*A minor correction to the video: The maximum hardware sensitivity using the InP input preamplifier is 7.5mV/Div at 256GS/s and using the SiGe input is 4mV/Div for the 128GS/s models. The remaining displayed vertical levels are DSP processed. The noise measurement is of course still correct and valid as are the shown ENOB and linearity numbers.*
The designers of the front-end amplifier want to make sure that no spec is over-stated. The sign of good, professional and ethical engineers. I applaud them for that.
256 GHz = 1/4 nano-second, sorry, light travels 3 inches in that 'TIME'. My brain says 'real-time' is just a phrase. [ built GHz+ tuned circuits and coded 'real-time' computer systems from '89 onward ] My ASD makes me say sh1t like that, great vid =]¬_D
Philip Rowney : nope :-) 1 GHz = 1ns (1 "foot", if you assume typical propagation speed) 256 GHz = 1ns/256 = 4fs. That's about one millimeter. (I'm ASD as well, but since I'm french, it's more like QSD hahahah) [sorry for the keyboard layout pun]
You know you've made it in life when Keysight will send you something like this. Not only the device but boards and decapped chips. Absolutely stunning.
ONE??? I need backups!!! Do ya think I can test my 555 timer prototype with this??? Or do I have to wait for the NEXT model? If I sell the house I could ALMOST afford to insure the first one... for the first year...
What an absolutely incredible engineering achievement. It takes a very special company with a wide range of talents and a wide variety of fabrication technologies/abilities to make a product such as this. I can only imagine the pride and accomplishment felt by the team who were fundamental to this device's creation. Great stuff 👍
10bit @ 113Ghz and 4 channels, I just can’t wrap my mind around that! Thanks for the video, no way I’d ever get to see one in use otherwise. I work with ultra low frequency stuff usually 10Ghz or lower so I guess that’s basically DC these days. Amazing.
That is by far one of the most amazing things I have seen on TH-cam. Being able to actually see that high frequency sinusoidal waveform, wow! That is serious engineering in all electronic voodoo disciplines. Thank you for doing this video, and you are correct, the engineers that designed this should be very proud.
I'm amazed that Keysight is letting the world see this equipment in such detail. Typically I would expect something so pricey and niche to be kept under lock and key. I can't believe all of that is required for a SINGLE channel of the scope. Absolutely incredible engineering. Totally worth every one of those 1.3 Million dollars.
You seemed extraordinarily comfortable waving that metal pointer around that million dollar oscilloscope's screen. If keysight had dropped something like that at my doorstep I would be afraid to even breathe near it.
I work as an engineer in a small company. Usually i do both schematics and layout (like i said small company). I would LOVE to some day layout a board like the digital processing board on this thing. It looks really fun to do. I wonder how long it took.
The PCB assembly alone is extraordinary. That A/D PCB must have taken a lot of planning and consideration for assembly, reflow, testing, etc. The design......obviously next level. Stunning opportunity to see technology like this.
Maybe a university or professor discount. Most of these high end instruments don't sell for MSRP, almost always negotiable for any business looking to aquire one.
Shahriar, every time i sit at my desk i feel good about what i have there about a house and a half worth of equipment and then i watch a video of yours.... takes me two weeks at least to recover from the shock
Man, that board design.. incredible. I'd love to just look at the routing. Of course the indium phosphide design is next level awesome, and look forward to other foundries opening this technology up. Huge props to Keysight...
You would not be able to get me to even look at this thing before i'm convinced that my ass is fully covered, insurance and liability wise :D I did not expect to be convinced that this machine is reasonably priced, and yet I am. Stunning stuff. Thanks for sharing.
Holy fucking shit, this thing is glorious. Every different board is a treasure of different engineering techniques. And kudos to keysight for the samples, I bet the people that worked on it would be happy to see someone gushing over it. And at this level you could take hi-res photos of each component and sell them to other companies, and yet they wouldn't have a shred of the technical know-how to produce this beauty.
Great video. Would like to see some applications which have demand for this type of equipment. Hope KS lets you have it again for some experiments once it is in production.
