EEVblog Electronics Community Forum
EEVblog => Other Blogs => Topic started by: Hugoneus on August 13, 2013, 03:03:13 pm
-
Experiments and Demo of an Agilent DSA-X 96204Q 160GS/s 62GHz Oscilloscope
In this episode Shahriar demos the world's fastest oscilloscope! The Agilent DSA-X 96204Q offers 160GS/s of conversionrate with a bandwidth of 62GHz on two dedicated 'RealEdge' channels. It can also provide 80GS/s conversion rate and 33GHz of bandwidth on four simultaneous channels. The unit demoed on this video is equipped with all available options and is valued at over 0.5 million US dollars. The block diagram of various sub-systems of the oscilloscope are presented and the principle operation of the instrument is explained. A 56Gb/s PRBS-15 signal is applied to the scope from a Centellax 2G2T5A (now Agilent N4975A) and the resulting data pattern is examined in real-time. Various scope functions are also presented as well as the capability to observer bit failures at baud-rate using this instrument.
As a second experiment, two 3.125Gb/s PRBS-7 data streams are simultaneously up-convered to 20GHz and 40GHz respectively by using a pair of MITEQ mm-wave DSB tripple balanced mixers and a pair of Avantek 20-40GHz YIG oscillators. The resulting two signals are combined by using a power-combiner and fed to the oscilloscope. The capability of the instrument to act as an ultra-broadband software-defined radio is demonstrated by recovering the the two PRBS sequences simultaneously through DSP post processing. The block diagram of this setup can be downloaded from The Signal Path website.
You can see the video here: [1 Hour & 10 Minutes]
http://thesignalpath.com/blogs/2013/08/13/experiments-and-demo-of-an-agilent-dsa-x-96204q-160gss-62ghz-oscilloscope/ (http://thesignalpath.com/blogs/2013/08/13/experiments-and-demo-of-an-agilent-dsa-x-96204q-160gss-62ghz-oscilloscope/)
More videos at The Signal Path:
http://thesignalpath.com/blogs/ (http://thesignalpath.com/blogs/)
-
Dammit Shahriar; how the hell do you get your hands on all this uber expensive equipment?
-
Dammit Shahriar; how the hell do you get your hands on all this uber expensive equipment?
Friends in high places! ;)
-
awesome! :-+
-
Anybody looking for some sticker shock? Look up the price of the signal generator he was probing with that beast of scope... >:D
I loved the section on the input/ADC. Thanks :-+
-
Anybody looking for some sticker shock? Look up the price of the signal generator he was probing with that beast of scope... >:D
I loved the section on the input/ADC. Thanks :-+
Well, you've got to connect something expensive to a half-a-million-dollar scope!
-
Yeah, I can't think of too many affordable ways to generate a signal worth sticking a 62 GHz scope on!
-
Yeah, I can't think of too many affordable ways to generate a signal worth sticking a 62 GHz scope on!
What did you think of the second experiment with dual 3.125Gb/s PRBS7 up-conversion to 20GHz and 40GHz bands?
-
Ah, that's the twin of the one i used to test the Jim Williams pulse generator. They didn't have that one ready yet. They were working on it, it was still in area51 ...
To see real pictures of that sampler hybrid :
https://www.eevblog.com/forum/testgear/agilent-dsa-x93204a-33ghz-80gss-teardown/msg195482/#msg195482 (https://www.eevblog.com/forum/testgear/agilent-dsa-x93204a-33ghz-80gss-teardown/msg195482/#msg195482)
-
My scopes just fart and shut off if I suggest using them as radios at any data rate! ::)
I'm amazed that you could recover so much data moving in and out of Fourier transforms like that. Of course, I have pretty much no experience in proper digital signal processing. Is there any chance we could get a look at your MATLAB code? I've used it plenty, but not for this kind of thing. I'd love to see what exactly you're doing.
-
My scopes just fart and shut off if I suggest using them as radios at any data rate! ::)
I'm amazed that you could recover so much data moving in and out of Fourier transforms like that. Of course, I have pretty much no experience in proper digital signal processing. Is there any chance we could get a look at your MATLAB code? I've used it plenty, but not for this kind of thing. I'd love to see what exactly you're doing.
