Keysight MXA Signal Analyzer / Spectrum Analyzer Review, Analysis & Experiments

[Hugoneus]

In this episode Shahriar reviews the long awaited Keysight MXA Signal Analyzer (N9020B). The new X-Series Spectrum Analyzers from Keysight offer an entirely re-designed GUI interface which supports multiple tabs as well as multi-touch interaction. The review is organized as follows:

1:07 –   Model comparison and overview of x-series spectrum analyzers.
7:08 –   Instrument physical overview.
11:49 – New GUI overview and demonstrations.
22:44 – Instrument block diagrams and analysis.
28:46 – Brief look inside the MXA spectrum analyzer.
32:53 – Back-scattering communication experiment block diagram and description.
37:16 – Phase noise characterization of doubler and amplifier.
51:54 – Complete back-scattering experiment measurements and analysis.
1:04:45 – Analysis of broadband modulated signals by using the S-Series scope and MXA Signal analyzer at the same time.
1:13:56 – Concluding remarks.

You can watch the video here: [1 Hour & 15 Minutes]

youtu.be/RIIv1KyoXss

The Signal Path
www.TheSignalPath.com
www.YouTube.com/TheSignalPath
www.Patreon.com/TheSignalPath

[videobruce]

From what I could read on that chart, the cheapest one was $13 large. 

[Hugoneus]

From what I could read on that chart, the cheapest one was $13 large. 

Yes, these are pro units for sure. The MXA that I tested with all the options is $150k at the minimum.

[BFX]

Yes I think I have to buy one to every my room 

[radar_macgyver]

Thank you for the well made video, it was most informative. We *may* (keeping fingers crossed) have enough money in our budget this year for a new instrument, maybe our existing MXA (2008 model) has enough trade-in value to get us a new MXA. I'll be sure to let the Keysight rep know about your video if we do!

[Hugoneus]

Thank you for the well made video, it was most informative. We *may* (keeping fingers crossed) have enough money in our budget this year for a new instrument, maybe our existing MXA (2008 model) has enough trade-in value to get us a new MXA. I'll be sure to let the Keysight rep know about your video if we do!

That would be great! Thank you.

[rachaelp]

Awesome video as usual Shahriar. I'm looking to buy a spectrum analyzer this year but unfortunately my budget won't stretch to an MXA.... I can always dream I guess...

[Hugoneus]

Awesome video as usual Shahriar. I'm looking to buy a spectrum analyzer this year but unfortunately my budget won't stretch to an MXA.... I can always dream I guess...

It all depends on what range of functions you need at the end. Either way, all X-series ones are pricey.

[technogeeky]

Awesome video as usual Shahriar. I'm looking to buy a spectrum analyzer this year but unfortunately my budget won't stretch to an MXA.... I can always dream I guess...

It all depends on what range of functions you need at the end. Either way, all X-series ones are pricey.

Coming from the software world, I really do not like how the price of software options causes the price of these devices to rise exponentially. At the very least, I would prefer that you at least had the option to build and develop the software on your own and take advantage of the hardware. I do understand that they have to use a pricing model that allows them to recoup the tremendous cost of R&D on the custom hardware they put into these devices. I also understand that most of their customers are corporations and laboratories that can (and do) afford the cost of these devices. If the lab needs it, they will pay for it. But it seems like the hobbyists and prosumers get left out of this at the end of the day, which is disappointing.

But man is that thing fast. The speed at which that device makes measurements and changes from one measurement to another is really astonishing.

[rachaelp]

Awesome video as usual Shahriar. I'm looking to buy a spectrum analyzer this year but unfortunately my budget won't stretch to an MXA.... I can always dream I guess...

It all depends on what range of functions you need at the end. Either way, all X-series ones are pricey.

I think the problem is finding the right balance between what I need for my work and what I want for experimentation/learning and just having fun with cool stuff. I'm a test equipment junkie and the really high end equipment like the MXA fascinates me but it's all out of my current budget. You certainly have access to a lot of really great kit, I have serious lab envy when I watch your channel!

Best Regards,

Rachael

[G0HZU]

Nice intro and overview as always.... however the doubler test setup for the phase noise test was a bit bizarre.

It looked to me like the amplifier after the doubler has a lot of gain (45dB?) and the doubler was being under driven to compensate for this. All very strange.