In my lab we have two URX Scopes 128GSps, they are not as fast as this one on the front-end side, but still impressive (the fastest is 33GHz). My opinion after 2 years of use is that they are undoubtedly amazing machines with excellent front-end and DAQ hardware. However (and I honestly hope that some Keysight colleagues will read this comment), I would have expected just a little bit more on the software and UI in general. I'm not saying that is bad, don't get me wrong. But we are talking about million dollars machines, I honestly would like to have the absolute best. In particular, I noticed that the stability should be improved, expecially while CDR or heavy measurements/math are used. In several occasions I had to forcefully restart the OS due to hangs. Furthermore, I'm sure that the UX (including responsiveness and even interface style) could be enhanced: controls are slightly less prompt than on the old infiniium scopes (which I loved, especially the 20 years old ones with that DAQ to GPU waveform bus
@@Sebas_Cba_82 I reply in English. In my team we develop solutions for high speed data links, including design of serialiser/driver ASICs, Si Photonics modulators and the relative integration modules. We currently target about 26Gb/s per lane. Hence, the UXR represents a valuable tool for the characterisation of signal integrity in such devices. We mainly use it for eye diagram testing and wide band phase noise analysis. These are quite common applications of the instrument. If you work with COTS transceivers, it is generally not required to own it, since those are originally qualified by the manufacturer. Good SI design methodologies and simulations are enough in that a case. The RTA features are instead very interesting in a variety of designs and test-benches involving RF signals. However, the purchase of a spectrum analyser could be discussed if time domain analysis are not required, this would probably allows spending a bit less.
I think that was the one I saw at the NIST antenna lab, along with a down-converter from Rohde Schwarz to bring the antenna's transmitted frequency down to something the Keysight scope could see. I think they said they could do work up to 300 GHz.
No word of boot time, fan noise or tap test on the inputs. Also I bet serial decode will be $800 optional extra! So I’m holding my purchase for now. :P
Just for some perspective.. you can capture light traveling less than 1mm with this puppy. That is a fast time domain. Useful at CERN. Probably overkill for weapons research :)
If you're still around in 25 years or so, Siglent will put this on your bench for $1M, which by then will the equivalent of $1k in today's money. But for now: wow.
plemli it will likely be cheaper, we are making major leaps and bounds with nano technology like graphene and carbon nano tubes.. getting to the point of making transistors.. in 10 years silicon may be outdated
Have you heard of the nano tin particles that are room temperature super conductor? And they are thinking with fluoride added into the matrix they can get it to be super conducting up to much higher temps
My impression of how the boards for such a high-end instrument would be made is this: Each board coming from the manufacturer would be X-rayed first for an initial visual inspection, and then would be passed through a flying-probe machine to check continuity of the traces. After the flying probe machine, the boards would then be placed in a thermal cycling oven and cycled for several hours and then tested again on the flying probe machine to ensure that the boards were manufactured properly and will have very high reliability over the service life of the instrument. Only then will the boards be passed through the silk screen presses and to the pick-and-place machines for the placement of small and dense components. The pick-and-place machine would be a high-reliability type like that made by Essemtec which has sub-micron alignment accuracy over the entire area of the board. To achieve this the machine would not only align to the fiducials, but also scan key areas of the board to account for any error in pad alignment before placing the components. The boards would then go to the first manual inspection, where the alignment of all of the components placed by the machine is verified and any missing components are placed manually. Boards which pass will then be manually populated with larger surface-mount components such as relays, coils, connectors and hybrid components, and then passed on to the reflow oven, which uses an inert atmosphere to mitigate any oxidation which could occur during the cycle. After reflow the board will then be sent for another manual inspection. At this point any through-hole components will be installed manually and soldered by a robot, after which the board will receive a further inspection. The boards which pass will then be throroughly washed and sent to a special facility with tight climate controls for integration of chip-on-board components. The process of integrating the chip-on-board components will be done entirely by hand, with each chip being secured with thermally conductive epoxy and then connected by wire-bonding. After the components are installed the boards will then pass through two independent inspectors, each of which will conduct the same integration inspection, and both must pass the job before the completed assembly will be encapsulated using laser welding. Each board after the requisite steps are completed will pass through several stations, each of which will conduct individual board-level checks to ensure that each board functions exactly as required. The individual boards will then be installed in a test instrument, where an initial calibration will be performed and operational checks carried out by hand. Sample boards from each run will then be burn-in tested for a full week and inspected once more before they are sent for integration into the product for sale.
Today I am trying to routing my first bga, with lpddr4 memory that works at 1.2GHz and TH-cam recommended me that video like "ha ha ha" watch what really cool engineers do.
And how do they test the 110 ghz signal generator when they need the signal generator to test the oscilloscope but also need the oscillopscope to test the signal generator?
my question is similar, do you already have a better specs scope or instrument that was used to design this and is just for internal use? or develop this using the same tools that you already have for sell?
He couldn't flip it anyway. These things need all kinds of support to be of commercial use. And the publicity he gives it is well worth one broken piece.
I'm new to electronics and am using analog oscilloscopes to do "Oscilloscope Music''. I would love to know how the high sample rate on this digital scope performs on this task. This is a beautiful piece of engineering. Thank you for showing us.