Going in and out of FFT in Matlab is lossless on it own. Of course, the filtering and shifting things around is not. :)
I can try posting the Matlab code that I wrote. I would have to add some comments to it so people can follow it.
-
Ah, that's the twin of the one i used to test the Jim Williams pulse generator. They didn't have that one ready yet. They were working on it, it was still in area51 ...
To see real pictures of that sampler hybrid :
https://www.eevblog.com/forum/testgear/agilent-dsa-x93204a-33ghz-80gss-teardown/msg195482/#msg195482 (https://www.eevblog.com/forum/testgear/agilent-dsa-x93204a-33ghz-80gss-teardown/msg195482/#msg195482)
Great, thanks for posting that. I should have asked for the ADC module also. I remember holding it in my hand while it was in development. At the time, 80GS/s was mind blowing.
-
Great video!
But for $500k you only get crappy microwave pop buttons?
-
http://www.microwavejournal.com/articles/9838-inp-hbt-chipset-enabling-high-bandwidth-real-time-oscilloscope-architecture (http://www.microwavejournal.com/articles/9838-inp-hbt-chipset-enabling-high-bandwidth-real-time-oscilloscope-architecture) There's some fairly in-depth description of how the sampler works.
http://poulton.net/ (http://poulton.net/) This Guy seems to be one of the engineers behind these super fast ADCs.
BTW: Lecroy claims to have 100 GSas/s realtime: http://teledynelecroy.com/100ghz/ (http://teledynelecroy.com/100ghz/)
-
http://www.microwavejournal.com/articles/9838-inp-hbt-chipset-enabling-high-bandwidth-real-time-oscilloscope-architecture (http://www.microwavejournal.com/articles/9838-inp-hbt-chipset-enabling-high-bandwidth-real-time-oscilloscope-architecture) There's some fairly in-depth description of how the sampler works.
http://poulton.net/ (http://poulton.net/) This Guy seems to be one of the engineers behind these super fast ADCs.
BTW: Lecroy claims to have 100 GSas/s realtime: http://teledynelecroy.com/100ghz/ (http://teledynelecroy.com/100ghz/)
Thanks for posting the articles.
I tried to invite Poulton to write a paper on ultra-highspeed ADCs for CSICS this year, unfortunately he was not able to participate.
As for LeCroy, I will believe it when I see it. I used their current >60GHz scope and did not find its performance to match that of Agilent. They use digital bandwidth interleaved technology and I could not get very good results for viewing high-bandwidth signals which spans multiple spectral bands. But competition is always good! :)
-
Video Demo & Experiments with the World's Fastest Oscilloscope!
Wasn't it Teledyne LeCroy having a much faster scope with 100GHz bandwidth and 240 GS/s?
http://www.elektronikpraxis.vogel.de/messen-und-testen/articles/413028/ (http://www.elektronikpraxis.vogel.de/messen-und-testen/articles/413028/)
-
Video Demo & Experiments with the World's Fastest Oscilloscope!
Wasn't it Teledyne LeCroy having a much faster scope with 100GHz bandwidth and 240 GS/s?
http://www.elektronikpraxis.vogel.de/messen-und-testen/articles/413028/ (http://www.elektronikpraxis.vogel.de/messen-und-testen/articles/413028/)
Not production and not true real-time sampling. It is block down-conversion and DSP combining, same as their 65G scope.
-
I really enjoyed the video, that is some serious stuff right there.
I started to look at the video and I just want to keep looking until the end. ( Which I did :) )
Good work !
David.
-
BTW: Lecroy claims to have 100 GSas/s realtime: http://teledynelecroy.com/100ghz/ (http://teledynelecroy.com/100ghz/)
s/Sa/Hz/ ...
-
But for $500k you only get crappy microwave pop buttons?
I don't think that's the sort of button they're using here. They are just recessed.
-
But for $500k you only get crappy microwave pop buttons?
I don't think that's the sort of button they're using here. They are just recessed.
Perhaps, but you can hear them popping, over the fans even! My G-mas microwave, 80s calculators, Speak and Spell etc.
