TNP #16 - Teardown, Repair & Analysis of a few mm-Wave (26GHz - 40GHz) Amplifier Modules

Поділитися
Вставка
  • Опубліковано 17 січ 2025

КОМЕНТАРІ • 100

  • @w2aew
    @w2aew 2 роки тому +40

    What I want to know is - how do you get such nice *glare-free* video of the FieldFox display?

    • @johnnycash4034
      @johnnycash4034 2 роки тому +2

      He uses baby wipes

    • @larslindgren3846
      @larslindgren3846 2 роки тому +2

      Some people think 40 GHz is too low frequency and are more interested in 400 THz technology.

    • @Thesignalpath
      @Thesignalpath  2 роки тому +29

      You walked right by it yesterday when you were visiting the lab! It is a dark cloth which I hang behind the camera to block any reflection from instrument screens. Makes filming difficult, but I think it is worth it. :)

    • @1903A3shooter
      @1903A3shooter 2 роки тому

      Hey Allen, W4GSM

    • @StephenHendersonActual
      @StephenHendersonActual 2 роки тому +4

      @@Thesignalpath @w2aew
      Hoping to see some videos of you two sharing thoughts and exploring the state of the art with your careers, hobbies, and EE on UA-cam (and anything else).

  • @RB9522
    @RB9522 2 роки тому +14

    The VR chip is marked 5V at the top right corner.

  • @xDevscom_EE
    @xDevscom_EE 2 роки тому +29

    Yes to quick videos, for people not constantly busy with RF even quick look on something teaches something new and interesting.

  • @JLCPCB
    @JLCPCB 2 роки тому +1

    Really interesting video! We like it! 😊

  • @pete3897
    @pete3897 2 роки тому +14

    7:50 - oh man I wish you had an overlay Shahrira. I have _no_ idea which part is the gate, nor which vertical line you're referring to! D'oh - I really wanna know! If only there were a pointer...

    • @Thesignalpath
      @Thesignalpath  2 роки тому +12

      I'll fix that for next time. The feed from the microscope is an HDMI line and I need to find a way to overlay on top of it during the video capture.

    • @bansci
      @bansci 2 роки тому +3

      @@Thesignalpath maybe have a screen with the live image on that you video while you talk over it? The microscope images won't be perfect, but you can point out everything interesting

    • @samfedorka5629
      @samfedorka5629 2 роки тому +2

      I'll try to explain. You see how there are 2 horizontal bond wires on the left and two on the right? Each one on the right goes to a little pentagon island in the middle, which is a rectangle, but the left edge of it is a triangle. The very left point of the island has a small horizontal line which goes to a vertical line. That is the gate.
      You can see the upper one, the vertical line on the gate has peeled off with the horizontal line (connected to the pentagon) and is now in the air. It took me a minute to notice it and I've had a HEMT class.

    • @pete3897
      @pete3897 2 роки тому

      @@samfedorka5629 Hey thanks Sam for the explanation. I'm not sure I would have seen that even with an overlay - I needed that additional explanation. But I see it now! Thanks for helping me learn

  • @JoshuaPalley
    @JoshuaPalley 2 роки тому +5

    At 8:21, there is an open circuit between the left two wire bond pads on the voltage regulator.

    • @michaelmullins8181
      @michaelmullins8181 2 роки тому +4

      I guess this is laser-cut to adjust voltage?

    • @JoshuaPalley
      @JoshuaPalley 2 роки тому

      @@michaelmullins8181 that makes a good amount of sense, given what we can see. Thanks!

    • @funkyironman69
      @funkyironman69 2 роки тому

      I think those are actually test points with marks from probing?

  • @JFirn86Q
    @JFirn86Q 2 роки тому +1

    Short videos are nice, but I love the long format ones. Thanks again!

  • @JWH3
    @JWH3 2 роки тому +1

    Definitely scatter more of these in with the longer content!

  • @PROWEB7
    @PROWEB7 Рік тому

    WOW very nice vidéo, Thank you for the patience, for opening the devices, and your explanations ... Thank you very much

  • @NonEuclideanTacoCannon
    @NonEuclideanTacoCannon 2 роки тому +8

    That capacitor at 6:56 looks like it has a low-res gravel texture applied to it.

  • @ranjitneelakandan9551
    @ranjitneelakandan9551 2 роки тому +2

    Really enjoyed it. Thanks for the time and effort!

  • @Darieee
    @Darieee 2 роки тому

    fantastic stuff ❤️

  • @electronic7979
    @electronic7979 2 роки тому +1

    Nice repair video 👍

  • @samanthaqiu3416
    @samanthaqiu3416 2 роки тому +17

    a very poorly taught aspect of electronic design, please make more content on this absolutely fascinating subject

  • @0MoTheG
    @0MoTheG 2 роки тому

    8:45 Look to the left between the two pads. Looks different than the other three. Just perspective?