Question on the calibration value measured The scope shows a value of 3.3ps for the fall time. This means the scope is sampling the edge much better than the 256GS/s spec. If it sampled the edge correctly using the 256GS/s speed then the 2 sample values are 3.9ps apart. So how is it able to measure 3.3ps for the fall time for the edge? Am I missing something?
isnt the spec for fall time the time it takes for it to fall from 90% to 10%? so if the signal goes from 95% to like 5% you could probably extrapolate that the fall time is quicker than the sample rate? also you can use multiple aquisitions to cheat your raw sample rate if your sampling timing is precise enough
Normally I would not think of a scope as a phase noise measurement system, but with the jitter analysis package I am thinking this could measure close in SSB phase noise on mmWave sources. Any thoughts on this?
Wow! What a absolute beast. Guys, if you want to see things even clearer then you can use a magnifying glass in front of your phones screen. It works really well. Thanks for the awesome video. =)
As I've got, it's just main low freq clock at 8 GHz (low noise PLL clocked by low noise OCXO), and higher clocks are formed just by cascaded doubling. This results, in theory, the ideal timings between cascades, by the definition of harmonics. The main comlicated technical issue is alingment of delays between cascades, due to the finite (and, in this case, actually very low, lol) speed of light. This results to extremely high geometric requirements to machined parts and traces.
28:00 Interesting to see that the base of these dont seem to be ceramic but ordinary PCB material? Dont think I have seen that before, but it would absolutely be cheaper though.. for them that is.
I am rewatching this video i guess for the 4th time as my company about to buy one :D I was wondering, how did they manage to isolate all 8 lines per ADC in the frontend module. They look like they are laid out very very closely. Any ideas?
I'm interested how the time domain waveform looks like after the samplers. It is new to me that the sample and hold circuit of the ADC is not on the same chip with the quantizers. Does it look like as a series of step functions with different amplitude or impulses with different amplitude? Could you help me out with good article about it?
As I've understood, basic idea of operation is simple: This scope is running all the ADCs at some one low frequency, and, in fact, they are semi-directly digitizing one input. The magic is, that they are sampling with small equal delay (clock phase shift -- more correctly) between each next ADC. This complicated golden S&H circuitry is doing the job of fast and precised sampling with different phases at this one low frequency. Then, output sample stream is formed just by simple placing samples of each ADC one after one. Thus, answer to your question is very simple: Signal after each sampler looks as time-quantized copy ("staircase") of input signal, with sampling frequency equal to full sample rate divided by number of "phases". Why isn't it possible for these signals to look like stream of short pulses? -- Because duration of these pulses might be in zeptosecond range (duration of actual slopes), what, in my opinion, is truly impossible, and not neccesary for operation of each next stage of samplers and ADCs.
Do you know what happen to the data after the dsp? It go to the CPU motherboard and processed by the CPU or the data is overlayed on the screen and the gui is runner's by the CPU?
Someone might have asked already, but on 13:08 you mention two "distributed" amplifiers in each arm of the hybrid for the 64 GHz clock. Do you mean "distributed" between the hybrid's arms? I do not see a reason for a broadband amp to amplify a filtered 64 GHz tone.
Oh, an oscilloscope! I was thinking it was a spectrum analyser but had to re-read the title. Holy poo, that's insane! The rise time must be in femtoseconds or attoseconds.
The thing with these is that I wonder how you can be really sure that what you see is the actual wave form. With low end scopes you can always take a higher end to confirm.
There must be an enormous amount of reagent grade silver in that mechanical attenuator. It is possible that the CMOS ASICs are made by GlobalFoundries as they have worked extensively in the past with Chartered.
still shocked and awestruck! mind blowing indeed. Are those signal lines in the acq board SIW or guarded striplines? Although the dispersion of SIW might be unacceptable here... Also, does it go down to DC or is there a minimum frequency input?
Let me guess - you have to send it back. :) Excellent review. Don't forget to grab one of eBay in 30 years for $500 to stick in the Lab. Mind you, I've a feeling that getting spare parts for the "repair video" may be an issue.. :)
The lowest end of infrared is 300GHz, so yeah almost touching that transition into literal optical radiation. I imagine that they are approaching the physical limit to what is possible with copper on a board.
Shahriar, It's great to see such an in-depth teardown of the tool. How long did they let you "play" with it? Do you have the contacts with Keysight through work such that they lend you such equipment?
You can bet the gubmint has at least 30 of these collecting mold in a storage container in the tropics somewhere... likely originally used for ballast in a submarine.
*A minor correction to the video: The maximum hardware sensitivity using the InP input preamplifier is 7.5mV/Div at 256GS/s and using the SiGe input is 4mV/Div for the 128GS/s models. The remaining displayed vertical levels are DSP processed. The noise measurement is of course still correct and valid as are the shown ENOB and linearity numbers.*
Lol sounds like keysight sent you an email
The designers of the front-end amplifier want to make sure that no spec is over-stated. The sign of good, professional and ethical engineers. I applaud them for that.