-
Update: It is possible that the RealEdge modules do not perform THA functions but rather block down-conversion. In this case, each of the 80GS/s acquisition boards digitizes 32GHz worth of bandwidth and recombining is done in DSP. This is similar to the method LeCroy employs in their scopes.
I have added a note to the video as well.
-
How is the teardown video coming?
-
How is the teardown video coming?
Ha! Sadly, not possible... It was very tempting, especially after the second beer.
-
How is the teardown video coming?
Ha! Sadly, not possible... It was very tempting, especially after the second beer.
I'll PM you my address, I'll get it apart before the Agilent hitmen get here! :scared:
-
Update: It is possible that the RealEdge modules do not perform THA functions but rather block down-conversion. In this case, each of the 80GS/s acquisition boards digitizes 32GHz worth of bandwidth and recombining is done in DSP. This is similar to the method LeCroy employs in their scopes.
I have added a note to the video as well.
Sounds reasonable for some reasons:
1. The 32GHz clock fed into the real edge module
2. http://www.tek.com/document/whitepaper/techniques-extending-real-time-oscilloscope-bandwidth (http://www.tek.com/document/whitepaper/techniques-extending-real-time-oscilloscope-bandwidth) They claim that there's a dip in the frequency response at 32GHz ... coincidence!
3. Service manual: "This Service Guide does not describe the RealEdge Technology assembly in any detail, as the technology is Agilent proprietary" Nothing to see here, go away.
It would have been quite challenging for marketing to explain why they're now using frequency interleaving, because they mocked it in http://cp.literature.agilent.com/litweb/pdf/5990-5271EN.pdf (http://cp.literature.agilent.com/litweb/pdf/5990-5271EN.pdf)
-
Update: It is possible that the RealEdge modules do not perform THA functions but rather block down-conversion. In this case, each of the 80GS/s acquisition boards digitizes 32GHz worth of bandwidth and recombining is done in DSP. This is similar to the method LeCroy employs in their scopes.
I have added a note to the video as well.
Sounds reasonable for some reasons:
1. The 32GHz clock fed into the real edge module
2. http://www.tek.com/document/whitepaper/techniques-extending-real-time-oscilloscope-bandwidth (http://www.tek.com/document/whitepaper/techniques-extending-real-time-oscilloscope-bandwidth) They claim that there's a dip in the frequency response at 32GHz ... coincidence!
3. Service manual: "This Service Guide does not describe the RealEdge Technology assembly in any detail, as the technology is Agilent proprietary" Nothing to see here, go away.
It would have been quite challenging for marketing to explain why they're now using frequency interleaving, because they mocked it in http://cp.literature.agilent.com/litweb/pdf/5990-5271EN.pdf (http://cp.literature.agilent.com/litweb/pdf/5990-5271EN.pdf)
Exactly.
The fact that Agilent has always claimed that block-down conversion is a bad idea, I took it for granted that they were not doing it. I had a hard time understanding how they were doing THA for 80GS/s ADCs with a 32GHz clock signal, which of course they are not!
What do you do for a living Lukas?
-
8)
-
Hurry up, if you buy one of these until September 30th, Agilent will throw in a free DSOX3034A! How can anyone pass this up? ;D
http://www.home.agilent.com/agilent/editorial.jspx?cc=US&lc=eng&ckey=2316099&nid=-33202.0&id=2316099 (http://www.home.agilent.com/agilent/editorial.jspx?cc=US&lc=eng&ckey=2316099&nid=-33202.0&id=2316099)
-
Ooh! I'll take six!
-
If you overload the input on that scope, somebody is going to kill you :)
I wonder how many of those they sell. I imagine most are rented
-
If you overload the input on that scope, somebody is going to kill you :)
I wonder how many of those they sell. I imagine most are rented
They have sold quiet a few already.
-
If you overload the input on that scope, somebody is going to kill you :)
Any idea how extensive the damage would be? I figure it's not exactly packed to the gills with input protection, that would totally destroy the bandwidth.... but how far would the damage propagate if it took 20V?