  • @bfx8185
    @bfx8185 2 роки тому +1

    Definitely yes for such videos! THX

  • @amirb715
    @amirb715 2 роки тому +3

    absolutely looking forward to more of these

  • @purerhodium
    @purerhodium 2 роки тому +1

    Looking at RF magic under the microscope is always fun.

  • @rfengr00
    @rfengr00 2 роки тому +1

    5:35 how do they bond down through a hole? I’ve done manual wire bonding before, and you’d have to get the wire threaded through the tool, and under the wedge. Crazy.

    • @luc7luc7
      @luc7luc7 2 роки тому +3

      Usually they only bond on one side then pass the wire manually through the hole and gap-weld on the other side.
      It could be placed manually and gap-welded on both sides but in this case it looks it was ball bonded on one side.

  • @itsamemario8414
    @itsamemario8414 2 роки тому +1

    If the last stage is blown, then it may be that the amp was operated with an input signal but no output sink so the large output signal reflected back to the output stage transistor.

  • @patsparks8731
    @patsparks8731 2 роки тому

    I think those are Fujitsu (FCSI) GaAs devices. Three wire bonds on that one pad caused it to lift. Thanks for posting this. Beautiful microscope BTW.

    • @Firerexify
      @Firerexify 2 роки тому

      Probably just a weak bond foot missed during QA process.

  • @adberco
    @adberco 2 роки тому +4

    Voltage regulator is a Motorola 7805, mask F77T :)

    • @Thesignalpath
      @Thesignalpath  2 роки тому

      Nice! Thanks.

    • @SeanBZA
      @SeanBZA 2 роки тому +1

      From the looks of it yes, and the thin film resistor network also looks like it is applying the minimum load specification, so they really wanted the amplifier to have a stable supply voltage, trimming the resistor to exactly get a specific current consumption out of the regulator.

  • @jessiepooch
    @jessiepooch 2 роки тому

    Fascinating stuff! Thanks.

  • @tlrptg
    @tlrptg 2 роки тому +5

    that's a 5 volt regulator. it's marked clearly on the wafer.

  • @Darkknight512
    @Darkknight512 2 роки тому +13

    5V regulator probably ;)

    • @rickoneill4343
      @rickoneill4343 2 роки тому +12

      Lol I was not sure if he was joking or not.

    • @gorak9000
      @gorak9000 2 роки тому

      Pretty tough to tell looking at a chip layout... /s

  • @justin.campbell
    @justin.campbell 2 роки тому +4

    Why does the inside look so scratched up?

    • @Thesignalpath
      @Thesignalpath  2 роки тому +5

      Probably just handing and during wire bonding.

    • @gorak9000
      @gorak9000 2 роки тому +3

      That happens when the witch is mixing them around in the cauldron casting the spell into them so they work at microwave frequencies. She does them in large batches to keep the production costs down.

  • @LutzSchafer
    @LutzSchafer 2 роки тому +3

    Shariar maybe you should consider using EDoF software to reconstruct your manual focussing? You then get everything in focus even with non equidistant slice snapshots.

    • @Thesignalpath
      @Thesignalpath  2 роки тому

      Yes, but then the feed won’t be in real-time. But I’ll look into it. Thanks!

    • @LutzSchafer
      @LutzSchafer 2 роки тому

      @@Thesignalpath with a cubic phase plate in the DC path and a simple inverse filter with the transfer function of the phase plate you can have life EDoF. Some 10 years ago this was a Zeiss project I did but it didn't make it to the market for some silly reason. I still have the Matlab simulation if you are interested to understand the principle of operation. Since there is a slight disadvantage that we used to call the "banana shaped" PSF, now you can get a more symmetrical phase plate that has a straight axial PSF. The phase plate you can buy from a company in Denver. Let me send you the Matlab code so you can get an idea. ....
      On second thought, the single shot EDF project was much longer time ago around the turn of the century. It was indeed targeted for the semiconductor market but didnt make it because then only the cubic phase plate (p = X^3+Y^3) was commercially available and due to its square shape would have cut off some 5% of the aperture, which at the time was too much reduction in lateral resolution by Zeiss standards. So the project got dropped as great cost because it was already in the pre-production phase. If you are lucky, I might find a prototype demonstrator of that time which was optimized to run with 5-10FPS when the post Pentium era just started... I'll send you an email when I get around to that.