I figured it was something like that. Much love sir keep up the amazing work. This one was UNREAL. Way over my head as usual but still love it.
256 GHz = 1/4 nano-second, sorry, light travels 3 inches in that 'TIME'.
My brain says 'real-time' is just a phrase.
[ built GHz+ tuned circuits and coded 'real-time' computer systems from '89 onward ]
My ASD makes me say sh1t like that, great vid =]¬_D
Philip Rowney : nope :-)
1 GHz = 1ns (1 "foot", if you assume typical propagation speed)
256 GHz = 1ns/256 = 4fs. That's about one millimeter.
(I'm ASD as well, but since I'm french, it's more like QSD hahahah)
[sorry for the keyboard layout pun]
Video reported for pornography
That analog front end is "sex on a stick "
NSFW... had to explain office-mate why i was drooling... dat asic thou...
So when does the 'giveaway' competition begin???
Lol
all that RF voodoo
Can I use this to trouible shoot my flashing led arduino project?
Yes, you can. And more...
that might be a little bit of overkilling it
Actually, you can't. It has only +/- 2V inputs.
not even a chance, i think you need something way better, to measure a 1hz signal
lol, it will probably show how the silicon in the led discharges.
You know you've made it in life when Keysight will send you something like this. Not only the device but boards and decapped chips. Absolutely stunning.
Its all relationships...
The entire list of potential customers could probably fit in a single coffee shop to talk about phase shifters in 110Ghz scopes.
And yet - we all want one! :)
@@trickyrat483 not want need. For what well that's a secret but I need one.
ONE??? I need backups!!! Do ya think I can test my 555 timer prototype with this??? Or do I have to wait for the NEXT model?
If I sell the house I could ALMOST afford to insure the first one... for the first year...
Tricky Rat I don't. That would be just plain junk to me.
Think about it. This scope could be an amazing way to measure whether a 9V battery still has some juice.
10 picoseconds - the time taken by light in a vacuum to travel approximately 3.3 mm
holy hell when you put it like that
but also the distance electrons move inside a conductor.
@@AnalogDude_ this is not true.
@@spdrfx it's true for a pure pure super conductor, CERN.
@@AnalogDude_is what environment would electrons move that speed?
I’d buy one, but I already have a 1054Z.
(sees ADC board back side) So that's where all the 100n capacitors went!
Thanks for the awesome video, love it!
What an absolutely incredible engineering achievement. It takes a very special company with a wide range of talents and a wide variety of fabrication technologies/abilities to make a product such as this. I can only imagine the pride and accomplishment felt by the team who were fundamental to this device's creation. Great stuff 👍
Nice sturdy-looking handles at the side of the instrument - to hold on to and keep you steady when you hear the price...
lmao. Those handles hold on to you :D
10bit @ 113Ghz and 4 channels, I just can’t wrap my mind around that! Thanks for the video, no way I’d ever get to see one in use otherwise. I work with ultra low frequency stuff usually 10Ghz or lower so I guess that’s basically DC these days. Amazing.
Give it a decade or so...
The video where Shahriar is literally nerdgasming for a full hour..... and so is everyone watching. Phenomenal technology!
Aliens don't have that sort of technology .
@@cuckingfunt9353 lets see what would be in new iPhones presentation...
Why did you blur out part of that ceramic package...???
That is by far one of the most amazing things I have seen on TH-cam. Being able to actually see that high frequency sinusoidal waveform, wow! That is serious engineering in all electronic voodoo disciplines. Thank you for doing this video, and you are correct, the engineers that designed this should be very proud.
I'm amazed that Keysight is letting the world see this equipment in such detail. Typically I would expect something so pricey and niche to be kept under lock and key. I can't believe all of that is required for a SINGLE channel of the scope. Absolutely incredible engineering. Totally worth every one of those 1.3 Million dollars.
I'm pretty sure they would bet their lives on it that on max 10 people on earth actually understands what's happening inside
You seemed extraordinarily comfortable waving that metal pointer around that million dollar oscilloscope's screen. If keysight had dropped something like that at my doorstep I would be afraid to even breathe near it.
I work with extremely expensive routers, totally normal. You‘re getting used to it. Two colleagues dropped a 250k router two years ago. It happens.
Wow, totally amazing teardown on a totally awesome oscilloscope. It's impossible to not be impressed!
I work as an engineer in a small company. Usually i do both schematics and layout (like i said small company). I would LOVE to some day layout a board like the digital processing board on this thing. It looks really fun to do. I wonder how long it took.
