This channel does something very interesting. Normally, the general public (as much as you can call this audience general) would never see the insides of such expensive and exotic devices. If it wasn't for this channel (and other like it) RF would always be magic, forever a black box. What videos like this can do is share the beauty in the engineering that happens behind the curtains. I know that my interest in RF has only grown with each video and I know that without them I would never appreciate and admire the work that goes in to making anything high frequency. I don't really know where I was going with this, but I'd like to thank you Shahriar.
Had videos like this, or even youtube itself existed when i was in school, i may have become some sort of RF specialist wizard! Alas, it took an extra 25 years, and I'm much less of a wizard. Still, I'm finding uses for this knowledge in my work and hobbies!
It's amazing how zooming in this deep shows exactly how that high/low diode diagram reflects back on the substrate. Usually there is some 'simple' diagram and then when you look at an assembly or a component there is all this extra stuff that can make it hard to identify what's what. This is almost a one-to-one implementation on that sensor! Seeing the construction like this also makes you appreciate all the manual high-precision work that has gone in to this, almost making you wonder how ~$4000 covers all of that...
*The intro animation was cut off accidentally on this video. :(* *Although no damage can be seen on the diodes, this unit definitely has a bad detector.*
Beautiful. Most comments asking for more in-depth analysis and tests to figure out the causa mortis (I must include myself into). But then, that would not classify into a TNP episode. On a side note, I cannot imagine how much EE Pooch has learned through the years, just breathing the air of Shahriar's lab!
Thanks! Would be interesting to see also thermocouple type power sensor. Also I couldn't spot any damage on a die to suggest overload? Maybe a only that spot on main substrate where resistors before the die?
I have a fully working U2001A. As you suspected Shahriar I'm sure it is the same or nearly the same hardware used for all power sensors in the family. Mine seems to work fine to 26.5 GHz, and maybe more. The levels are uncalibrated past 6 GHz but the input impedance and response looks very good.
My brain: "He has already done one of these before." Also my brain: "WOO HOO! I get to learn more AND get reinforcement learning!" This was super interesting as always. I'd love to learn more about the second microscope too! Thanks a million, Shahriar! I wish I could find words to express how much your videos have broadened my enjoyment of the electronics hobby and multiplied my skills & abilities.
Funny, Maxim, LT and AD parts. Now they're all Analog Devices parts. Wonder whether there are any Hitite parts on there too (only 9 minutes in) should be interesting.
What would be the benefit of using ribbon bonds over standard bond wires on the chip? The insertion loss of a ribbon bond compared to regular bond (even hamburger bond) is marginal and is typically only seen at E- or W-band to squeeze the extra bit of bandwidth in. This is a 6 GHz sensor though. Couldn't see any epoxy for die attach so are the ribbons being used to secure the chip down? Power handling capabilities?
Substrate most likely is sapphire, even back in HP times, they used that stuff everywhere around RF signals, i.e. pi attenuator stages in their large multistage attenuators in RF signal generators/spectrum analyzers etc. Sure that none of these maxim switches were blown? These things (Maxim and DG ones) fail quite often, most Keithley 20xx DMMs failures are due to leaky analog switches, since the diodes under the microscope look quite intact....
I wonder if they use Scott Zerodur for this application as well? It is a glass ceramic with high transparency, high dielectric strength and an extraordinarily low thermal coefficient of expansion. The latter property has made it an invaluable material for mirrors used in space-borne optics.
@Joel Dunsmore Could it be polycrystalline sapphire? That should make the dielectric constant isotropic (get rid of birefringence). I don't see why single-crystal material would be needed for low loss.
Nice Video! You mentioned that the diodes operate in a square law region. I was wondering what kind of diodes these are and why they have a square law region?
Another Pooch appearance! Yay! I think according to union rules, once he appears on screen he need to get paid in cat treats for his performance. What an amazing bit of mechanical design, tiny spring-a-ma-things just ejecting themselves during disassembly. Not hard to imagine how much manual labor is involved in the production of small, precise instruments like this.
Thanks for the nice video, very interesting. Confused: Why is the high sensitivity path after the attenuator, I would think that would be the low sensitivity path?
