TSP #27 - Experiments and Demo of an Agilent DSA-X 96204Q 160GS/s 62GHz Oscilloscope

[@EEVblog]

They are usually quite involved to take apart, & I wasn't going to risk it on mine. Mine was only worth $140K, the $500K model probably doesn't exist in Australia as a demo unit. Usually you order one and wait 3 months for them to make one for you from scratch. I had it sitting on the back seat of my car to get from the home to the lab because the transport box was too big to fit in my car boot, scary when it's worth 80 times what your car is worth.

[@Thesignalpath]

Essentially, yes. As the temperature of the electro-magnet rises, the resistance of the wires increases. This in turn causes less current to flow through the coils at the same constant voltage provided by the power supply. Since the YIG oscillator is tuned with magnetic field (thus as a function of current), the frequency shifts.

[@Thesignalpath]

This it is the scope equipped with the fastest real-time sampling ADCs at 80GS/s. LeCroy uses 40GS/s ADCs with 4-band block down-conversion and DSP processing to achieve an effective sampling of 160GS/s. Agilent does this with only 2-band block down-conversion and 80GS/s ADCs. This makes a difference in performance (aside from noise and ENOB). I hope this helps.

[@Nermash]

This was really stretching my very modest knowledge of digital signal processing to the limit... Pure unadulterated fun and by far and long the most expensive scope I have ever seen. Nice to see that Agilent still has that screaming madness quality:)

[@1hdsquad]

Oh my god. Ive got several oscilloscopes and power supplies and stuff at home, I study electrical engineering, I earn money with pcb design and programming, I can pretty much fix everything around, but this... Thank you very much for this video, making me feel very very small ;) I liked the open bottle directly next to 0,5*10^6 $ at the end.

[@gamccoy]

This was the BEST hour I've spent all week. I loved this video! I actually liked the last experiment more than the scope demo, though. In the end, all the scope did was capture data for off-line processing, which is nice but... I am very interested in the DSP lecture and wish you could spend some time on that topic but at a level most of can digest. Thanks again. You never disappoint.

[@JimGriffOne]

I had to read the description more than 5 times to make sure I was reading it correctly!

[@billmoran3812]

Awesome! It was just as amazing to me as the first time I saw a triggered scope in operation. (Tektronix 545). That was in 1965. El chronic sh as been quite a ride over the past 50 years.

[@NoahFect]

Quick tip: when you hook up two instruments with a piece of coax, short the coax center pin to its shield with a pair of tweezers or a piece of wire first. The coax dielectric can store enough charge to cook one of those 80 Gs/s samplers to a toasty crisp. Or at least make the connection to the cheaper box first. :)

[@rapsod1911]

FFT is calculated on taken data so if taken data is not ok FFT can't be ok. 500k scope, priceless presentation :)

[@dBm_drillingbits]

Nice material Sahriar. Big Thumbs Up as the Aussie guy says!

[@Thesignalpath]

The description is now added to the KZbin video also.

[@mevmevmev]

Did you consider renting a fog machine and running a tube behind the scope? Then press the footswitch just after you make a connection to the front panel..

[@douro20]

If I remember right, Agilent has a joint venture with X-FAB which manufactures the ASICs in Germany.

[@Thesignalpath]

Don't get me wrong, LeCroy scopes are not bad, but in terms of sheer performance, I have observed clear advantages with the Agilent model.

[@Thesignalpath]

Perfect! Thanks for posting this. I hope your comment gets up-voted so other people can read it also.

[@BarriosGroupie]

I hope you do more stuff to show off this amazing instrument.

[@Thesignalpath]

LeCroy uses Digital Bandwidth Interleave (DBI) technology, and does not actually sample at 160GS/s. I have used their scopes and did not get the same performance as I have seen on Agilent.

[@electronicatutorial]

This is just Agilent kicking some butt! in response to the HD 65Ghz from Lecroy they introduced not long ago.

[@Razor2048]

Will Agilent ever consider having a sale or a coupon code that will allow us to get one of those scopes for under $1000?

