power on settings there is a 5-10dB slope rising from 0 Hz to 2.1 GHz.  Is there a setting to correct this?

In general I would like to store a trace and then subtract it from another stored trace or the current input trace.  Is this possible?  If so, how?

[rf-loop]

I notice that with the input terminated with 50 ohms and the unit at the factory preset, power on settings there is a 5-10dB slope rising from 0 Hz to 2.1 GHz.  Is there a setting to correct this?
In general I would like to store a trace and then subtract it from another stored trace or the current input trace.  Is this possible?  If so, how?

Base noise can not correct to flat. (oh well... it "can" but then we loose most important things, signal level flatness what is real important. )

It is natural and quite common in spectrum analyzers that base noise level rise when we go to higher frequencies.

This do not affect signal level  accuracy. It is corrected by internal flatness table (1) -  for every single individual SA (this is done by factory and this is one important part of factory calibration data (discussed also in some "hack" threads.
Also PreaAmplifier flatness is corrected so that measured signal levels are in specifications. And as can see it also produce bit different base noise.  If we look more carefully this base noise it have quite complex shape with many hills and valleys.

This is one individual SA noise floor but other SA may be some amount different.

SSA3021X (previously in this thread:"Reply #212 on: June 18, 2016, 11:54:05 AM" )

Do not worry it is not flat. (as long as it is inside specs.)


SSA3000X DANL





Edit/add: Keysight N9320B DANL specs.


(1) every SSA3000X have after factory calibration process individual level and flatness correction table. It is in folder "calib". 
Example every 52 attenuator level (0-51dB) have its own 8kb level correction table. Also there is many other factory cal data.

[Dubbie]

I was measuring an oscillator today and was pleased to see how accurate the reported frequency was.
The SSA is running internal reference. My counter is running external 10MHz GPSDO lab reference.

looks like about 1Hz accuracy @ 8MHz

[rhb]

I have not seen any reference to how to submit bug reports.  Because Siglent support staff read EEVblog, this seems as good as any.  If there is a specified protocol for reporting bugs please direct me to it.

I had setup and was looking at the passband of a 2 pole LC filter using a log X scale. I had normalized the test setup with the filter shorted and then inserted the filter.  I was averaging 999 traces.

When I changed the X scale to linear the output was renormalized to a flat line with a lot of noise at the extremes of the X axis.  Setting it back to log did not remove the 2nd normalization.  So I had to repeat the entire setup procedure just to change the horizontal scale.

I am using the 1.2.8.2 firmware installed by the factory in May 2017.  The seller does not allow returns if the buyer has modified the unit, so I do not want to update the firmware until I have satisfied myself that there are no show stoppers.

Markers have a similar problem in that they are mapped to screen coordinates rather than display coordinates, so changing between log and linear X places the markers in the wrong location.

[rhb]

Here are 2 JPEGs showing the problem.  The only difference is a single button push to change the X scale.  I'd like to note that the previous times I've done this there was a flat line in the center of the linear X plot.

Despite being very meticulous about keeping detailed notes and trying to do exactly the same thing, I have found it difficult to consistently reproduce problem behaviors.  This suggests to me an uninitialized variable issue.  I've been sole support for several million lines of other people's code, so I have a pretty good intuition about such things.  I'd like to suggest Siglent invest in a static analyzer package to check there source code.

BTW Previous examples did not have the spikes decorating the peak response.  No idea yet what is going on with that.

[rhb]

When I changed the scale I dislodged the signal connection.  Here's what things typically look like.

[tautech]

I am using the 1.2.8.2 firmware installed by the factory in May 2017.  The seller does not allow returns if the buyer has modified the unit, so I do not want to update the firmware until I have satisfied myself that there are no show stoppers.

That's very unreasonable, you should contact them and insist that you can install at least 8.3 FW that seems it may be addressing some of your issues.
8.3 link:
http://siglentamerica.com/gjjrj-xq.aspx?id=6381&tid=15

I have found it difficult to consistently reproduce problem behaviors.

So did I until I found ways to return the unit to startup settings and become more familiar with the UI.
It does take some little time before you don't get lost and with tits in a tangle.

As we have no idea from your profile where you might be it's hard to even suggest which Siglent branch might best help you. They have paid tech support staff waiting to help.