That is a truly beautiful piece of equipment. It is more of a piece of art work than any other equipment I have ever seen.
The PCB assembly alone is extraordinary. That A/D PCB must have taken a lot of planning and consideration for assembly, reflow, testing, etc.
The design......obviously next level. Stunning opportunity to see technology like this.
Imagine the path tracing nightmare at 110GHz....this is nerdvana
If engineers had a collective standard of beauty, this would exceed it by several orders of magnitude.
Next level GHz. :) I'd love to see some wallpaper sized (4K+ res please) photos of those RF parts and chips.
I'm speechless, with tears welling up.. What an impressive design.
Can I get a student discount on one of these puppies?😂
Sure, 25%. Your total is 975 thousand dollars please. ;)
Maybe a university or professor discount. Most of these high end instruments don't sell for MSRP, almost always negotiable for any business looking to aquire one.
You might need a small loan from your richt dad
Dude, I cried of this comment because in Ukraine I couldn`t get a student discount even for the public transport. F****** PUBLIC TRANSPORT, CARL!
I don't think they'll negotiate with me. "How about $100" "... are you serious?" "Okay fine, $1000, final offer" "*hangs up phone*"
And here I was feeling fancy with my 200mhz Siglent.
I am considering buying a 200 MHz Siglent. Are you happy with it?
@@mmaranta785 Yes its superb. I use it a lot both learning electronics and debugging arduino circuits thanks to the i2c and SPI decoding.
Are you sure you are allowed to show this before 10PM?
Or at the very least it should be age restricted..... ;-)
"The following video may not appropriate for all audiences..."
I need to take a shower after this video
@@KeysightLabs do you work with any universities in germany? I would love to work with and learn about all this equipment
You should have written "edge restricted" :-D
@@ayrendraganas8686 in Germany R&S usually has a better foothold
I wasnt expecting to see that monster in your bench that fast.
Only the best for Shahriar!
@@KeysightLabs I could use a little test equipment love. You guys have the best stuff on the planet.
pretty slick engineering no question.......hell of an analysis/demo too
Shahriar, every time i sit at my desk i feel good about what i have there about a house and a half worth of equipment and then i watch a video of yours.... takes me two weeks at least to recover from the shock
I'm going to wait until the Owon version comes out for $300
Probably gonna wait 10 or 20 years, lol.
//edit
Seems a little generous. maybe more like 30 or 40 years.
Being generous, I didn't expect this level of performance for another 2 or 3 years....
Hopefully equiped with at least 24bit A/D-converters.....
If input is 2mV- 500mV/div there is no need for 24bit A/D converter
All the hardware desgn is Keysight, but now a days, with all the piracy going on, you may be right.
Great video as always shahriar. Just could you please zoom in more when you are analyzing the front end circuitry.
Man, that board design.. incredible. I'd love to just look at the routing. Of course the indium phosphide design is next level awesome, and look forward to other foundries opening this technology up. Huge props to Keysight...
Thanks!
You would not be able to get me to even look at this thing before i'm convinced that my ass is fully covered, insurance and liability wise :D I did not expect to be convinced that this machine is reasonably priced, and yet I am. Stunning stuff. Thanks for sharing.
Can't wait for the full review and giveaway :)
Anyone know where I can download the schematic and PCB layout for this.
you wont see that for 20 years after it is obsolete
Thank you for making it possible for the rest of us to see inside and outside a scope of this possibilities and price range.
Holy fucking shit, this thing is glorious. Every different board is a treasure of different engineering techniques.
And kudos to keysight for the samples, I bet the people that worked on it would be happy to see someone gushing over it. And at this level you could take hi-res photos of each component and sell them to other companies, and yet they wouldn't have a shred of the technical know-how to produce this beauty.
Great video. Would like to see some applications which have demand for this type of equipment. Hope KS lets you have it again for some experiments once it is in production.
In my lab we have two URX Scopes 128GSps, they are not as fast as this one on the front-end side, but still impressive (the fastest is 33GHz). My opinion after 2 years of use is that they are undoubtedly amazing machines with excellent front-end and DAQ hardware. However (and I honestly hope that some Keysight colleagues will read this comment), I would have expected just a little bit more on the software and UI in general. I'm not saying that is bad, don't get me wrong. But we are talking about million dollars machines, I honestly would like to have the absolute best. In particular, I noticed that the stability should be improved, expecially while CDR or heavy measurements/math are used. In several occasions I had to forcefully restart the OS due to hangs. Furthermore, I'm sure that the UX (including responsiveness and even interface style) could be enhanced: controls are slightly less prompt than on the old infiniium scopes (which I loved, especially the 20 years old ones with that DAQ to GPU waveform bus
En que aplicaciones usas estos instrumentos con especificaciones tan especiales?
