Keysight MSOX-6004A Test to Impress Win

[EEVblog]

What's that beast of a modular PSU?

[Howardlong]

The display is indeed enormous and awesome.

I have a similarly sized display in the MSOX7104B I have, and it's taken me a while to get it set up in a way that it doesn't takeover the bench, or alternatively is so out-of-the-way that it becomes inconvenient to control without having to lean out of he chair every few moments.

For smaller lunchbox style scopes I've used gas spring articulated VESA mounts, but the 7104B and I assume the 6004 is a bit too heavy and combersome to jury rig something up, although I did try. My solution now is a slide-out shelf on runners and a turntable that's about 8" above the bench, so I still enjoy the all important bench space and benefit from thr "enormous and awesome" display.

[memset]

Wow, such a nice, overpowered and overpriced toy!
Although best thing to get as a prize, I wonder who would prefer X6000-series scope over DSO-S series for the money?

Similarly priced, DSO-S looks to be better in all features. Except for the weight and waveform update rate, that's 450000/s for hardware-based X6000 and only 700/s for software-driven DSO-S.
But hey, for the same price magnitude you're getting loads of memory, full compatibility with InfiiniMax probes, crazy 10-bit ADC and OCXO-based timebase.
And I didn't mention Windows 7 enabling you to browse Internet directly on the scope screen pretending you're working hard.

Just can't understand why Keysight set their prices so close.

[Berni]

That is a pretty neat solution for holding a scope. I keep it on a equipment shelf behind the bench and that only works without leaning when sitting directly in front of it, while the bench is 3m long. Still means I can make a bigger mess before I decide I can't work anymore and clean it.

I am curious to see your vesa mounted arrangement

Edit:
There are bad sides to these S scopes tho. They take long to boot, they are noisy and power hungry and you don't get a nice fuzzy graduated waveform on them. My MSO6000 gets mo e use than my MSO9000 for those very reasons

[KE5FX]

Trying to understand the limitations of the FFT. I have never owned an SA, so my ability to compare is near zero. First look seems like it is slow and has a high noise floor - but keep in mind that this is a new place for me that I have very little understanding and not even sure where to look first.

See if there's a "Precision" option that you can turn on in the scope's settings/configuration menu.  Agilent added this feature to one of the later firmware releases for the DSO/MSO6000 scopes, which was extremely awesome of them.  It allows FFTs to be performed on the entire acquisition record rather than just the screenspace data buffer. 

It may not be included or relevant in the newer-generation models, but if it is, you should familiarize yourself with it.

[Howardlong]

I am curious to see your vesa mounted arrangement

Here are some details. In the first link, I had the 7000 on top of the monitor, but it was inconvenient to keep having to get up, so typically I found I ended up using the smaller form factor x3000 on the vesa mount bracket shown in the second link. The Tek MDO3000 comes with a vesa mount already fitted, and I have both the MSOX3054a and the MDO3000 on the same clamp stand on separate gas spring articulated arms.

To finish it off, both scopes are on quick release brackets so if I need to use the scopes elsewhere I can, and if I want to put another monitor on the arm for pcb/schematic layout for example I can easily do that too.

https://www.eevblog.com/forum/testgear/new-dso-for-work-any-opinions-welcome/msg861879/#msg861879

https://www.eevblog.com/forum/testgear/new-dso-for-work-any-opinions-welcome/msg871104/#msg871104

http://youtu.be/IbPh7-DGkC8?t=50s

[memset]

My MSO6000 gets mo e use than my MSO9000 for those very reasons

Sure it's better to have both, but if you'll be choosing one to spend big money to...

Here are some details.

Mein Gott! Howardlong, so many scopes

[Howardlong]

Here are some details.

Mein Gott! Howardlong, so many scopes

So I can fix the ones I break ;-)

[Berni]

Sure it's better to have both, but if you'll be choosing one to spend big money to...

Well if you spend that much you can still afford a Rigol 1000Z for the quick poke around jobs.

I still wouldn't mind a low end S series for a good price. They have software unlockable bandwidth to 8 Jigaherz and i might happen to know a little bit how the licensing system works.

[memset]

I still wouldn't mind a low end S series for a good price. They have software unlockable bandwidth to 8 Jigaherz and i might happen to know a little bit how the licensing system works.

Yep. BTW, 6000 X-series are also all-software up to 6GHz.