[rhb]

Given the risk of a buyer bricking the unit during an upgrade, I don't think the seller is being unreasonable. I plan to discuss with the seller today. This is more nuisance than anything else.

I've taken to power cycling the unit to be assured of reproducible initial state. I do wish there were more information about the hardware.  For example if the unit has power but is "off" is the internal reference running?  I have all the stuff on my bench on a power strip so that I can turn it all on and off at once.  That's obviously not desirable if the soft power was done to keep the oscillator running. 

I'm used to HP & Tek gear and manuals.  I have a Motorola spectrum analyzer,  but it needs work.  My reason for buying this was repair fatigue and the cost of new from Tek and Keysight.  So  very poorly done manuals  without so much as a block diagram is discouraging.  Were I making the purchase in a professional context rather than a hobby context I should not have considered  Siglent after reviewing the documentation and seeing how skimpy and error ridden it was.

It would be helpful if Siglent were a bit more detailed in their release notes.  Rigol didn't even have any when I complained about the glacially slow transfer speed to USB on my DS1102E.  I did eventually install a FW update, but it did not resolve the problem.  I'm looking seriously for a replacement, but am going to ask for a demo before I commit and a service manual is going to be a requirement.  I was told by Siglent support that the service manual for the SSA3021X had not been finished despite the product being on the market for 2 years.

I'm in the US.

[Bicurico]

There is little chance of bricking the SSA3021X during a firmware upgrade, especially if you already are at v8.x.

It would take extra effort to deliberatly brick it - it would be easier to fry the input stage.

That seems to me as a totally ridiculous demand of the seller: if there was any risk, then Siglent themselfes would not offer a public firmware upgrade. It would be done by authorised centres, like what is done in some cases with premium brands.

Also, tell your seller that the warranty is covered by Siglent, so if the upgrade would brick the device, then Siglent would have to repair it.

It makes no sense at all using a SSA3021X with an older firmware, especially if you encounter bugs which have been fixed in newer FW releases.

Regarding the manuals, it is always complicated to keep up the manual, when developing the firmware. Yes, they should finish it and keep it updated along with the firmware releases, but the truth is, that the device is pretty self-explanatory. I never fully read the manual. I guess using the device just makes sense.

Regards,
Vitor

[nctnico]

I'm looking seriously for a replacement, but am going to ask for a demo before I commit and a service manual is going to be a requirement.  I was told by Siglent support that the service manual for the SSA3021X had not been finished despite the product being on the market for 2 years.

What do you expect to find in a service manual? You might find a procedure to check the device and determine which board to replace in case of a failure but that's it. Service manuals with detailed repair procedures and schematics are something from a long forgotten past. Repair is still possible but you'd have to reverse engineer the device and have a good understanding of how it should work in principle.

[W7NGA]

Just received my SSA3021X and running it thru some testing. No, I don't care how loud the fan is (it isn't), nor do I care how long it takes to boot (reasonable). However, I am thrilled with the performance, especially considering the latest pricing and free Tracking Generator. I'll still use my HP 8711C Network Analyzer for antenna design, but my beloved HP8594E Spectrum Analyzer will most likely get banished to the back of the bench! Siglent has done an amazing job with this analyzer ... every ham radio operator (and retired RF engineer with a love for instrumentation) should own one!

W7NGA
Seaside, Oregon

[trevwhite]

That's good to hear. I actually took a flyer and ordered a Siglent SSA3021X myself. It is a lot of money for a Siglent product but with all the information on this forum it seemed an okay gamble.

I do have a niggle wondering if I should have looked at the FPC1000 a bit more. Am thinking the FPC1000 would be great build quality but everything is an option and I think it probably would have cost a lot more money. The SSA3021X appears to offer real bang for buck.

[rhb]

I had a nice chat with the seller's tech support after which I updated the FW to 1.2.8.5a.  My unit was shipped a few days before he updated all the units in stock.

The first USB flash drive I tried repeatedly hung the unit so that I had to remove the power.  Not even the power switch had any effect.  I was on the phone to the tech at the time.  Later I tried a different drive and the update went smoothly, but the problem shown in #708 still persists.

I plan to discuss further tomorrow before returning the unit.  There are some things that suggest a hardware fault.