@@Sebas_Cba_82 I reply in English. In my team we develop solutions for high speed data links, including design of serialiser/driver ASICs, Si Photonics modulators and the relative integration modules. We currently target about 26Gb/s per lane. Hence, the UXR represents a valuable tool for the characterisation of signal integrity in such devices. We mainly use it for eye diagram testing and wide band phase noise analysis. These are quite common applications of the instrument. If you work with COTS transceivers, it is generally not required to own it, since those are originally qualified by the manufacturer. Good SI design methodologies and simulations are enough in that a case. The RTA features are instead very interesting in a variety of designs and test-benches involving RF signals. However, the purchase of a spectrum analyser could be discussed if time domain analysis are not required, this would probably allows spending a bit less.
Wow what an amazing instrument! Really loved the breakdown of the different boards!
I think that was the one I saw at the NIST antenna lab, along with a down-converter from Rohde Schwarz to bring the antenna's transmitted frequency down to something the Keysight scope could see. I think they said they could do work up to 300 GHz.
No word of boot time, fan noise or tap test on the inputs. Also I bet serial decode will be $800 optional extra! So I’m holding my purchase for now. :P
Just for some perspective.. you can capture light traveling less than 1mm with this puppy. That is a fast time domain. Useful at CERN. Probably overkill for weapons research :)
Saw this at a conference today, the money they spend to get their products reviewed is certainly well spent.
If you're still around in 25 years or so, Siglent will put this on your bench for $1M, which by then will the equivalent of $1k in today's money.
But for now: wow.
plemli it will likely be cheaper, we are making major leaps and bounds with nano technology like graphene and carbon nano tubes.. getting to the point of making transistors.. in 10 years silicon may be outdated
Have you heard of the nano tin particles that are room temperature super conductor? And they are thinking with fluoride added into the matrix they can get it to be super conducting up to much higher temps
@@greenthizzle4 from a physics standpoint, i dont think silicon will just be replaced
drkastenbrot it's possible given recent advancements in nano technology
drkastenbrot 25 years is a long time when we are making progress at the rate we currently are..
Woah, that's a crazy device indeed. The complexity of that project, damn.
I wish I had a second life after the one I'm living to have a chance to work on these designs. Wow. Mind blogging.
My impression of how the boards for such a high-end instrument would be made is this: Each board coming from the manufacturer would be X-rayed first for an initial visual inspection, and then would be passed through a flying-probe machine to check continuity of the traces. After the flying probe machine, the boards would then be placed in a thermal cycling oven and cycled for several hours and then tested again on the flying probe machine to ensure that the boards were manufactured properly and will have very high reliability over the service life of the instrument. Only then will the boards be passed through the silk screen presses and to the pick-and-place machines for the placement of small and dense components. The pick-and-place machine would be a high-reliability type like that made by Essemtec which has sub-micron alignment accuracy over the entire area of the board. To achieve this the machine would not only align to the fiducials, but also scan key areas of the board to account for any error in pad alignment before placing the components. The boards would then go to the first manual inspection, where the alignment of all of the components placed by the machine is verified and any missing components are placed manually. Boards which pass will then be manually populated with larger surface-mount components such as relays, coils, connectors and hybrid components, and then passed on to the reflow oven, which uses an inert atmosphere to mitigate any oxidation which could occur during the cycle. After reflow the board will then be sent for another manual inspection. At this point any through-hole components will be installed manually and soldered by a robot, after which the board will receive a further inspection. The boards which pass will then be throroughly washed and sent to a special facility with tight climate controls for integration of chip-on-board components. The process of integrating the chip-on-board components will be done entirely by hand, with each chip being secured with thermally conductive epoxy and then connected by wire-bonding. After the components are installed the boards will then pass through two independent inspectors, each of which will conduct the same integration inspection, and both must pass the job before the completed assembly will be encapsulated using laser welding. Each board after the requisite steps are completed will pass through several stations, each of which will conduct individual board-level checks to ensure that each board functions exactly as required. The individual boards will then be installed in a test instrument, where an initial calibration will be performed and operational checks carried out by hand. Sample boards from each run will then be burn-in tested for a full week and inspected once more before they are sent for integration into the product for sale.
That's absolutely crazy. I really have no idea how they measured those parts during the development.
Today I am trying to routing my first bga, with lpddr4 memory that works at 1.2GHz and TH-cam recommended me that video like "ha ha ha" watch what really cool engineers do.
Did the first revision work? :D
Just an amazing video and piece of technology. Thank you for sharing this!
This is INSANE !!! Cant wait for the Chinese clone for $19.95 with free shipping !