Probably glue to increase reliablity by preventing the chip from breaking off the board due to thermal cycles. That is a common failure mode of BGA soldering.
![[TH] 2024 PMSL SEA W2D5 | Fall | พลาดวันนี้ จะต้องเสียใจ](http://i.ytimg.com/vi/A5-Cuo12p6A/mqdefault.jpg)
This channel does something very interesting. Normally, the general public (as much as you can call this audience general) would never see the insides of such expensive and exotic devices. If it wasn't for this channel (and other like it) RF would always be magic, forever a black box. What videos like this can do is share the beauty in the engineering that happens behind the curtains. I know that my interest in RF has only grown with each video and I know that without them I would never appreciate and admire the work that goes in to making anything high frequency.
I don't really know where I was going with this, but I'd like to thank you Shahriar.
You are welcome my friend and thank you for the kind words.
@@Thesignalpath thank you for your videos
Fascinating to see how complex and beautiful the implementation is. I have new respect for "just a power sensor".
Had videos like this, or even youtube itself existed when i was in school, i may have become some sort of RF specialist wizard! Alas, it took an extra 25 years, and I'm much less of a wizard. Still, I'm finding uses for this knowledge in my work and hobbies!
It's amazing how zooming in this deep shows exactly how that high/low diode diagram reflects back on the substrate. Usually there is some 'simple' diagram and then when you look at an assembly or a component there is all this extra stuff that can make it hard to identify what's what. This is almost a one-to-one implementation on that sensor! Seeing the construction like this also makes you appreciate all the manual high-precision work that has gone in to this, almost making you wonder how ~$4000 covers all of that...
*The intro animation was cut off accidentally on this video. :(*
*Although no damage can be seen on the diodes, this unit definitely has a bad detector.*
Thanks for another power sensor teardown, enjoy them all!
I wasn't sure what the "Square-Law Region" meant. Had to look that up. Learned something new today.
Slick answer, Glen. (yes, a nickle for every time you heard that...)
Beautiful. Most comments asking for more in-depth analysis and tests to figure out the causa mortis (I must include myself into). But then, that would not classify into a TNP episode. On a side note, I cannot imagine how much EE Pooch has learned through the years, just breathing the air of Shahriar's lab!
Thanks! Would be interesting to see also thermocouple type power sensor. Also I couldn't spot any damage on a die to suggest overload? Maybe a only that spot on main substrate where resistors before the die?
beautiful machined parts!
Fascinating, all you need now is an SEM....cheers.
Starting to see why these are so expensive!
Very interesting. Thanks for sharing this. Also I think that has to be one of the longest Pooch cameos.
I just learned that this channel used to be called The Noise Path.
How interesting.
I do reckon TSP sounds much better.
The channel is called The Signal Path. The noise path are the shorter videos on the channel.
Very cool, thanks Shahriar!
A peek into the workings of the rf dark arts is very welcome, especially with your explaination. Thanks.
I have a fully working U2001A. As you suspected Shahriar I'm sure it is the same or nearly the same hardware used for all power sensors in the family. Mine seems to work fine to 26.5 GHz, and maybe more. The levels are uncalibrated past 6 GHz but the input impedance and response looks very good.
My brain: "He has already done one of these before." Also my brain: "WOO HOO! I get to learn more AND get reinforcement learning!" This was super interesting as always. I'd love to learn more about the second microscope too! Thanks a million, Shahriar! I wish I could find words to express how much your videos have broadened my enjoyment of the electronics hobby and multiplied my skills & abilities.
@MichaelKingsfordGray My brain has no idea what you're talking about.
Very cool to see, thank you!
Funny, Maxim, LT and AD parts. Now they're all Analog Devices parts. Wonder whether there are any Hitite parts on there too (only 9 minutes in) should be interesting.
I missed pooch getting a shock on the intro =D Anyways thanks for the great video! =D
What would be the benefit of using ribbon bonds over standard bond wires on the chip? The insertion loss of a ribbon bond compared to regular bond (even hamburger bond) is marginal and is typically only seen at E- or W-band to squeeze the extra bit of bandwidth in. This is a 6 GHz sensor though. Couldn't see any epoxy for die attach so are the ribbons being used to secure the chip down? Power handling capabilities?
