EEVblog  

Hi Dave, I love the Knife as a pointer! :^) I think for future teardowns, you should use a wetted soldering iron with a little solder ball on the tip, as your pointer.... Make sure to "skim" the boards surface with it. Then wait for all the comments to roll in. HA HA!!! But in all seriousness, Love the videos!! Keep up the good work.

46:08 In this topology we have to consider wave propagation. The butterfly structures are transmission lines stups which, depending on the frequency, acts like a short or an open. Mostly these structures are used as shorts for a certain frequency. The flare out a the end of the stups makes the damping characteristics in the transfer function more wider compared to a non flared out stup. The connections between the butterflies are much more thinner than a 50Ohm transmission line should be. The idea is to increase inductance so the butterflies can provide much more damping due to the higher impedance caused by the inductance.

How often do people get to see inside an $18K spectrum analyser? I think it's very worthwile and very interesting.

Yo dawg, we heard you like shielding, so we put shielding in your shielded shielding so you can have a shielded shield in yo shielding!

RF design is wizardry

Those copper tapes at 41:00, I think, during the metal shield is being soldered, the air inside will expand, so the hole is used for the air relief, and once technician done with soldering, he/she simply seal it off with a that copper tape.

I hope my relatives never find out what I consider pornography.

" There so..shield that, ya know, we're talking about 20GHz here, put a lil' bit of spit back on that, she'll be allright, no worries!". LMAO, hilarious

Min 26:50 Very likely the reference generator loop - This takes in an external 10 MHz or uses the 10 MHz VCTCXO (internal reference for basic stability). The phase noise at far offsets (> 30 -40 Hz) is achieved with the 40 MHz VCXO. Min29:11 - Silver plated Beryllium Copper for springiness (The white material on board is silver coating which gives lower loss than Gold coating (due to Nickel layer underneath). Min31:17 - Filled via - with conductive or non-conductive (most likely) plug. Min31:42 - Prevent leakage with Rogers 4003 or 4350 outer layer with Glass epoxy layer sandwiched? Min34:08- Near DC to a few GHz Amplifier, L38 in view is high value L with a ac Impedance limiting Resistor in parallel (alternate would be conical inductors). Power supply on the left (passed from backside?) Min34:30 - Low Pass Filter to limit the Input BW. Min35:11 - Local oscillator Amp for Mixer mixing a Low Pass Filter (0 -3 GHz?) to limit the Input BW. Min38:06 - Single Frequency Amp LEE-39+- 7GHz? Min39:08- Power supplies and bias generators? Min47:02- Coupler for LO null? Min47:33- LPF, Min48:35- First Mixer followed by LPF & BPF (U shapes), LPF s are cascaded for different cut off frequencies. Smaller the size-higher the freq. Min49:23- 2.8 -8 G , 940 MHz(?) IF Mixer. CRO-3750 for 2nd LO?

32:15-33:30: "424" is a Hittite digital attenuator. "H270" parts are Hittite 2 way switches, "1GM1" part is an Agilent 20Ghz active mixer. 33:25: "184D" is an inductor for the "N4" darlington gain block. 33:40: Another "424" Hittite step attenuator 34:15-34:20: More monolithic darlington gain blocks. 35:10: "H564" parts are Hittite LNA amplifiers.

I wonder if that guy from Agilent China fully understood what you were going to do with that expensive piece of kit :) - or he is an engineer himself and proud of a rare chance to show off his work. I am a firmware engineer and I know that the guy who designs the stickers on the product gets all the attention, but a beautiful board layout is rarely appreciated.

These transmission line magics never fail to amaze me.

The holes on the gold cans may be vents. When they are soldered the air expands. What do you think?

Hello, Dear Dave! I can't understand english at all but, I just enjoy watching... It's very sad to know that I am losing a lot of valuable information. 😢 All your knowledges is pure unattainable gold for me. I am a apassionate maker electronics, and you are an inspiration to me, brother! 💖 Greetings from Córdoba, Argentina.

I have no idea what I would use a tool like that Spectrum analyzer for, but as a engineer and fitter/turner, I love the machine work and construction of it!, just amazing tech!