[rx8pilot]

What's that beast of a modular PSU?

That is a Chroma DC electronic load. Up to 8 channels - a fantastic instrument for the multi-channel power systems I have been designing.

[rx8pilot]

There's a good paper linked by rf-loop in this post:
https://www.eevblog.com/forum/testgear/siglent-ssa3000x-spectrum-analyzers/msg965873/#msg965873

Excellent paper, thank you for the link. If the paper was not nerdy enough - I was reading it at a Weird Al Yankovic concert last night for bonus nerd points.

I have a similarly sized display in the MSOX7104B I have, and it's taken me a while to get it set up in a way that it doesn't takeover the bench.......

So far, with the few hours I have in front of this scope, I am slowly developing a flow and physical layout. The scope, in general has been placed out of reach from where by DUT would be. At first it was just where I arbitrarily set it up. What I found myself intuitively doing is using the front panel to do an initial setup. When I was focusing on probing (relatively far from the scope) - I just kept my wireless mouse close by for tweaks. The voice control was reasonably handy for simple things like start/stop and single shot while both hands were on probes. I kind of like having the scope out of the way and using the mouse/voice - that arrangement gives me some space for the DUT and all the leads coming/going.

Wow, such a nice, overpowered and overpriced toy!

Keysight gave me a good deal.  Only cost a few weeks of pestering friends, family, and complete strangers for votes.

Although best thing to get as a prize, I wonder who would prefer X6000-series scope over DSO-S series for the money?

Similarly priced, DSO-S looks to be better in all features. Except for the weight and waveform update rate, that's 450000/s for hardware-based X6000 and only 700/s for software-driven DSO-S.
But hey, for the same price magnitude you're getting loads of memory, full compatibility with InfiiniMax probes, crazy 10-bit ADC and OCXO-based timebase.
And I didn't mention Windows 7 enabling you to browse Internet directly on the scope screen pretending you're working hard.

Just can't understand why Keysight set their prices so close.

It's hard to say for sure. I was looking at getting a round of financing about a year ago and that was partially for a new high-end scope. The The S-series was high on the list and the X6000 was not on the list at all. During my research, I found how difficult it is to pick an 'all-around' scope that can go from power-electronics to high-speed digital. There are so many little details that add and subtract capability in different areas. Certainly one consideration is the liability of having one very expensive scope vs 2 or more more task dedicated scopes. In SMPS design, the scope is at more risk and does not need anything close to 6Ghz. The high-speed scopes tend to focus on signal integrity for digital systems. The X6000 fits somewhere in the middle and is good at being the only scope on the bench.

While I don't know for sure, it seems the X6000 is much easier to get discounted relative to the S-Series. The MSRP's seem rather close and the S-Series has a winning front end and gobs of memory. Hard to say if the slower update rate would be a problem or not.

In the end, I never business planned to win this scope from Keysight. Now that I have it, I can spend my T&M money on other things - starting with an awesome probe kit. If I had to pay for the scope, it would take a lot longer to go beyond basic probing. In reality, I probably would have looked for a round of financing and used it on a broader variety of lower-cost gear. My money has to cover much more territory than T&M - the whole business is always screaming for more stuff - computers, software, pick and place parts, labor, wire stripping machines, laser engraver, air conditioning upgrade, etc, etc. This thing is a real world win of $40k+ taxes - massive bonus and just the beginning.

See if there's a "Precision" option that you can turn on in the scope's settings/configuration menu.  Agilent added this feature to one of the later firmware releases for the DSO/MSO6000 scopes, which was extremely awesome of them.  It allows FFTs to be performed on the entire acquisition record rather than just the screenspace data buffer. 

It may not be included or relevant in the newer-generation models, but if it is, you should familiarize yourself with it.

I looked at the impact of using hi-res mode and how it may impact FFT performance. What seems to be hidden or at least hard to understand - is how the system is using it's memory. I find myself trying to figure out how long will it take to fill up the modest 4Mpts of memory at a particular sample rate. It would be nice to be able to specify how much time I need to capture and the sample rate would adjust even if I was on a much faster timebase. Or, if I am looking at a span on the FFT - the scope would adjust to an appropriate minimum sample rate. These may be the wishes of an newbie beginner though.

Curious if anyone knows if it would be possible to measure output power of WiFi, Bluetooth, etc - what instrument is needed for those measurements?