@@DrJoelVNA Thank you!
Substrate most likely is sapphire, even back in HP times, they used that stuff everywhere around RF signals, i.e. pi attenuator stages in their large multistage attenuators in RF signal generators/spectrum analyzers etc.
Sure that none of these maxim switches were blown? These things (Maxim and DG ones) fail quite often, most Keithley 20xx DMMs failures are due to leaky analog switches, since the diodes under the microscope look quite intact....
I wonder if they use Scott Zerodur for this application as well? It is a glass ceramic with high transparency, high dielectric strength and an extraordinarily low thermal coefficient of expansion. The latter property has made it an invaluable material for mirrors used in space-borne optics.
@Joel Dunsmore Could it be polycrystalline sapphire? That should make the dielectric constant isotropic (get rid of birefringence). I don't see why single-crystal material would be needed for low loss.
Nice Video! You mentioned that the diodes operate in a square law region. I was wondering what kind of diodes these are and why they have a square law region?
Great question DF. I personally though that "square law" referred to RoBlox in some way. Seriously, does sound interesting. Cheers!
These are shottki diodes. The output voltage = input voltage square X coefficient
Cool video, thanks :)
So cool
The detector seemed fine. I wonder what was the actual malfunction of this device.
I think there is actually a burn mark between the detector chip and the interface horn...
The detector is bad for sure. I have verified this with another sensor.
There seems to be a long crack between the connector and the transition from the coax input side to the IC (upper side of the quartz substrate)?
Beautiful videos.. thanks!
I think the choppers are also to reduce the 1/f noise by increasing the signal frequency. Any comments?
Welcome to The Noise Path !? 😂
Awesome autopsy 😌
These are still available from Keysight: $3866 new.
Very cheap
Which microscopes and cams are you using?
Another Pooch appearance! Yay! I think according to union rules, once he appears on screen he need to get paid in cat treats for his performance. What an amazing bit of mechanical design, tiny spring-a-ma-things just ejecting themselves during disassembly. Not hard to imagine how much manual labor is involved in the production of small, precise instruments like this.
My cat, like pooch, is also always planning his next attack. But I would add this one caveat - when awake.
Thanks for the nice video, very interesting. Confused: Why is the high sensitivity path after the attenuator, I would think that would be the low sensitivity path?
Blackfin is actually a general-purpose CPU and DSP in one package.
Wonderful microscopy work! What make/model microscope do you use?
I would also like to know
Is it possible with devices to measure reflected power of microwave in waveguide ?
Hey! teach me how to change coefficients in power sensors (N8481a...N8487a)
Anyone recognised the low frequency chips at the input that are chopper amplifiers according the block diagram?
👍👍
I need to watch this video atleast 3 times to understand wtf is going on lol. I'm kinda slow
Three (3)? You a Mensa member? Good on you :O) It takes me (3) tries just to get the video to play right :O\ Cheers CTG.
I have a question that has been bothering me for a while. Is your cat's name Pooch or PUCCH?
It is Pooch. :)
Krap! We've all been asking the WRONG questions.. Then Bartosz shows up and "scoops us"? Very good BK. You RoCk! 10 points awarded. Cheers.
10:11 Someone could please tell me the name / manufacturer/family of this connector ?
Missed I where is the problem of this detector? I don't se any damage.
ah I see your comment now :D
What is the purpose of the black material covering the pins of the processor?
Probably glue to increase reliablity by preventing the chip from breaking off the board due to thermal cycles. That is a common failure mode of BGA soldering.
Mostly for protection against shock and thermal cycle. People probably drop these things occasionally.
Your signal to noise ratio is falling!
Sadistic engineer thinking a ZIF flatflex is the best in-production programming interface
👍👍💖🙏💖👍👍
I almost destroyed one at work old hp one
John Cage intro
6Ghz, is that all? I mean consumer CPUs run at 5 Ghz. Why don't we make computers out of 66 Ghz RF electronics?