[@yoramstein]

What about a tear down? Lets send it to Dave Jones :-)

[@urdnal]

Bet you this one won't be "a little how ya doin' "

[@Thesignalpath]

That would be very impractical. The instrument processes a minimum of 2.56Tb/s of data! What computer is going to process that data for you? All the processing (and buffering of the data) is done by FPGAs and DDR4 memory on the acquisition boards.

[@CxC2007]

I remember that I saw this equipment in a video from Altera, explaining the goodness of their transceivers, while showing an eye diagram

[@ronaldlijs]

Fantastic video! Keep them coming, I liked your previous one around connectors, GREAT STUFF!!!

[@1337NoMad1337]

Wonderful video. Plus you did exactly what I hoped for in the end :D Keep up the good work, and thank you for sharing your knowledge!

[@chrislemus9392]

Im very fortunate to get to utilize this piece of equipment where I work.

[@Galfonz]

Now that's how to show off an instrument. Very good video.

[@ve2um]

The inputs go directly to Agilent's custom built BGA chips.. Hey boss, i've connected you brand new DSA-X 96204Q scope 62 GHz input to the Hydro power to watch the waveform... It made a small poof and i don't see any 60 Hz waveform on the screen. Are you sure there's power into that outlet ?

[@ayrendraganas8686]

the safety circuits to protect the hardware from that would reduce signal quality.

[@Thesignalpath]

It is not a matter of 'not good enough'. It is the wrong instrument for such an experiment. The Agilent PNAX, for example, would be the right instrument to investigate the differences between connector grades.

[@Richardincancale]

Hi Shahriar - glad you're back! Please can you walk us through the Matlab code you demo'd? It's one of those packages that's so hard to get into without some concrete examples like this.... Thanks again!

[@Thesignalpath]

That would be a first for Agilent. I could not find any documentation for this. If anyone knows, please let me know. If Agilent is doing this, they are doing it better than LeCroy (based on my experience). Also, they are doing it only once (two band combining).

[@Thesignalpath]

Sorry... Have been very busy. Rest assured the channel is not dead!

[@rfengr00]

Did it have the 89600 VSA software on it? That would be cool to try with 65 GHz of bandwidth. I got to examine one of the InP MCMs when Agilent gave a demo of this scope. Very impressive.

[@assadij]

Shahriyar. In your matlab section, to what you were referring to as complex conjugate, it is in fact only aliasing effect. The FFT should only be represented up to nyquist. ;) Cheers.

[@fr0zenst4r]

drinking a nice brown ale and watching futurama on a half million dollar oscilloscope... dude U LIVIN THE LIFE.

[@smunaut]

You have the IF->RF leakage meaning you get the PRBS spectrum at baseband as well, then the LO->RF leakage meaning you see the tones as well on the scope. And then you have the LO harmonics meaning you get those spectrum at all multiples of the LO freq.

[@matrinGD]

That is some awesome equipment and a very good presentation. I do have some follow up questions. You mention that the oscilloscope software filters the signal above 62GHz. Does it use analog hardware filters to filter above the Niquist frequency as well? I could not spot any on the diagram. But I don't have any electronic components knowledge, so I probably missed them.

[@slap_my_hand]

i'll just keep saving money for a 1054Z.

[@Ziplock9000]

Mine arrived today. Best kit I've ever bought. Price / Performance ratio is insaine

[@mixolydian2010]

Thanks a lot its an amazing thing where would these machines be mostly found my little brain cant imagine, Thanks again for the videos you are definitely pushing my boundaries...

[@JorgeRodriguez-kw7pc]

1/2 Million dollars TV set !!!!! Juajuajua Great show and Tell !!! Careful "Agilent commandos will drop from the ceiling if you broke this break the seal"

[@andrewmcneil]

That is an impressive piece of equipment. Now if I had half a mill, scope or Ferrari mmmm! Thanks for sharing

[@madinatore]

Change in Frequenzy due to a change in termperature that caused a change of resistance in the magnetic coils? Couldn't follow all of the content you delivered, but does it have any sort of theft protection?

[@rotlerin]

High end oscilloscope - high end beer. Newcastle Brown . . . . . wicked!!