[BFX]

Curious if anyone knows if it would be possible to measure output power of WiFi, Bluetooth, etc - what instrument is needed for those measurements?

You need an HF Power Meter  ... little expensive unit.
For common usage is enough homemade power meter based on AD8318.
Really nice base part is this Chinese board from eBay.
http://www.ebay.com/itm/142020799172

And congratulation MSOX-6004A is really nice piece of gear.

[Berni]

You certainly can look at wifi and other 2.4GHz stuff. If you a are interested more than just tx power but actually demodulating it, then you might want to look at Keysights VSA software. It can be used to turn my MSO9000 in to a vector spectrum analyzer with bit error rate meshurement and everything. It runs the oscilloscope data trough FFT to do it, but it gets better performance when run on a actual spectrum analyzer(still uses FFT but works better because the signal is downcoverted and digitized with a high resolution adc)

I don't know if it can work on your scope but it's worth a look.

[KE5FX]

I looked at the impact of using hi-res mode and how it may impact FFT performance. What seems to be hidden or at least hard to understand - is how the system is using it's memory. I find myself trying to figure out how long will it take to fill up the modest 4Mpts of memory at a particular sample rate. It would be nice to be able to specify how much time I need to capture and the sample rate would adjust even if I was on a much faster timebase. Or, if I am looking at a span on the FFT - the scope would adjust to an appropriate minimum sample rate. These may be the wishes of an newbie beginner though.

Hi-res mode isn't the same thing.  If there's a "hi-res mode," and if it works like the one in the previous 6xxx scopes, then it trades off aliasing protection for a few more simulated bits of ADC resolution, and has no useful effect on the FFT display.  You're better off ignoring it.  The "precision" setting, on the other hand, decouples the FFT input size from the screenspace waveform buffer.

On the MSO/DSO6000 series, the FFT rate and length is determined by a combination of the horizontal time/div setting, the selected FFT span width, and whether the 'precision' control is turned on.  What I usually do is enable the precision feature, assuming I care more about the resolution and noise floor than I do about acquisition time.  Often that's the case, but not always.  Then I set the desired span width and adjust the time/div control for a good compromise between bin density and acquisition time.  Finally, if I need averaging to reduce the noise floor further, I'll mess with that.  This may all work entirely differently on your model, of course. 

Curious if anyone knows if it would be possible to measure output power of WiFi, Bluetooth, etc - what instrument is needed for those measurements?

You can probably do this but it may not be straightforward. 

Ideally you'd select an equivalent "resolution bandwidth" or "analysis bandwidth" that's wider than the modulation type you're looking at, so that the entire waveform (or at least the majority of it) is represented in a single FFT bin, or more precisely within the response of a flat-top window.   The amplitude can be translated directly into signal power based on the load impedance.  With most traditional spectrum analyzers you don't get resolution bandwidths on the required order of 100 MHz, so you'd typically look at the amplitude in a narrower bandwidth and scale it up to the signal bandwidth by 10*log10(signal BW / resolution BW) dB.  This is made possible by the fact that modern modulation types look like noise, but it's also complicated by the same fact.  The effective noise bandwidth isn't quite the same as the nominal bandwidth rating for most real-world filters, whether they're physical filters or FFT windows.

Even if the scope won't let you position your entire signal in a single wideband FFT bin, you could just capture a burst in the time domain and convert its amplitude into RMS power based on its crest factor without bothering with the frequency domain at all.  The trouble with that is that you'd need to know the crest factor for the modulation type in question.  Hopefully the scope can do the RMS conversion for you, but you'll still need to reconcile your measurement with the relevant regulations and standards, which will be written in terms of mask shapes, duty cycles, antenna gains, and other stuff.

So, the TL;DR is that you'd want a software package that knows how to measure occupied bandwidth and power for the specific wireless industry standard you're working with.  Capturing the raw data is only the first step.  You don't need a dedicated power meter, given a sufficiently-killer scope, but you do need to think carefully about how to interpret the data you get.

[Wuerstchenhund]

While I don't know for sure, it seems the X6000 is much easier to get discounted relative to the S-Series. The MSRP's seem rather close and the S-Series has a winning front end and gobs of memory. Hard to say if the slower update rate would be a problem or not.

The DSOX6k gets pretty large discounts simply because it's a tough sell, mainly because it's not exactly the type of scope that people expect in this price class. It's main strength is the waveform update rate but that's generally not a critical factor in this price class.