Thanks Dave! I have the N9340A HSA that I use for filter and duplexer tuning and general purpose work in public safety radio installation and maintenance. I paid a little over $2500 for it used on eBay. It’s definitely paid for itself in the 10 years or so that I’ve owned it. Great device. I love the fact that you allowed us an in-depth view INSIDE the N9344 series SA. This allowed me to avoid taking my SA apart and most probably destroying it just to see the inside magic of Agilent. Thanks for a phenomenal channel !

Little hole in cans will be to let hot gases escape to stop then lifting & blowing holes in the solder during reflow

What I want for christmas is a video of the face of an agilent calibration technician while watching this video...

Should have turned it on and done a quick function over view. Its ALWAYS good to know what something working right works like... before pointing out details of how it gets there.

The structure at 46:55 is a directional coupler. A wave traveling down one of the lines will couple across to the other line causing a wave to travel down it in the opposite direction.

36:07 is a slotline hairpin bandpass filter. And the one above it is a Chebychev/elliptic lowpass filter. It's easy to tell low pass from band pass, just view it as a dc line. if it passes dc (there's a direct connection with no lines to ground, it's a low pass). The chebychev/elliptic low pass is like a normal filter, but the parallel element is actually a resonator instead of a single reactive element. Really nice tear down though. Subscribed!

Wow, those guys at Agilent sure give the impression of being masters of their craft.

I love the smell of expensive hardware in the morning.

Not just a screwdriver but a jack knife!! 18,000 bit of kit there. Thanks Dave, for taking the time to show us what's inside one of these units.

The base manufacture cost of this is probably around the $5K+ mark.

UPDATE: Yep, it works!

I thought that youtube had rules against uploading porn...

@40:50 I assume that the pinholes in the shielding cases are to equalize the air pressure during and after soldering :-) It is a real beauty Dave!

Just like Norman Schwarzkopf stopped short of Bagdad, YOU stopped short of going beyond the copper tape and into THE GOLDEN CHAMBER!!! YOU'RE FIRED!!!

I believe the LO at 40:04 indicates the Local Oscillator signal. MCL39 is a DC to 8 GHz monolithic amplifier.

If you search "distributed element filter", you will find a wikipedia article which shows pictures and descriptions of the PCB filters. Several of the pics are from the N9344C.

the knife pucker factor across those RF traces is just...keeping me on edge!

The parts labeled four digits on top and four on the bottom are Agilent chips, using the same kind of part numbering scheme that HP used back in the day. The ones with ceramic tops are most certainly InP devices which are manufactured at a plant in Germany which, if I understand correctly, is a joint-venture with X-Fab.

I wish I could understand most of what you said, but I still enjoy watching you take such a pricey piece of equipment apart!

Oh how I love you! I recently discovered your channel here and I spent hours upon hours watching your videos. While I know nothing of the inner workings of electronics I do love learning about many things. Thank you very much for your work!

47:07: Yes, microstrip directional coupler. Exactly what I was thinking. Maybe this is being used to isolate the incoming signal from a local oscillator just before a mixer diode. I DO love that "kite tail" bandpass filter a little later (50:00), though.

The IC shown in 32:00 can be Hittite part HMC424LP3 digital attenuator, H270 can be HMC270ms8 6GHz SPDT switch.

Yep, it took forever to upload and process a 2.5GB HD file. I started it Tuesday night my time. Had to leave it running all night and still wasn't finished processing until 10am this morning.

Wow! That's a lot of pretty impressive stuff there. No wonder it's quite expensive. I just can't imagine the development costs for this analyzer alone. I recently found this blog and find all this teardown videos very interesting (including all other EEV videos). Even as an mech. engineer with limited knowledge in this area i just love them. Keep up the very good work!! / Mats from Sweden.

As said by a couple of others, the Hnnn parts are Hittite parts. I saw: HMC349, HMC700, HMC773, HMC564, HMC424, which are actually pretty easy to identify if you know what you are looking at (and now, you do!). The "LO" you pointed out has nothing to do with local oscillation, those are input LNAs to set the noise figure at the input connector. Cheers!