I'll be very interested in what W7NGA has to say after he's spent a few weeks using the unit.  There are not many options I can afford.  The HP8594E seems to be the best and I've already been through used gear dying shortly after the warranty expired.  I eventually fixed it (bad solder joints), but it was a pretty tedious task and required another not quite fully functioning unit to accomplish.

[tautech]

I've taken to power cycling the unit to be assured of reproducible initial state.

There's no need to do that, use the Preset button that when new is set to factory default but is user definable for your favourite setup.

I do wish there were more information about the hardware.  For example if the unit has power but is "off" is the internal reference running? I have all the stuff on my bench on a power strip so that I can turn it all on and off at once.  That's obviously not desirable if the soft power was done to keep the oscillator running. 

Most modern equipment doesn't like hard switching, instead use the power button, AFAIK the internal ref is only active when the unit is booted. Even as such the unit need be On and warmed for spec to be guaranteed.

I'm used to HP & Tek gear and manuals.  I have a Motorola spectrum analyzer,  but it needs work.  My reason for buying this was repair fatigue and the cost of new from Tek and Keysight.  So  very poorly done manuals  without so much as a block diagram is discouraging.  Were I making the purchase in a professional context rather than a hobby context I should not have considered  Siglent after reviewing the documentation and seeing how skimpy and error ridden it was.

There's further documentation that might help prepared by the guys in Siglent Ohio here:
http://www.siglentamerica.com/prodcut-pxzl.aspx?id=5113&tid=227&T=2

I plan to discuss further tomorrow before returning the unit.  There are some things that suggest a hardware fault.

Take some more time to get your head around this unit, faults are unheard of.

I've already been through used gear dying shortly after the warranty expired.

I've got Siglent DSO's running for 5 years in tech schools that have never given any problem. None.

Little has changed in the marketplace since these were released, they offer great bang for buck and are well spec'ed. Sure, some take a while to get to grips with them to where they can acquire the results needed.

Of course I seem one eyed......but I think back to when rf-loop had not long had his and his classic HP analyser went down the road when he knew that these were a good replacement for it.

[rhb]

@tautech I  don't want to quote your entire post so:

1) The system locked up so that the PRESET had no effect within the first few hours of operation.  That was the first instance,  The power switch still worked, but the static IP assigned had changed.  The static IP has been changed again with no obvious anomalous behavior.  As Siglent's documentation is quite poor, using the soft power switch seemed the prudent thing to do in trying to precisely reproduce anomalous behavior.

2)  You don't know and neither do I.  It would make a great deal of sense for the reference to run continuously.  Almost all instrument behavior is obvious if you understand the design.  But Siglent are not telling.

3) The QST article was interesting.  I'd missed that.

4) It hung so hard while I was on the phone to the seller's tech support I had to remove power.  Not even the power switch would work.  This appears to have been caused by the USB flash drive I was trying to use to update the FW.  It happened as soon as I tried to start the FW update process.   A different USB drive worked fine.  The USB in question works fine with Solaris & OpenIndiana, but is not seen by Windows 7 Pro.  This is my first encounter with a flash drive doing that.

5)  I bought a 60's vintage dual trace 60 MHz Dumont 1060, my first triggered sweep scope, in 1990 with a 30 day warranty from the seller.  About 6 weeks later it died.  Having paid over $300 this was rather painful. It sat dead for several years until I picked up a semi-functional Tek 465 in trade for a couple of 141 MB ESDI disks.  That was sufficient to allow me to track down the bad solder joints in the horizontal section of the Dumont.  Once that was fixed I went through and fixed the problems with the 465.  After 20-30 years, lots of stuff is prone to failure at any time. The delay line on the 465 was not properly terminated.  When I went to resolder the resistor it fell out onto the bench when I melted the solder on the first lead!  I'd be very concerned if 5 year old gear died.

I fully agree that it's a very nice unit with excellent specs.  The only serious competitor is the Rigol DS815-TG.  Everything else is too much for my hobby budget.    I've noticed some noise appear that I'd not seen before using kludged test setups.  So some proper test fixtures are very much in order as this *is* RF.

I invite you to test changing the horizontal axis.  Any loose quartz crystal you have on hand will do for the test.  Setup so you can see the resonances modes of the crystal and then change the X axis.

[tautech]

rhb
Yes a crook USB stick could definitely cause the freeze and be primarily responsible for your bad first experience.