I looked at the impact of using hi-res mode and how it may impact FFT performance. What seems to be hidden or at least hard to understand - is how the system is using it's memory. I find myself trying to figure out how long will it take to fill up the modest 4Mpts of memory at a particular sample rate.

If I remember right FFT on the DSOX6k doesn't use above 1Mpts of memory.

You certainly can look at wifi and other 2.4GHz stuff. If you a are interested more than just tx power but actually demodulating it, then you might want to look at Keysights VSA software. It can be used to turn my MSO9000 in to a vector spectrum analyzer with bit error rate meshurement and everything. It runs the oscilloscope data trough FFT to do it, but it gets better performance when run on a actual spectrum analyzer(still uses FFT but works better because the signal is downcoverted and digitized with a high resolution adc)

I don't know if it can work on your scope but it's worth a look.

The DSOX6000, like all DSOX models, is an embedded scope (I think running WinCE) on which you can't install 3rd party applications. You'd need an Infiniium scope (like the MSO9000), which is Keysight's Series of scopes running desktop Windows, to install the VSA software directly.

The DSOX6k can still be used with the 89600B VSA software if its installed on a separate PC. Performance will be limited though as there's no way to get real-time data out of the scope fast enough for streaming (it's limited to USB 2.0 and a not exactly fast 1000Mbps Ethernet interface).

[memset]

The DSOX6k gets pretty large discounts simply because it's a tough sell, mainly because it's not exactly the type of scope that people expect in this price class. It's main strength is the waveform update rate but that's generally not a critical factor in this price class.

I'd prefer high-speed interface (PCIe/10ge, etc) decoding / trigger for a 6GHz-class scope.

[Wuerstchenhund]

The DSOX6k gets pretty large discounts simply because it's a tough sell, mainly because it's not exactly the type of scope that people expect in this price class. It's main strength is the waveform update rate but that's generally not a critical factor in this price class.

I'd prefer high-speed interface (PCIe/10ge, etc) decoding / trigger for a 6GHz-class scope.

Yes, I'd too. I'd also want decent FFT (>25Mpts), a wide range of triggers and search tools, and some serial data/signal analysis tools at least as option, and MathLab integration (on the scope).

FFT on the DSO-S is very nice 

[rx8pilot]

If I remember right FFT on the DSOX6k doesn't use above 1Mpts of memory.

Ouch. Where would I find that data? I cannot really figure out what is happening under the hood.

I have been binge reading/watching about FFT and SA's over the weekend. While I am still under-educated on the topic, it appears that the FFT on the X6000 is rather limited. The good news is that I don't really need it for what I am doing right now. What I would like, however, is to find an SA/Signal Analyzer/VNA etc that has a good triggers and a trigger output so that I can time correlate events between the time and frequency domain.

[Wuerstchenhund]

If I remember right FFT on the DSOX6k doesn't use above 1Mpts of memory.

Ouch. Where would I find that data? I cannot really figure out what is happening under the hood.

http://literature.cdn.keysight.com/litweb/pdf/5991-4087EN.pdf?id=2456396&cc=GB&lc=eng

Page 34:

Enhanced FFT:                Record size: Up to 1-Mpts resolution via precision mode

Not a lot 

I have been binge reading/watching about FFT and SA's over the weekend. While I am still under-educated on the topic, it appears that the FFT on the X6000 is rather limited. The good news is that I don't really need it for what I am doing right now. What I would like, however, is to find an SA/Signal Analyzer/VNA etc that has a good triggers and a trigger output so that I can time correlate events between the time and frequency domain.

That depends on what you want, i.e. BW, RBW range, RF performance, and budget.

[rx8pilot]

Enhanced FFT:                Record size: Up to 1-Mpts resolution via precision mode

I am not sure what 'precision mode' is - there is no mention of it in the manual. Curious if the 450k/s update rate reduced the need for large records to be stored for FFT. Even though it's taking a small bite of data - it's taking a lot of bites?

The only source of high frequencies I have right now sends out 1.485Gbps data. I was curious how the FFT would do with this. After adding some filtering (averaging), it was easily able to see the 3rd harmonic at 6Ghz. It looks like it will be a useful general purpose tool, but I doubt it will replace an SA as I advance my studies. Being able to get a quick spectral picture out past 6Ghz.

That depends on what you want, i.e. BW, RBW range, RF performance, and budget.