I bet those "access holes" under the copper tape patches are for pressure equalization so that hot air can expand during soldering and contract when it cools down afterwards.

I love seeing inside this crazy kind of RF stuff

Microwave and RF designing have always reminded me sth unearthly and sacred as we get close to light speed. You did a great job of dismantling this weird device.

At 40 minutes there is a section where LO is Local Oscillator, RF is exactly what it says, and the chip i think is an RF Mixer for down conversion. Z1 is the impedance matching on the output.

Thanks for your videos! I'm a young electronics designer and I find you vlog really amusing and interesting. Keep it awesome! Greetings from Spain!

High timebase stability. The GPS signal is used to discipline the reference oscillator.

Agilent must be crazy to put their beautiful Spectrum Analyser under the knife. Truly an RF work of art.

Agilent gets free advertisement from this, why tek wouldnt want that baffles me, any one buying this product would want to know how it was put together, so be it if you break 1 if it gets them 10 more sales, since these don't get the privilege of large scale mass production price cutting

High-end gear, high-end walk through. Thanks, Dave. Looking forward to a Fundamentals Friday, too.

Cool teardown! I can hardly wait to see what you do for the next fundamentals Friday. Would you consider doing some of the theory behind the pcb based band pass filters? Thanks a bunch!

If it looks like a phillips, and smells like a phillips, chances are it's a JIS or a posidriv, or occasionally one of the weird ones like frearson or quadrex.

I want the input board looks like under all that shielding! I bet those holes under the copper tape are to equalize pressure during solder and assembly.

Yes, almost looks conformally coated, that's what I thought at first glance too. Just really shiny gloss solder mask.

Uh, I'm guessing the last "cube-conduit" simply does a phaseshift which is welcome to passivate connections. Thats exactly the same with this "snail-trail" and the "hand-held-fan-lookalike" that are always dimensioned for one quarter of the wavelength of the voltage frequency. Cool Vids, I really enjoyed it. Thumbs up!

Very interesting seeing the changes to these RF devices over the years as bandwidth goes higher and higher.

12:40 I love that inductor. So cozy in there

One of the best videos on RU-vid!

The pcb filters are cheaper to mass produce, but almost always under perform discrete component based filters. The killer is the losses in the pcb material. The alternating pattern filters are probably interdigital band pass filters. The ones with the pizza slice caps may look like low pass filters. The one with the smooth curved inductors is probably f

Something to consider with a spectrum analyser like this is it probablly has serveral different routes a signal can take through the system depending on the operating frequency. You can't reasonablly make one signal chain work from 1MHz to 20GHz.

Hey Dave, just wanted you to know that the Stiff Co-axial cables are 99% copper and another wire because just got myself one and cut it.

31:00 next to the added trim jumper is an area that looks to have been laser cut.. During Bench build, Tuning assembly? The shinny dots may be through connections as Z axis shielding around 46mins some of the this shielding looks tobe drilled threw some not, random.. 0iC A Ground Plane, MayBe...

The white package at about 20:00 is almost certainly a mixer from Mini-Circuits.

I don't think you mentioned this, but also note the symmetry of the filter at 34:30. That's not just for simplicity or looks, but to mitigate the effect of unequal stage loading between orders of the filter. Also I could be wrong, but the "zigzags" or rather meander in for instance the bowtie stub filter at 46:10 are rather to ensure the correct length of trace (or rather in this case, waveguide) (eg 1/4 wavelength or whatever is needed) between elements rather than act as elements themselves. It's interesting however that they don't have chamfered edges in the RF sections which is where I'd _really_ expect to see it. As for that little loop at 50:17, my best guess would be maybe some sort of a pi network.

After not seeing a eevblog video for a few days, this one popped up and it is seriously well above the average! (which is awesome). Cheers for the awesome experience of looking at some RF voodoo circuit design, Dave. I've been taking electronic stuff apart since I was a toddler, I can only hope to one day make a living doing it and with items this exotic. Mega props Dave, your videos inspire me to learn more about this awesome subject. Peace

Another fantastic teardown. Would love to see it working again after re-assembly lol. You really should get a laser pen for pointing with. I've watched a lot of your videos and noticed you always leave marks on the pcbs etc by using a screwdriver as your pointing device :P Keep up the great work, I look forward to your videos!!