At this level of price and function, I'd be very afraid to hook anything up to it! but it would look soooo good on my bench!
But how do hey test the test instrument?? It's already so massively difficult to make this, how can they measure and qualify it's gain, linearity etc?
they have 110 ghz signal generators, too.
And how do they test the 110 ghz signal generator when they need the signal generator to test the oscilloscope but also need the oscillopscope to test the signal generator?
@@Jonas_Meyer thats what the 110Ghz Spectrum Analyzers are for
Like other comments say, use a NIST traceable source and see what shows up on screen! Also, lots and lots of simulations during the design process.
my question is similar, do you already have a better specs scope or instrument that was used to design this and is just for internal use? or develop this using the same tools that you already have for sell?
I just started to know what oscilloscopes actually are since 3 days ago. But I can still imagine how crazy those numbers are.
Here i am.. just two weeks ago😅😅
"You're only ever as good as the tools you use." *Laughs in poverty*
07:06 Ok now, who's going to area 51, sector 3z, cell alpha to tell the poor alien, that someone stole their technology?
😂😂😂😂😂
I wouldn't even know how to turn this thing on let alone use it for anything. But it doesn't make me want one any less! 🤑
I hope they let you keep it 🤓
I wonder how much skill and esteem from manufacturers is required to receive such instrument for a review and a teardown. Of course a lot!
He couldn't flip it anyway. These things need all kinds of support to be of commercial use. And the publicity he gives it is well worth one broken piece.
I'm new to electronics and am using analog oscilloscopes to do "Oscilloscope Music''. I would love to know how the high sample rate on this digital scope performs on this task. This is a beautiful piece of engineering. Thank you for showing us.
Question on the calibration value measured
The scope shows a value of 3.3ps for the fall time. This means the scope is sampling the edge much better than the 256GS/s spec.
If it sampled the edge correctly using the 256GS/s speed then the 2 sample values are 3.9ps apart.
So how is it able to measure 3.3ps for the fall time for the edge?
Am I missing something?
isnt the spec for fall time the time it takes for it to fall from 90% to 10%? so if the signal goes from 95% to like 5% you could probably extrapolate that the fall time is quicker than the sample rate? also you can use multiple aquisitions to cheat your raw sample rate if your sampling timing is precise enough
Normally I would not think of a scope as a phase noise measurement system, but with the jitter analysis package I am thinking this could measure close in SSB phase noise on mmWave sources. Any thoughts on this?
Wow! What a absolute beast.
Guys, if you want to see things even clearer then you can use a magnifying glass in front of your phones screen.
It works really well.
Thanks for the awesome video. =)
I sure could have used this when I was calibrating noise sources used in radio astronomy. That cosmic microwave background can be a real bear.
Amazing! You just almost restored my faith in humanity!
Would be interesting to know what is the timing source used in this instrument.
platypus' heartbeat
As I've got, it's just main low freq clock at 8 GHz (low noise PLL clocked by low noise OCXO), and higher clocks are formed just by cascaded doubling.
This results, in theory, the ideal timings between cascades, by the definition of harmonics.
The main comlicated technical issue is alingment of delays between cascades, due to the finite (and, in this case, actually very low, lol) speed of light.
This results to extremely high geometric requirements to machined parts and traces.
28:00 Interesting to see that the base of these dont seem to be ceramic but ordinary PCB material? Dont think I have seen that before, but it would absolutely be cheaper though.. for them that is.
Well, this is simply amazing! Is there a reason for the front end to be so messy? Why not make all four channels symmetrical?
Do they offer a volume discount?
Yes, how many do you want? :D
Lol. You guys know how to market (and make amazing products). It's nice to dream.
Keysight Labs I want 10,000 units, can you do that for me at $10,000 please? If you take a little loss I'm sure you can write it off on taxes
yep :) with such a volume with handles you have a right to ask for the discount.
These are number i never expected to see in my lifetime and i have not even seen the video yet :P
Very impressive piece of hardware. Thanks for the concise and accurate information.
About halfway through the video now.. what would the practical application(s) of an oscilloscope like this be?
@Maytemberr😂😂😂😂
Maytemberr yes if you tried you fried a 110ghz scope is 0.5vpp maximum well you west 1.3M
I need it to make sure my 32768 crystal is accurate so my alarm clock goes off on time.
Me: Hey boss, why are you firing me?
Boss: You were 4.3 picoseconds late for work today!
T3chNiqueBeatz design and testing of ultra high speed networks and other communications systems
What's the power consumption of that? Wouldn't be surprised if it tripped a normal 15A/115V breaker the moment you turn on the 2nd channel. :)
Shahriar, how does KS achieve such phenomenal phase noise?
How is the main 64GHz clock derived?