Would it not have been better to drive the doubler properly and dispense with the Avantek amplifier? The phase noise test would have had a bit more credibility. Also I suspect that the phase noise of the analyser itself would be playing a part in your tests. Especially at offsets very close to carrier.

Finally, I can't understand why you think your backscatter test was 'really testing' the dynamic range of this analyser. It didn't look to be a tough test at all. There are various ways to really test the dynamic range of this analyser but I don't think that was one of them.

[Hugoneus]

Nice intro and overview as always.... however the doubler test setup for the phase noise test was a bit bizarre.
It looked to me like the amplifier after the doubler has a lot of gain (45dB?) and the doubler was being under driven to compensate for this. All very strange.

The amplifier has about ~25dB of gain at 8GHz. The doubler would have been sufficient without the amplifier.

Would it not have been better to drive the doubler properly and dispense with the Avantek amplifier? The phase noise test would have had a bit more credibility. Also I suspect that the phase noise of the analyser itself would be playing a part in your tests. Especially at offsets very close to carrier.

The purpose of the amplifier was to show the rise of the noise floor and its impact on the phase noise measurement. The whole point was to show how three regions of the phase noise measurements are impacted by various effects:

1) Close in phase-noise is dominated by the instrument's own internal phase noise and thus the two measured curves line up. I offered this as a puzzle to the viewers.
2) Mid-range phase noise is scaled mathematically based on the consequence of the doubler.
3) Far out phase noise is increased beyond the expected 6dB due to the excessive noise from the amplifier.

This is a valuable test to teach people what to watch out for. Other than that I know how to optimize phase noise if it was the only goal.

Finally, I can't understand why you think your backscatter test was 'really testing' the dynamic range of this analyser. It didn't look to be a tough test at all. There are various ways to really test the dynamic range of this analyser but I don't think that was one of them.

The backscatter experiment does not really test the dynamic range of the instrument during regular measurement but it shows the limit during real-time measurement and that was the point. I chose the TX/RX power to be close to the limit of the real-time digitization dynamic range. Also, I need to make a fun experiment not just "verify" the spec, anyone can just look that up from the datasheet.

[G0HZU]

The amplifier has about ~25dB of gain at 8GHz.

Are you sure about that? That amplifier should have a low noise figure, maybe 4 or 5dB so it would need to have about 45dB of gain to produce the high noise level on the analyser via the resistive splitter and the coax that feeds to the analyser. How else do you explain the -35dBm of noise power (in a 6GHz BW) that you measured on the analyser?

Your tests suggest to me that the amplifier probably has a gain of about 45dB if we assume a sensible/realistic noise figure of (approx) 5dB for such a wideband amplifier.

This would produce the 6GHz wide noise pedestal you saw on the analyser if you take into account the loss in the splitter and coax after the amplifier.

[technogeeky]

The amplifier has about ~25dB of gain at 8GHz.

Are you sure about that? That amplifier should have a low noise figure, maybe 4 or 5dB so it would need to have about 45dB of gain to produce the high noise level on the analyser via the resistive splitter and the coax that feeds to the analyser. How else do you explain the -35dBm of noise power (in a 6GHz BW) that you measured on the analyser?

Your tests suggest to me that the amplifier probably has a gain of about 45dB if we assume a sensible/realistic noise figure of (approx) 5dB for such a wideband amplifier.

This would produce the 6GHz wide noise pedestal you saw on the analyser if you take into account the loss in the splitter and coax after the amplifier.

The datasheet for the amplifier is available. It's the AGT-8235 part. We never actually saw the exact setup in the video (due to Shahriar's usual meticulous attention to detail I assume this is an oversight). But we can assume that the gain is somewhere between 45dB and 25dB. The voltage control of the gain is inverted and cutoff; so at 0V gain control the gain is 45; any increase in voltage decreases the gain until you get down to 25dB. It appears the noise goes up when you do this (from 5.5dB to 7dB).

So I guess the phase noise shown must come from a combination of phase noise from the doubler and the amplifier together? Or the amplifier is not functioning perfectly?

[Hugoneus]

Thanks for the details. I was quoting the gain from memory and I could be off. But I will measure it tonight.

[G0HZU]

The datasheet for the amplifier is available. It's the AGT-8235 part.