[@johnyu2799]

It is amazing that 64Gsps is a reality!!! effectively baseband goes up to GHz range. You have demonstrated the amount of data can be directly demodulated! The only limits are the power and memory needed!

[@opablo_gm]

How can you get your hands on this kind of equipment ? Is Agilent letting you play with this for a few days so they get free(damage risk) internet presence for their brand ? What could have happened if you dropped it ? Are there ensurance services for this kind of stuff/situations ?

[@ScottieNiven]

EEVBlog had one like this and agilent told him he was not allowed to take it apart. He just took the top cover off and had a quick look inside.

[@ArthurBenemann]

Awesome video! I wish you where the professor in my communications class.

[@nlhans1990]

Isn't the 'harmonics' in the 60GHz region a second product of the DSB mixer, like 2*fc +/- fm, and 3*fc +/- fm, etc. albeit with decreased amplitude?

[@tmmtmm]

The extra signals on the FFT would be harmonics and the 3.125ghz signal making it to the output, right?

[@Plutonion2]

HI Shahriar .Nice to know your still about : ) .:I got withdrawl symtoms from lack of your fabulous videos .

[@smunaut]

So whic of the 3 kind of RF connectors did you use for this ? :) And what connector is on the YIG ? (that looks like SMA but I guess it's too high in freq could be any of the one with the same 'thread').

[@baroque_engineer]

"Oops, what did I do?" *FFT image disappears* - that was the most scary moment in a whole video. :)

[@KLATUBARARA1]

Thank you so much, that helps me with my Ethernet PHY. Ecelent docu.

[@Thesignalpath]

Nope, not dead. Just very busy... I'll try so make my next video this weekend.

[@ToniT800]

what was in the description????

[@ahmedgaafar5369]

Very enlighting

[@OneCoolDude08]

What do you think of Rhodes and Schwartz scopes?

[@NoahFect]

He wanted to do an inverse FFT to get the (real-only) time-domain data back, though, so he needed to mimic the conjugate mirror effect imposed by the original forward FFT. He could also have used a Hilbert transform to move the whole problem to the complex plane, but this was probably easier.

[@TheRadeonVideo]

This thing can do more GS/s in realtime than my tek scope in equivalent mode..

[@Zeratuhl]

I think he means implementing it in the form of pci-e cards and just talking to the card via pci-e for easier display and usage on an existing Win7 workstation. All the FPGAs and memory for processing data could still be on the board, similar to how pci-e hdmi capture cards do encoding right on the board.

[@chilldog1234567890]

Can you do a tear down of this scope?

[@vaualbus]

So said the fastest production scope so technicaly agilent or lecroy have some r&r scope more fast that that, or some prototype?

[@GregoMorgan]

A single probe for this thing costs three times as much as the oscilloscope I use.

[@AlexForencich]

That's what I figured. I suppose running the power supply in constant current mode would prevent that from being an issue.

[@Thesignalpath]

It was lent to me by Agilent.

[@vaualbus]

Just a question, How you get the scope?

[@NerdNordic]

Is the frequency shift due to the magnets warming up?

[@walterharriman]

Holy wow. I'll just keep saving for my Rigol DS1074Z...

[@srscricket]

I am drooling uncontrollably.

[@OneCoolDude08]

Thanks for doing the math for us!

[@gorak9000]

Ooh, I got really excited when you said you were going to use the scope as the SDR to recover the data, and then you used Matlab :`( I still wish you were the prof that taught me DSP - would've been a whole lot more useful course. Funny thing is, I'm actually writing GPIB software for another scope in the Infinium line to automate some data collection for my MSc. project. It's amazing how the actual scope software looks identical to our scope, except ours is only 3Ghz and 8Gsa/s, and runs win98

[@gkasprow]

I also hoped that the scope can do it itself :)

[@vaualbus]

In theory you could rune matlab on the scope?

[@programagor]

So very cool. Given that it runs Win7, could you install MATLAB onto it and do the SDR directly, without the need of transferring the waveform to your PC?