Looks like I gotta hold out for the handheld version for use in the field.
😂😂😂😂after for years it's not yet available buddy.. maybe in twenty yeard😂😂😂
I am rewatching this video i guess for the 4th time as my company about to buy one :D I was wondering, how did they manage to isolate all 8 lines per ADC in the frontend module. They look like they are laid out very very closely. Any ideas?
Oh! II've found the same in the street at a few yards from where I live. Unfortunately, the screen was broken.
Great video (as always)! Do you think you could make an introductory video on distributed element circuits?
The scope that we deserve but not the one we need right now :)
Holy ASIC's Batman!!
I'm interested how the time domain waveform looks like after the samplers. It is new to me that the sample and hold circuit of the ADC is not on the same chip with the quantizers.
Does it look like as a series of step functions with different amplitude or impulses with different amplitude? Could you help me out with good article about it?
As I've understood, basic idea of operation is simple:
This scope is running all the ADCs at some one low frequency, and, in fact, they are semi-directly digitizing one input.
The magic is, that they are sampling with small equal delay (clock phase shift -- more correctly) between each next ADC. This complicated golden S&H circuitry is doing the job of fast and precised sampling with different phases at this one low frequency.
Then, output sample stream is formed just by simple placing samples of each ADC one after one.
Thus, answer to your question is very simple:
Signal after each sampler looks as time-quantized copy ("staircase") of input signal, with sampling frequency equal to full sample rate divided by number of "phases".
Why isn't it possible for these signals to look like stream of short pulses? -- Because duration of these pulses might be in zeptosecond range (duration of actual slopes), what, in my opinion, is truly impossible, and not neccesary for operation of each next stage of samplers and ADCs.
Do you know what happen to the data after the dsp? It go to the CPU motherboard and processed by the CPU or the data is overlayed on the screen and the gui is runner's by the CPU?
Someone might have asked already, but on 13:08 you mention two "distributed" amplifiers in each arm of the hybrid for the 64 GHz clock. Do you mean "distributed" between the hybrid's arms? I do not see a reason for a broadband amp to amplify a filtered 64 GHz tone.
Any technology sufficiently advanced is indistinguishable from magic!
Oh, an oscilloscope! I was thinking it was a spectrum analyser but had to re-read the title. Holy poo, that's insane! The rise time must be in femtoseconds or attoseconds.
The thing with these is that I wonder how you can be really sure that what you see is the actual wave form. With low end scopes you can always take a higher end to confirm.
It's not the actual waveform. When getting close to the cutoff, everything looks like a sine wave.
That's the point of the calibrator.
There must be an enormous amount of reagent grade silver in that mechanical attenuator.
It is possible that the CMOS ASICs are made by GlobalFoundries as they have worked extensively in the past with Chartered.
We sometimes work with other ASIC manufacturers, but for some small-batch super-high-tech stuff we use our own fab in Santa Rosa, CA.
still shocked and awestruck! mind blowing indeed. Are those signal lines in the acq board SIW or guarded striplines? Although the dispersion of SIW might be unacceptable here... Also, does it go down to DC or is there a minimum frequency input?
at 43:00 what's that frequency multiplier? Is it handmade?
Let me guess - you have to send it back. :)
Excellent review.
Don't forget to grab one of eBay in 30 years for $500 to stick in the Lab. Mind you, I've a feeling that getting spare parts for the "repair video" may be an issue.. :)
Tricky Rat I wouldn't send it back, I would go into hiding
Almost literally DC to daylight woah
I still want to see the scope probe that covers THAT!
0 to 770THz. They’ve got a way to go.
Daniel Segel I trust it will be done in 30 years.
The lowest end of infrared is 300GHz, so yeah almost touching that transition into literal optical radiation. I imagine that they are approaching the physical limit to what is possible with copper on a board.
Shahriar, It's great to see such an in-depth teardown of the tool. How long did they let you "play" with it? Do you have the contacts with Keysight through work such that they lend you such equipment?
what the actual...
What a beast!
Great video too.
😎 OWON salió del planeta..... Impresionante tecnología, y es de minimo 5 años atrás, que se estará cocinando ahora?
Great explanation how it works. Thank you.
This scope would be perfect for that super discriminating audio engineer looking to get out ALL the artifacts from his audio amplifier :-)
😂😂😂😂
38:12 have a look at the circular scratches in the conductor surfaces- is that milled?
I think I will pick up couple of them just to play around with!
TNX FOR ANOTHER GREAT VIDEO !
73 N8AUM
You can bet the gubmint has at least 30 of these collecting mold in a storage container in the tropics somewhere... likely originally used for ballast in a submarine.
Wtf you smoking?
WOOW What an UNBELEVABLE TECH :O