Realy nice that they let u open ther toys, awsome marketing. I honestly would not even consider a unit like that until i saw how much work they actualy put into making it worth the money.

Dave, are we going to be treated to one of your famous high speed reassembly videos with the whistly audio and final power on at regular speed? Hope so, those are fun.

Holy crap, what a lucky man Dave!

This kind of RF electronics design really is magic! Great vid Dave :)

At 40:53 those are not access holes. Those are to relieve/vent any pressure that might build up during the soldering process of that large shield. Then they plug it post production for RF reasons.

17:42. Those are ground connections and a Samtec connector used for LVDS. It is an expensive connector.

Just re-watched this today, and I am still at total AW of that level of voodoo going on in there.

Dave, I noticed that on the Spectrum Analyzer Wikipedia page a picture from this teardown is currently at the top of the page!

The analyzer is just a badass, never saw such complicated microstrip filters

Using the sharpened knife is even better, at one point you could see it making a scratch right across one of the RF filter magic patterns.

I wonder if the converters are JESD204B on it to the FPGA, The low I/O count to the other blocks would tend to suggest that. The main board shielded blocks are probably precision clock reference generators, shielded for the femto second order performance.

E.T can phone home with that thing .... LOL!

at 46:58 seems to be a directional coupler, used most to detect vswr in this case maybe to create a derivation for measuring the sinal level or something..

The analogue part looks like some alien tech to me. So amazing. Hope I can work on one of these someday in the future

6:00 Ladies and gentlemen, I'm honored to present you, the main processor for the T-600!

Incredible video. Always learning with you. RF master stuff here :)

The structure of the paired lines at 47:00 is likely a directional coupler.

at 47:02 is a directional coupler i think (could be wrong but don't think so). directional coupler is meant to "sample" a small fraction of the RF wave propagating from input to output. Common coupling port can be 20 dB or more.

I believe that "From control board's left side" is the proper use of the possessive.

I'm all for taking nearly everything apart, however, I ALWAYS turn in on to play with it first!

My guess. After the solder solidified the contracting air in the container might be enough to flex the PCB and possibly damage some ceramic parts inside or out.

Awesome, good to hear, keep it up!

Not likely. Waveguides are really only practical at microwave freqs due to the physical sizes needed. a wave guide for 1MHz would be enormous.

The info in wikipedia about non lumped pcb components shows some of the Agilent parts shown in this video. Go to wiki and type in Distributed_element_filter.

Hilarious at 6:19 aaaaahhhhh uuffffffffffffffff ooohhhhhhh I bet your wife is not jalous anymore when you play Doctor with those devices ^^ Nice Teardown by the way, and nice from Agilent to send you this HSA!! I wish i could have one one day....

fantastic cheers, thing of beauty

I hope they can buff out all the scratches. Granted not on critical paths and probably deemed as a demo unit for the rest of it's life but would it be so hard to use a plastic trimmer as a pointer?

At 30:44 exactly. Is that soldering of the connector broken, or it'a a play of light. That would whoop down the signal to the ground (that kind of crack takes 10 GHz to dust). 31:01 look at that stub tuning at the right center :D. 31:11 it is actually a tuning method for that high frequencies. If made on Teflon or on ceramic, it would be more precise with epsilon, and hence less tuning. Would not say a word anymore. I just had a bear after a long day in open pit coal mine. Best Regards, thumbsup

The acces holes are for air escape while soldering them in place.

This wasn't a review though, it was just a teardown. When it comes to these teardowns it's to show us stuff that you normally would never get a hand on and everything inside. So it's for precisely the reason of it being expensive.

Yes, I added those.

Amazing......u reminded me my previous job as test engineer....lethal stuff..

Thanks for the teardown. If I tore down our N9344 the boss would fire me! It is a great instrument.

I might try a laser pointer, that could be cool, but probably a bit wobbly under the x10 macro lens.

46:58 isn't that just a coupler? Seems the upper branch goes left to the board-to-board RF connector, while the lower continues right.