    • @Thesignalpath
      @Thesignalpath  2 роки тому

      @@LutzSchafer Very interesting! Feel free to drop me an email. :)

  • @joes5669
    @joes5669 2 роки тому +3

    Believe it or not, I have had some success popping off ball bonds (and other types of bonds) by hand under a similar microscope for troubleshooting purposes (like pulling drain bonds off to see which HPAs are shorted, that sort of thing). I have also successfully changed the bonds on a resistor biasing network by hand. Of course, this assumes one end of the bond is still attached to something. Obviously, this is only for fault analysis, and I wouldn't ship anything like that! I bring this up because I'm curious to see if you might be able to press the gate bond back into place to return some functionality to the amplifier. I use a very fine needle chucked up in a pin vise.

    • @frosty129
      @frosty129 2 роки тому

      it was hermetically sealed anyway, so after the teardown it was hopeless to begin with

    • @Firerexify
      @Firerexify 2 роки тому

      @@frosty129 hermitcally sealing a unit is not done for tuning performance, its to protect it environmentally.
      All tuning is completed before the seal. So if he did have the balls to tweezer bond that wire bond on the the MMIC face (a daunting task bonding on the a MMIC face)
      Then in theory it should work great.
      But you can see a burn mark on that same MMIC in question. So most likely it requires a mmic replacement.

  • @quantumelectricaldynamics4928
    @quantumelectricaldynamics4928 2 роки тому +1

    Why are there 2 gates? A matched pair or does that geometry have a better impedance? I’d like to see what happens if you take out that last stage and bypass it!

    • @simontay4851
      @simontay4851 2 роки тому

      By hand? I don't think so.

    • @quantumelectricaldynamics4928
      @quantumelectricaldynamics4928 2 роки тому +1

      @@simontay4851 The micro strip is wide enough you could put a piece of copper tape across the gap with tweezers and use a toothpick and silver epoxy for contact. After removing the transistor and it’s wire bonds with a tweezer of course. Those things were put in there by hand. Just takes a steady one. Impedance marching won’t be great with the extra matching network, but you could calculate roughly how wide you need to cut the tape maybe skip over the stepped impedance as long as the open stub created is too short to resonate in band. Would just be cool to see if it works.

    • @bansci
      @bansci 2 роки тому

      Agreed, would be great to see if bridging out the final stage you could still have a half decent amplifier. I'm guessing the mismatch would lead to large reflections, causing other issues though

    • @gorak9000
      @gorak9000 2 роки тому

      @@quantumelectricaldynamics4928 They were put there "by hand" with the help of a wire bonding machine, not literally "by hand". You think silver epoxy and a random chunk of copper tape is good enough to not have all kinds of odd parasitics at 40GHz??

  • @TheElectronicDilettante
    @TheElectronicDilettante Рік тому

    5volt (time index 8:29)? Is there any way of fashioning a vacuum chamber to place these hermetically sealed devices in that would allow for disassembly/reassembly? It’s hard to conceive how devices like these are produced by the same species that brought us Hometown Buffet and Jerry Springer. Thanks for the great videos!!

  • @Richardincancale
    @Richardincancale 2 роки тому

    Is F I on the dies for Fairchild Industries?

    • @luc7luc7
      @luc7luc7 2 роки тому

      Fujitsu, most likely an FHX13x

  • @anthon5504
    @anthon5504 2 роки тому +1

    yee, another video!
    enjoy and appreciate it very much, thanks!
    since im not that rich, i'll pay you with a nice thum

  • @frankbumstead3838
    @frankbumstead3838 2 роки тому

    Yes, very interesting. But your very long videos are also good.

  • @askild_eide
    @askild_eide 2 роки тому +1

    But what about the small 10MHz to 26GHz amplifier? Did it have the same fault as the other one?

    • @Thesignalpath
      @Thesignalpath  2 роки тому +1

      It looks like it has the beginning of a similar failure.

  • @raa9774
    @raa9774 2 роки тому

    Hello, I have a faulty Fluke 355. Does anyone know where to find the service manual of it. I have searched all over the internet but didn't find nothing. I would really appreciate your help. Thank you very much

  • @NiHaoMike64
    @NiHaoMike64 2 роки тому +1

    On the unit with the bad stage, couldn't you just remove that stage and jumper across it?

    • @Thesignalpath
      @Thesignalpath  2 роки тому +3

      It will disturb the matching of the final stage, not to mention that you will lose some gain.

    • @simontay4851
      @simontay4851 2 роки тому +2

      One does not "just remove and jumper across" such microscopic components. He would need a wire bonding machine at the very least. ...and the rest.

    • @benmodel5745
      @benmodel5745 2 роки тому +2

      @@simontay4851 He did tweet about owning a wire bonder last year, soo....

    • @Thesignalpath
      @Thesignalpath  2 роки тому +10

      Yes, I do have a wire bond machine. Maybe I will give it a try in some future video.

    • @askild_eide
      @askild_eide 2 роки тому +2

      @@Thesignalpath We do want to see you using the wire bonder, so this could be a good candidate, and in best case, you end up with an amplifier with 3/4 of original gain. And worst case, it don't work at all, so same status as now.