Thanks. I estimated the gain of the amplifier after making a few reasonable assumptions. A broadband $$$ Avantek amplifier like that will have a noise figure of about 4 or 5dB.

I assumed that there was about 9dB total passive loss in the splitter/cable after the amplifier.

The noise power measured on the analyser in the 6GHz BW was -35dBm and 6GHz BW = 98dB Hz. So the amplifier gain would be approx:

Gain = (-35 - 98 - 5 + 9 - (-174)) = 45dB

This seems to agree with the Avantek datasheet if the amp is set to the default gain?

Note: looking at the typical gain in the datasheet graphs it is more like 50dB gain for that model at 8GHz. Maybe there is more loss in the splitter/cable part of the test setup that I first thought?

It seems a curious choice for a post doubler amplifier because it would mean you would have to really reduce the drive level to the doubler to get 0dBm on the analyser at 8GHz. No wonder the rejection of the 4 GHz fundamental looks poor from the doubler if it isn't being driven into its normal working region?

[technogeeky]

The datasheet for the amplifier is available. It's the AGT-8235 part.

Thanks. I estimated the gain of the amplifier after making a few reasonable assumptions. A broadband $$$ Avantek amplifier like that will have a noise figure of about 4 or 5dB.

I assumed that there was about 9dB total passive loss in the splitter/cable after the amplifier.

The noise power measured on the analyser in the 6GHz BW was -35dBm and 6GHz BW = 98dB Hz. So the amplifier gain would be approx:

Gain = (-35 - 98 - 5 + 9 - (-174)) = 45dB

This seems to agree with the Avantek datasheet if the amp is set to the default gain?

Note: looking at the typical gain in the datasheet graphs it is more like 50dB gain for that model at 8GHz. Maybe there is more loss in the splitter/cable part of the test setup that I first thought?

It seems a curious choice for a post doubler amplifier because it would mean you would have to really reduce the drive level to the doubler to get 0dBm on the analyser at 8GHz. No wonder the rejection of the 4 GHz fundamental looks poor from the doubler if it isn't being driven into its normal working region?

Re-reading this post for clarity, I do think that you have downplayed the fact that he's really trying to show off the capabilities of the instrument, and he's not trying to maximize the capability of the DUT. I don't actually even understand enough yet to have the kind of intuition you two have about the capabilities and noise figures of these instruments, but I do think he did a good job of showing off the instrument with a demonstration of a practical circuit.

In any case; Shahriar, did you have time to investigate? What, if anything, did you find?

[Hugoneus]

The datasheet for the amplifier is available. It's the AGT-8235 part.

Thanks. I estimated the gain of the amplifier after making a few reasonable assumptions. A broadband $$$ Avantek amplifier like that will have a noise figure of about 4 or 5dB.
I assumed that there was about 9dB total passive loss in the splitter/cable after the amplifier.
The noise power measured on the analyser in the 6GHz BW was -35dBm and 6GHz BW = 98dB Hz. So the amplifier gain would be approx:
Gain = (-35 - 98 - 5 + 9 - (-174)) = 45dB
This seems to agree with the Avantek datasheet if the amp is set to the default gain?
Note: looking at the typical gain in the datasheet graphs it is more like 50dB gain for that model at 8GHz. Maybe there is more loss in the splitter/cable part of the test setup that I first thought?
It seems a curious choice for a post doubler amplifier because it would mean you would have to really reduce the drive level to the doubler to get 0dBm on the analyser at 8GHz. No wonder the rejection of the 4 GHz fundamental looks poor from the doubler if it isn't being driven into its normal working region?

Re-reading this post for clarity, I do think that you have downplayed the fact that he's really trying to show off the capabilities of the instrument, and he's not trying to maximize the capability of the DUT. I don't actually even understand enough yet to have the kind of intuition you two have about the capabilities and noise figures of these instruments, but I do think he did a good job of showing off the instrument with a demonstration of a practical circuit.

In any case; Shahriar, did you have time to investigate? What, if anything, did you find?

The test was done to show specific performance of the MXA, not optimize the DUT as I mentioned before. The gain of the amplifier has no bearing at all on the test itself, because that is not what is being measured. I still have to measure the gain, but I suspect that it will match the datasheet.