I'm not so sure the ref needs to run continuously to ensure accuracy, evidence posted by others and my own experience (limited) indicate it's very accurate from power on. I've not tried to investigate drift but it could be an interesting little exercise. 
Did you see there was 3 pages of usage notes in the Siglent FAQ link ?

I'll have another fiddle with the operation you suggest over the weekend....busy today.

[rhb]

I already had gotten an RMA before I ever stuck a USB drive in it.  So the initial hang was not related.   More curious is that I just used that  USB drive to transfer the RMA  PDF from my OpenIndiana system to my Solaris system for printing.  No issues, but neither Windows 7 Pro nor the Linux distros I tried can even see the stick.

I just sent the unit back and asked for a replacement instead of the refund I'd initially requested.  At this point I have no doubt that there is a HW issue.  Hopefully most of the problems will disappear.  Since i can talk to it over the LAN I can fix minor annoyances easily.

I'm quite amazed that the Siglent documentation is so crappy.  I was looking at the Rigol DS815-TG user manual and it's striking how much more effort they put into their manual  even if not to HP standards.

The [CENTER FREQ] units menu came up this morning with no units and would not respond until I pressed the soft power switch.  I also had a very noisy trace with a cable from the TG to the SA when I turned on the TG after [PRESET] which went away when I rebooted it.   There were some other odd things that happened, but without knowing more about the design it's pointless to speculate.  I took lots of photos which I sent to the seller's tech support along with detailed descriptions of what I'd done.

I didn't notice there were 3 pages to the link, but after spending so many hours using the unit I'd figured out how to do everything listed there of interest to me.  I did not do anything with the EMI or AMK options other than the waterfall display which my SDRplay RSP2 does much better.  I have no reason to buy those options, so no point in learning to use something that is going away.  If the trial timer were based on usage rather than power up time I might eventually find a use for them.  But I can write a quick script to do anything they do in most cases and an SDR will take care of the rest.

Thanks for your help and interest in my travails.

[rf-loop]

I already had gotten an RMA before I ever stuck a USB drive in it.  So the initial hang was not related.   More curious is that I just used that  USB drive to transfer the RMA  PDF from my OpenIndiana system to my Solaris system for printing.  No issues, but neither Windows 7 Pro nor the Linux distros I tried can even see the stick.

Whole USB drive with SSA3000X need be in  FAT32 file system!

[rhb]

It is.  I transferred the RMA PDF after I had reformatted the USB stick.  When I inserted it in the SSA3021X it was as formatted when I received it with some reports from a Stanford research consortium on it.  These were commercially produced drives with a logo showing the year and the files.   I had used it once to do a firmware update on my Rigol DS1102E.  Saving some data on the Rigol deleted all the consortium reports and theses.  At the time I blamed it on the Rigol, but I need to do some testing with a USB hard drive. It may well be that the USB stick was wonky.  Using a different USB stick, the SSA3021X worked fine.

[hendorog]

It is.  I transferred the RMA PDF after I had reformatted the USB stick.  When I inserted it in the SSA3021X it was as formatted when I received it with some reports from a Stanford research consortium on it.  These were commercially produced drives with a logo showing the year and the files.   I had used it once to do a firmware update on my Rigol DS1102E.  Saving some data on the Rigol deleted all the consortium reports and theses.  At the time I blamed it on the Rigol, but I need to do some testing with a USB hard drive. It may well be that the USB stick was wonky.  Using a different USB stick, the SSA3021X worked fine.

I'd suspect the USB stick - The company I'm currently work for ordered some commercially branded USB sticks with some marketing videos etc on them. However the failure rate was very high and they stopped handing them out.

In general with USB sticks, I use them for copying files between machines, but I never trust them as a reliable backup.

[nctnico]

Yeah, those no-name USB sticks are utter crap and I always throw them in the bin. So far Kingston USB sticks have served me very well.

[N8AUM]

I noticed a while back that my 3032X and someone else's 3021X running 1.2.8.5 that signals at -125dbm are about 2db lower than what they should be and at -135dbm its over 10db lower when RBW is 3Hz or 1Hz. 10Hz RBW and wider show the correct level.
If the answer is in AN-150 I just don't see it.

Vidas  N8AUM