6+Ghz, with some analysis capability and tracking gen. $5k to $15k (obviously looking at used options, but not decades old stuff). Right now, probes are by far the higher priority.

[rx8pilot]

....wellllll look what I just found. The 'PRECISION' option that feeds more (much more) data into the FFT engine. The default I think is around 64kpts for the FFT and the precision takes that to 1Mpts. No wonder I was struggling to see anything! Much better! The downside is that is gets rather slow. Finding glitches in the frequency domain is not going to happen at this speed. This is memory dedicated to the math functions and is a sample reduced data set from the main acquisition memory. I guess that is why it gets slow.


Pic1: Precision off, not enough data for FFT.

Pic2: Precision on, but at 100kps - nothing

Pic3: Precision on at 1Mpts has useful amount of data available to accomplish FFT.

Now I am looking at the various measurements, filtering, peak detect, etc, etc not that I can get some spectral data.

[KE5FX]

....wellllll look what I just found. The 'PRECISION' option that feeds more (much more) data into the FFT engine.

Yes, that's what I just explained earlier.  It was a big help when Agilent added this feature to their older 6000-series models, but I doubt that even 1% of their customer base has used it or even noticed it.

Going to get a bit preachy here  and suggest that just about everyone in this thread spend some significant time studying basic digital signal processing theory in general and Fourier transforms in particular.  It's a shame to see people criticize an incredibly awesome piece of gear because it doesn't support a "25Mpt FFT" or some other such ridiculousness. 

Especially for those who have been given free access to these scopes, please take the time to learn what these tools can do, learn how they work, and learn as much as you can about how they're implemented. 

Seriously.  This is some amazingly cool stuff, and it will only get better as your understanding increases.  You guys with these scopes have been given a kingly gift.  Don't waste it.

[rx8pilot]

Yes, that's what I just explained earlier. 

That is why I went on the hunt for it. If you said nothing, I would not have stumbled across that any time soon. I did find it in the manual and better understand what it does. Great tip - the FFT has much more limited use without it. The extra data does slow it down to the point where it only updates/sweeps once every few seconds - limiting its use. I don't think there is a magic menu for that. This precision feature is not at all obvious to me, thanks for pointing it out.

Going to get a bit preachy here  and suggest that just about everyone in this thread spend some significant time studying basic digital signal processing theory in general and Fourier transforms in particular.  It's a shame to see people criticize an incredibly awesome piece of gear because it doesn't support a "25Mpt FFT" or some other such ridiculousness. 

Especially for those who have been given free access to these scopes, please take the time to learn what these tools can do, learn how they work, and learn as much as you can about how they're implemented. 

Seriously.  This is some amazingly cool stuff, and it will only get better as your understanding increases.  You guys with these scopes have been given a kingly gift.  Don't waste it.

That is why I have been binge learning on the FFT topic the past few days. While that is only enough to scratch the surface, I have definitely covered some ground. The new surface knowledge has left me wanting more but I have to get back to revenue generating work now unfortunately. The more daily features have been fantastic. I was able to load up all 16 channels of the logic analyzer and the 4 analog channels to examine an entire fault sequence on my power management product for the first time. Being able to time correlate analog events with the i2c data is such a time saver. It is so much easier to look at actual timing from a fault condition to uC interrupt and the i2c responses during that time.

There are a few scenarios where the system I am working on is locking up or dealing with a short period of data errors and I suspect some glitches are on the i2c lines or possibly the 5v power rail. This is a power management device with a lot of high-current current switching and DC-DC conversion. Tons of opportunity for the delicate analog areas or i2c to couple some stray EMI.  This may be a good chance to take advantage of the fast update rate to catch a one time problem. To stimulate the problems, I have to set up a sequence of external problem scenarios and don't want to miss the problem when I finally get it to happen. The plan is to record the good i2c in the AWG and modify it with various problems so that I can work out various advanced triggers while playing it back. Once I feel like I have the triggers setup I can start trying to stimulate the problems - which could take a long time. Being able to see a dozen elements at the same time should help isolate the root  cause. Learning the details of the instrument is the only way to sort these kinds of things out.

This instrument will not be wasted at all.

[tautech]

I was able to load up all 16 channels of the logic analyzer and the 4 analog channels to examine an entire fault sequence on my power management product for the first time.

That we MUST see. Screenshot please.