  • @antebrizic886
    @antebrizic886 2 роки тому +1

    Where are you getting those devices from ?
    I'd like to build myself a home lab but the prices of instruments and parts are really through the roof ):

    • @derrekvanee4567
      @derrekvanee4567 2 роки тому

      Keysight has a giveaway every year I've been enyer8ng for years but the rf stiff and pro stuff ain't cheap. Get a da1xx diy scope decent dmm a breadboard and hunch of components for following some,tutorials thst interest you.

  • @benmodel5745
    @benmodel5745 2 роки тому

    Jeez that's a cool microscope

  • @KeritechElectronics
    @KeritechElectronics 2 роки тому +3

    The blackest of the black arts in analog electronics. This shade of black is so black that the only blacker thing would be a perfectly black body.

  • @1nn0centBystander
    @1nn0centBystander 2 роки тому

    Dear Community,
    Would you advise me on troubleshooting and calibrating my 1993 Studiomaster P7 analog mixing console?
    Theres absolutely NOTHING out there. I wish to troubleshoot the channel strips and start using my multimeter. I just need some guidance or point me in a right direction if you know a suitable guide/video. Im quite new to audio tech so still on beginner level.
    Many thanks, any suggestions appreciated. Really keen to understand this so to pass this knowhow in our studios.

  • @EvanZalys
    @EvanZalys 2 роки тому +1

    Those HEMTs may not be pseudo morphic. They look like fujitsu/eudynia devices. See FHX45X

    • @Thesignalpath
      @Thesignalpath  2 роки тому

      I think that could be right! Thanks for the info.

    • @EvanZalys
      @EvanZalys 2 роки тому +2

      @@Thesignalpath I've spent many hours doing horrible things to those HEMTs... cleaving them in half, epoxying them to the cut edge of dice... I'd spot them from a mile away. That said, I think there's internal connection between the gate pads. We've used them with only one gate pad bonded, though maybe the performance suffers. I did manage to spot a little black mark in the channel near one of the gates, so maybe it shot through there.

    • @Thesignalpath
      @Thesignalpath  2 роки тому

      @@EvanZalys As you said the two gates are internally connected. Using only of the two would likely hurt the device fMAX. The gate is "peeled" off almost in the bad die.

    • @0MoTheG
      @0MoTheG 2 роки тому

      thx I was wondering was a "gas pm" is.

  • @rajkiran3989
    @rajkiran3989 2 роки тому

    Would love to know about your study back ground. What books did you study etc...love from india.

  • @DavidMG99
    @DavidMG99 2 роки тому

    👍👍

  • @donaldviszneki8251
    @donaldviszneki8251 2 роки тому +1

    I want to see fabrication of these devices. If you could make that video, I would be blown away.

  • @tamaseduard5145
    @tamaseduard5145 2 роки тому

    👍💪🙏

  • @SanthoshKumar-mr5ss
    @SanthoshKumar-mr5ss 2 роки тому

    5v regulator!?

  • @donaldviszneki8251
    @donaldviszneki8251 2 роки тому

    DIC 3D effect is pseudo-3D. DIC is AFAIK really only useful for transparent specimens, though the surface roughness visualization was a nice surprise.

    • @Thesignalpath
      @Thesignalpath  2 роки тому

      It is useful for both transparent and non transparent subjects but as you said the primary benefit for top illuminated subjects are for roughness and surface analysis. The quasi-3D effect is neat to see in this sample probably because it is not very planar.

    • @donaldviszneki8251
      @donaldviszneki8251 2 роки тому

      @@Thesignalpath you should do a video about it! Afaik DIC's benefit is visualizing changes in refractive index

    • @donaldviszneki8251
      @donaldviszneki8251 2 роки тому

      @@Thesignalpath Does your setup allow you to use a single wavelength light source?

    • @Thesignalpath
      @Thesignalpath  2 роки тому +1

      @@donaldviszneki8251 I do have filters I can use on my current light source.

    • @donaldviszneki8251
      @donaldviszneki8251 2 роки тому

      @@Thesignalpath I'd like to see how that looks. This is my first time realizing that DIC is good for anything beyond transparent specimens. I'm still skeptical, but I've now read that grades and to some extent even distances on the scale of the wavelength used to illuminate the specimen.

  • @timpattinson1894
    @timpattinson1894 2 роки тому

    No sound?

  • @Alex-bi8ob
    @Alex-bi8ob 2 роки тому

    These Junkosha cables are not cheap either :)

  • @c4t4l4n4
    @c4t4l4n4 2 роки тому +1

    5000 milli volt circuit board for sure. 😁

  • @derrekvanee4567
    @derrekvanee4567 2 роки тому

    Freaking rf wizards making me go all wtf am I wasting my life with xyz