Can this be done? How is the connection made with e.g. Rigol DSA815?
What manufacturers provide external tracking generators with good feature set, that have good price range, and can be connected to Rigol DSA?
Is it relatively better feature/price wise to go for an external or is the internal TG option a better deal?
My understanding is that external TG is more expensive, but provides more features.
Can someone share experience and provide a recommendation?
Can this be done through LXI or IEEE-488?
Not that I know of. On the backside you have an input for the 10 MHz reference, but that's about it. That is still going to be using the internal synthesizer of the DSA815.
Can this be done through LXI or IEEE-488?
I don't think so
A tracking generator like the name is a generator that follow (track) exactly the frequency of the SA.
The option on the Rigol is not expensive and is strongly advised when you buy one if you want to test filters ....
So an internal tracking generator is the best option?
I read somewhere that the internal TG is feature-limited. So therefore I was looking what the difference is between an external TG and an internal TG, and what the respective benefits are.
Do you know about the limitations of the internal TG? And if these are acceptable or if one needs workaround through an external TG?
Lets put it this way: I can count, and I got the DSA815 with the tracking generator option. Well worth it. Because no, there is no viable way you are going to get the same kind of functionality on the cheap.
Lets look at it another way. Assume you have an external sweep generator. How are you going to feed it into the DSA815? Answer: you are not, because there is no input for it on the DSA815.
I am conveniently ignoring the matter of how to synchronize that sweep because that doesn't even matter if you cannot get your external sweep signal somehow into the mixer in an orderly fashion. Synchronize the sweep using LXI might be viable in principle (just not implemented in the DSA815
), but getting that externally generated swept clock into the DSA ... not so much.
Isn't the synchronisation important only?
The external sweep would pass through the DUT, and the spectrum analyzer measures in the same frequency range as the through LXI synchronized sweep frequency?
Or am I missing out on something?
Your external sweep generator and your DSA815 agreeing on "the sweep starts .... NOW" is insufficient. If you intend to synthesize a similar-yet-not-the-same clock inside of the dsa815 and use that for your local mixing I suspect you are out of luck.
Anyways, the point of this type of synchronization is totally, fully and utterly 100% moot since the DSA815 does not support any synchronization like this.
So the answer is: it is definitely not possible in practice on a DSA815. On theoretical devices everything is possible, that why they are theoretical.
Okey, then I need to know that I have to go for that extra option when buying a spectrum analyzer.
No direct need yet for a SA, but maybe in the future I will need this, to complete my lab setup
Let's say you use the 10MHz Ref in on the SA and on the signal generator. Then you use LXI to set the SA to Zero-Span and the generator to the same freq. Wait a bit. Then you set the next freq on the SA and the generator. And so on. All computer controlled via LXI or GPIB.
If there are slight deviations between how the sig gen generates it's tone from the 10 MHz and how the SA does it, this could be calibrated before by doing a similar sweep without DUT. Then you could wiggle the freq of the generator or SA a bit at each point till they exactly match. This deviation could then be stored and used afterwards during the real measurement.
I think this could be done and could give good results. But it would probably be awfully slow.
Let's say you use the 10MHz Ref in on the SA and on the signal generator. Then you use LXI to set the SA to Zero-Span and the generator to the same freq. Wait a bit. Then you set the next freq on the SA and the generator. And so on. All computer controlled via LXI or GPIB.
If there are slight deviations between how the sig gen generates it's tone from the 10 MHz and how the SA does it, this could be calibrated before by doing a similar sweep without DUT. Then you could wiggle the freq of the generator or SA a bit at each point till they exactly match. This deviation could then be stored and used afterwards during the real measurement.
I think this could be done and could give good results. But it would probably be awfully slow.
maybe this will clear some things up:
Think of a spectrum analyzer operating on the X axis with a synchronized
band pass filter with gain that is
only open at the position in time/frequency of the horizontal trace, which translates to a certain frequency of interest. This is pretty hard to keep in sync unless you're controlling the trace speed and position with the tracking generators sweep pulse and amplitude.
Tell you what. I have a DSA815-TG. I have a DDS that I can command around 10 MHz. For example 9 - 11 MHz. I can send scpi commands to the dsa815. So, what do I do to enable this cleverness you propose?