-
If you overload the input on that scope, somebody is going to kill you :)
Any idea how extensive the damage would be? I figure it's not exactly packed to the gills with input protection, that would totally destroy the bandwidth.... but how far would the damage propagate if it took 20V?
It all depends on what vertical setting it is set to. It has a front-end mechanical attenuator, so you could damage the resistors in the 67GHz attenuator if it is set to high attenuation. But if the scope is set to 2mV/Div, then you can directly damage the front-end. If an overload condition is observed the scope quickly jumps to high-attenuation however.
-
Update: It is possible that the RealEdge modules do not perform THA functions but rather block down-conversion. In this case, each of the 80GS/s acquisition boards digitizes 32GHz worth of bandwidth and recombining is done in DSP. This is similar to the method LeCroy employs in their scopes.
I have added a note to the video as well.
Sounds reasonable for some reasons:
1. The 32GHz clock fed into the real edge module
2. http://www.tek.com/document/whitepaper/techniques-extending-real-time-oscilloscope-bandwidth (http://www.tek.com/document/whitepaper/techniques-extending-real-time-oscilloscope-bandwidth) They claim that there's a dip in the frequency response at 32GHz ... coincidence!
3. Service manual: "This Service Guide does not describe the RealEdge Technology assembly in any detail, as the technology is Agilent proprietary" Nothing to see here, go away.
It would have been quite challenging for marketing to explain why they're now using frequency interleaving, because they mocked it in http://cp.literature.agilent.com/litweb/pdf/5990-5271EN.pdf (http://cp.literature.agilent.com/litweb/pdf/5990-5271EN.pdf)
Note also in the video that the FFT of the noise floor, before applying the frequency translated PRBS signals, shows peaking around 32GHz, indicating that there may be some DSP boosting going on there to try to flatten out a dip at the point where the bands are split.
-
Note also in the video that the FFT of the noise floor, before applying the frequency translated PRBS signals, shows peaking around 32GHz, indicating that there may be some DSP boosting going on there to try to flatten out a dip at the point where the bands are split.
Yes, I am convinced that you are right about this. Thank you.
-
Hurry up, if you buy one of these until September 30th, Agilent will throw in a free DSOX3034A! How can anyone pass this up? ;D
http://www.home.agilent.com/agilent/editorial.jspx?cc=US&lc=eng&ckey=2316099&nid=-33202.0&id=2316099 (http://www.home.agilent.com/agilent/editorial.jspx?cc=US&lc=eng&ckey=2316099&nid=-33202.0&id=2316099)
Its funny that you say that, because we bought a bunch of stuff and have a few of these little guys sitting around now. They offer this promotion on a few items.
-
Yeah, it seems like you can get a free DSOX3034A by spending as little as US$18,581... equivalent to more than 40% off. It's actually a very good deal assuming you need a Pulse Function Arbitrary Noise Generator. ;)
-
Is pretty remarkable,
Just imagine in 40 years with graphene transistors or other technology we might all have a cellphones (or eqiv) that can sample/produce up to 100Ghz for some sort of ultra fast software defined radio RX&TX communications.
-
As now have the sure to said the real edge do step down conversion or is just a track and hold amplifier?
Some one have some more info on that?
Just to know theorically how much an adc of this machine cost (50k, 60k)?
-
As now have the sure to said the real edge do step down conversion or is just a track and hold amplifier?
Some one have some more info on that?
Just to know theorically how much an adc of this machine cost (50k, 60k)?
All evidence points to the use of a block down-conversion scheme. Agilent's 80GS/s ADC is a large array of S/H+ADCs all running at a lower speed. If I recall correctly, there are 160 (or more) that are used to make up the 80GS/s base sample rate.
-
Shahriar this oscilloscope i no more the more fats production oscilloscope, apparently agilent have released a new series of oscilloscope.
The DSAZ634A apparently tha have better performance thant the other scope.
At first the biggest difference we can see from the agilent picture is that now there isn't no more the seperate connector for high speed signal and not.
Just only the channel connector.
The cost without option is just "only" 453k of dollars!
I wish you know something more on this scope and what is new from the predecessor.
Best regards, Alberto.