[@AureliusR]

Without looking at the other comments... because the YIG oscillators are magnetically tuned, and contain large coils of copper wire (from what I've seen), and since the current to tune them ends up being a couple of watts, I imagine the frequency shifting is because the copper is heating up and thus changing its DC resistance, changing the amount of current that is flowing and thus the tuned frequency. I guess this is why we always see them as part of a some sort of feedback system (like a PLL) to help compensate for these changes over time?

[@Thesignalpath]

Sounds good! :)

[@AureliusR]

@@Thesignalpath So I am correct? :D Are there any other effects?

[@OneCoolDude08]

I remember seeing equipment that costs $27,000 in the 1987 Tek catalog on eBay now for about $60, lol. I would be willing to bet in 40 years this scope will most likely be worth less than $1,000.

[@warrentb1]

So, how long until a hobbyist can pick one of these up on eBay or future equivalent for a couple hundred space dollars?

[@commodork]

Now I know what the Nissan GT-R of test equipment looks like.

[@Thesignalpath]

You can buy 5 Nissan GT-Rs with the price of one of these scopes. :)

[@kalhana_photography]

more like F1 car of test gear!

[@daveblane6442]

+commodork -- a what??

[@michveldvvid]

im glad you didn't say nissan skyline lol

[@DavidLeeMenefee]

Very nice, thumbs up!

[@vaualbus]

On the not high frequency input, How is lower frequency you can apply to that? Of course the input are 50ohm impedence.

[@EEEAmin]

hai man, thanks for the rep..cause every 1/2 day i go to you channel to see whas new......cause you know so much about RF stuff can i ask what was your PHD about and is there any way that you can make a video about Network analyzer

[@warrentb1]

the funny thing is I bet a couple guys with big mustaches in the 70's said the same thing about a piece of drool quality test equipment that is currently available on ebay. I'll give it a couple decades, though since its so rare it'll probably be thousands of spacebucks for collector quality alone

[@code123ns]

So glad that I got that one right... and I'm a software developer guy ^.^

[@AlexForencich]

Well, you would see a difference, but it would not be nearly as obvious as it would be with an instrument more suited to the task. The measurement would also be more difficult to perform. A good VNA can have around 100 dB of dynamic range. A scope like this one has an ENOB of 7 or 8 and so has a dynamic range of around 40 dB. Also, a VNA has a built-in signal source so it can sweep across its entire frequency band and measure the response accurately. An external source is required for a scope.

[@CxC2007]

How many of this units does Agilent expect to sell per year ??, I guess not many... telecommunication industry ?

[@Thesignalpath]

I would have loved to! But alas, it was not possible.

[@EEEAmin]

Hai is there any way ..you can make a video about Network Analyzer

[@npiltan]

is it available in university?

[@tmmtmm]

maybe also intermodulation distortion?

[@Thesignalpath]

Oh now you tell me! ;)

[@rfengr00]

Did it have the 89600 VDa

[@chilldog1234567890]

how could you afford all those gears?

[@bacphan7582]

when your scope has higher BW than your Spectrum analyzer

[@richardgray8593]

Did you buy this, or is it a loaner?

[@TMuel1123]

Dont switch it on, take it apart! This box is just awesome.

[@AlexForencich]

And that's without all the additional options. It's another $51,099 to unlock 2 gigapoints of memory per channel, for example.

[@ahmedgaafar5369]

Hi Sir, Is there any possibilty to share the matlab code so i can study how did you recovered the prbs code in detail....thanx in advance.

[@cocacolarrrrrrrr]

Is there any way to export the time domain data of the waveforms?

[@Thesignalpath]

Yes of course.

[@cocacolarrrrrrrr]

How to export the data? I was using the DSA90604 digital signal analyser

[@helmut666kohl]

Having an open beer in front of that beast and weaving your hands around… Like playing paintball in the Louvre :-) Cool video though - where is the teardown?

[@BalticLab]

Would you mind sharing your Matlab code?

[@anirbansengupta1693]

Can this oscilloscope measure EVM or plot constellation diagram of a modulated signal?

[@Thesignalpath]

Of course. With the Keysight VSA software suite.

[@anirbansengupta1693]

@@Thesignalpath That's great. Does it come with different license package and we have to pay for it separately or it is integrated with it by default?