Because I think I see big gaping holes in the plan, but I am always happy to be proven wrong if that means enabling extra measurement goodness using gear I already have.
Think of a spectrum analyzer operating on the X axis with a synchronized band pass filter with gain that is only open at the position in time/frequency of the horizontal trace, which translates to a certain frequency of interest. This is pretty hard to keep in sync unless you're controlling the trace speed and position with the tracking generators sweep pulse and amplitude.
To make this work you have to set the SA to zero span. That means, the SA doesn't sweep anymore, but just listens on the one set freq. So you don't have to get sync between a sweep on the sig gen and a sweep on the SA: both work on just one freq.
When you have got the response on this one freq, you move both devices to the next freq and so on. Slow, but it should work.
Think of a spectrum analyzer operating on the X axis with a synchronized band pass filter with gain that is only open at the position in time/frequency of the horizontal trace, which translates to a certain frequency of interest. This is pretty hard to keep in sync unless you're controlling the trace speed and position with the tracking generators sweep pulse and amplitude.
To make this work you have to set the SA to zero span. That means, the SA doesn't sweep anymore, but just listens on the one set freq. So you don't have to get sync between a sweep on the sig gen and a sweep on the SA: both work on just one freq.
When you have got the response on this one freq, you move both devices to the next freq and so on. Slow, but it should work.
I take it you plan on hacking that 815? Not going to work, way too much trouble. How do you intend on controlling the filters, how far to push those filters, not to mention stopping the 815 trace (or internal dsp conversion process)? The list goes on! How do you plan on testing for this response, making decision on response.. where to you go next?
Back in the day, late 70's, I set up a test jig using a Tek 465 and wavetek sweep generator and some filters for a special test jig for looking at some audio characteristics.. but did not have any real calibration to work with. It was only for comparative analysis from one device to another.
I take it you plan on hacking that 815? Not going to work, way too much trouble. How do you intend on controlling the filters, how far to push those filters, not to mention stopping the 815 trace (or internal dsp conversion process)? The list goes on!
Zero span is a common function, all reasonable SAs offer it. So no need to hack anything for that.
How do you plan on testing for this response, making decision on response.. where to you go next?
You set the sa to show the max signal level at the given freq. Then you wait until both instruments are truely tuned to this freq. No other response checking needed. Then you move both instruments to the next freq via LXI. So no decisions to be made.
Finally someone who is on the same page here. Of course it should be possible by LXI.
Well then, I guess it would be good in a "perfect world" Never have used the LAN extensions, guess I'm way too old school.
Finally someone who is on the same page here. Of course it should be possible by LXI.
Why make it simple if you can make it complicated
The TG option on a Rigol 815 is not expensive, don't waste you time building like they say in France "Construire une usine à gaz"
Had a much longer reply, but why talk someone out of a unique idea?
Just out of general interest: What kind of minimum and maximum sweep times are you going to get? And how are you going to actually display the trace? If the answer is "on the PC" then fair enough. If the answer is on the DSA, then please tell me more about this feature on the DSA815. What SCPI command sequence is going to accomplish that?
Lots of smart people here with 50 different ways to skin a cat... Not a whole lot of wise people here that realize that just because you can doesnt mean you should. Its a $200 option(15% extra) , and it just works, you dont have to spend hours scripting, worrying about ethernet latency bla bla... sweeping almost offensively slow if youre setting frequency, using zero span, reading the value, and moving to the next freq. Having to code even more when you want to normalize your TG... Or just go to the TG menu, turn the output on, jumper the TG to the input, and hit normalize...
Source: Ive spend the last couple weeks programming labview using LXI to manually trigger and record data from an N9914 since it doesnt do recording.
just get a noise source and skip the whole tracking thing. I have tracking on my SA, yes it is helpful, but there are times when I use the noise source and get the same plot. There's one on ebay that works ok if you find the mods dropping it to 12v and add a high quality noise diode.
The other element missing here is the normalization. Before monkeying about with any actual measurement, one will have to get a baseline set of data to normalize to (as it's doubtful any low cost signal source will be perfectly flat over frequency) then have that ready to go for calculating the delta.
Honestly, it sounds like the OP would be better off with one of those home-brew Spectra-Scope options that use the X-Y function of the scope to become a pseudo-spectrum analyzer with some additional monkeying about.