[G0HZU]

Yes, but in the video you appeared surprised how noisy the amplifier was and made comments about how poor the doubler performance was. Avantek don't make noisy amplifiers and MiniCircuits make decent doublers. But they only appear decent if they are used sensibly. Your test setup was far from 'sensible'. I've designed plenty of doublers using MiniCircuits parts over the years for many products and so anyone like me is going to spit their drink out when they watch your doubler test and listen to your commentary and see the spectrum on the analyser.

[diyaudio]

Yes, but in the video you appeared surprised how noisy the amplifier was and made comments about how poor the doubler performance was. Avantek don't make noisy amplifiers and MiniCircuits make decent doublers. But they only appear decent if they are used sensibly. Your test setup was far from 'sensible'. I've designed plenty of doublers using MiniCircuits parts over the years for many products and so anyone like me is going to spit their drink out when they watch your doubler test and listen to your commentary and see the spectrum on the analyser.

please make us a video and go into detail about it.. we waiting. 

[Hugoneus]

Yes, but in the video you appeared surprised how noisy the amplifier was and made comments about how poor the doubler performance was. Avantek don't make noisy amplifiers and MiniCircuits make decent doublers. But they only appear decent if they are used sensibly. Your test setup was far from 'sensible'. I've designed plenty of doublers using MiniCircuits parts over the years for many products and so anyone like me is going to spit their drink out when they watch your doubler test and listen to your commentary and see the spectrum on the analyser.

You are still missing the point. The element of "surprise" you refer to is the flair I am adding to make the test exciting and engaging. I know very well that I am purposely under-driving the doubler. I am trying to incorporate specific failings into the test so I can both teach something as well as test the instruments.

Either way, I think everything should be clear by now.

[G0HZU]

You are still missing the point. The element of "surprise" you refer to is the flair I am adding to make the test exciting and engaging. I know very well that I am purposely under-driving the doubler. I am trying to incorporate specific failings into the test so I can both teach something as well as test the instruments.

Oh wow... yet again on the internet I'm the only slow one. I'm supposed to be able to tell apart 'flair and pretend surprise' from (deliberately) incompetent use of high performance RF modules

diyaudio:   please make us a video and go into detail about it.. we waiting. 

Look, I sometimes comment and contribute to these teardown videos. Sometimes I'm critical and sometimes I'm not. Maybe you should take the time to read my technical contribution to this one below. Maybe I deserve more respect than your 'doubting' popcorn icon.

https://www.eevblog.com/forum/testgear/video-teardown-and-repair-of-an-agilent-e4433b-esg-d-synthesized-signal/

I don't do presentations, I'm an engineer usually hidden in a lab far away from the general public. My presentations skills are poor and so are my social skills. My video editing and IT skills are even worse. I really don't think you want to see me do a video about doublers.

My background is in RF design and what is relevant to threads like this is that I have experience of designing high performance RF converters for use with a digital IF. My experience goes back here to about 1990. So over the years I've designed lots of high performance gear that resembles todays signal/spectrum analysers. So looking inside teardowns of modern RF gear like this is interesting to me. If anyone is going to be critical of anything odd in the presentations then I guess it is often going to be me.

[Hugoneus]

Nothing wrong with criticism, just commenting on the style of it. Please continue to lend the forum your insight.

[diyaudio]

You are still missing the point. The element of "surprise" you refer to is the flair I am adding to make the test exciting and engaging. I know very well that I am purposely under-driving the doubler. I am trying to incorporate specific failings into the test so I can both teach something as well as test the instruments.

Oh wow... yet again on the internet I'm the only slow one. I'm supposed to be able to tell apart 'flair and pretend surprise' from (deliberately) incompetent use of high performance RF modules

diyaudio:   please make us a video and go into detail about it.. we waiting. 

Look, I sometimes comment and contribute to these teardown videos. Sometimes I'm critical and sometimes I'm not. Maybe you should take the time to read my technical contribution to this one below. Maybe I deserve more respect than your 'doubting' popcorn icon.

I don't doubt you, you were a bit harsh your respect was honored a result of your "down to earth come back", not your contribution. just saying.

[Z80]

It's easy to be an armchair critic, but not so easy to step up and produce a quality Youtube channel.  I for one appreciate the time and effort put into your videos, thanks and keep up the good work.