Yeah a packaged part will never work. You will ruin the module beyond repair trying what you described. Not to mention the parasitics at microwave. I don’t believe that can be done. And if you determine the part number of the bad die and purchase a replacement or identify a suitable alternative, which is probably the only viable path forward, you’d need to scrape out the bad one and silver epoxy down and cure the new one. Very delicate process. You need an industrial oven to cure it properly. Obviously not an option so you have to let it self cure by letting it air dry for a few days, which still might not work. Then wire bond it. But you don’t have the equipment needed for these modules, so unfortunately you are probably SOL.
What about making up an external box with amplifier, switches, doubler, filter and whatnot... then modifying the module to always run through the low band path? Chip out the first amp and switches and replace with some easier to work with substrate, use conductive epoxy and copper ribbon to bond over all the gaps?
The 5th channel on the Quintbrick mixer and IF digitizer is for use in the 4 port PNA-L. As a low cost 4 port, one incident channel is shared for all 4 ports with the splitter before the port switch. It makes TRL family calibrations impossible without another method to determine switch terms. PNA-X and probably the regular PNA have 8 receivers on the 4 port models.
I had the power supply fail on our N5230A. The replacement was quoted at 2K, but instead I just took it apart and replaced the two bootstrap capacitors, fixing it for about $2. Thanks for your very helpful videos.
Reminds me of the point to point wiring seen in the Vacuum Tube era, just a bit smaller and all gold plated. Great content and details, looking forward to the next video!
Hey Shahir, great video. I suggest you attempt to buy the die and perhaps go to one of the unis near you and see if they have a wire bond machine? might make a good video as well.
Beautiful man, even if this doesn't work, I really learned so much watching this video. first time vie ever seen someone go this deep on an RF repair! using a small QFN swicth will be interesting!
Excellent analysis as usual. Wire bonding machine on alibaba for a bit less than $2000 would be the trick. No point buying it unless you're absolutely positive a die is available for purchase.
I'm a bit conflicted - On the one hand, I would like to see a repair with a QFN-Chip to see how well a packaged chip that is 21 years younger can hold up against a bare die. And also to see if such a device can be repaired in this way (if so, ebay-prices for defective network-analyzers will spike again, I think :D). But on the other hand, I would also like to see how a wirebonding-machine works and how the dies and ceramic modules are being put into the metal casing. As well as how to remove the die without having debris flying all over the place, possibly causing further damage. Either way, it'll be a very interesting 2nd part. :)
Nice video Shahriar, thanks. Interesting - never seen that kind of mesh bond before. Know of any test equipment breakers that have any used complete modules? There area few on ePay that come up. On the other hand I'd like to see a 'hobbyist'-style best-efforts fix with a packaged SPDT switch.. but that's a bit of a one-way mission.
Your videos are so damn clear ! And one has to marvel of the beautie of the engineering of the whole thing. Then you say 'jam it in there', I nearly got hernia :D
If you totally destroy the switch in the repair can you still bodge the pass through so that it still works under 10.5GHz.? If yes you have nothing to lose.
Probably, but there's also the possibility that the switch has some attenuation that is required in the signal-chain to be within an area that can be regulated by the ALC-Circuits.
You could look at the input/output loss of the good (identical) switch to determine RF loss. Touch a probe on both sides to look at delta of relative RF strength. You could then put a SMD attenuator to replace the bad switch to accomplish what George suggests while addressing your concerns.
If you short the signal to both the straight path and the x2 path all the time i.e. bypass the input side switch, would the working output side switch provide enough rejection of the x2 signal when it is not wanted? If not, could the power to the x2 amp be disabled by external logic when it is not required? Would it wake up quickly enough? If not, could the mystery circuit mentioned as an attenuator at 34:31 be brought in to play to reduce the x2 signal level when below 10.5GHz? I assume the input side switch must be there for a reason, but I don't know exactly why.
A special saturday night treat! A repair video. Most excellent! Followed the block diagram along with you, I liked this way- laggy pc or not, still learned a great deal Test the mesh for open (moderately) separated-waveguide RF transmission attenuation too! (Along with the other suggestions last videos!) 40:41, the damage, hypothesis to what caused that? Was that more of an "etch" damage or "crusty" damage? Probably irrelevant, but, curious none the less As far as proceeding on part 2, it sounds like your plan of attack is the best choice (considering the equipment, or lack there of, at the moment -it would be VERY cool if you got a wire bond-er though!) ..bout the only other thing that may be best, is just finding a new module -maybe the none "L" versions would have a comparable one that could be found as "parts/scrap" and used just as well Anyways, great video! As usual Learned a lot and loved the look into the high tech side of electronics. Absolutely beautiful. *Thank you, again and again
Wire bonding is an art and even with all the equipment to remove the old die (typically achieved by scraping the die away with a scalpel), vacuuming out the cavities, applying new epoxy and placing the new die (taking care not to damage any airbridges, scratch any tracks/components and not chip the edges of a die). THEN you can think about wire bonding (which will require lots of time setting up the machine (ultrasonics, flame off, pressure, timings etc). Trading favours with an established company that has the equipment and trained personnel in this field soon becomes a lot more attractive! Perhaps they could trade wirebonding for a “sponsored video”?
Thanks for the great video. Started watching late last night but just watched the final 10 minutes. I'd look into getting a wire bonding machine anyway. If we've nothing to lose and can't obtain a replacement die *AND* you have a suitable candidate in qfn for why not depot the switch in it (that would be interesting in its own right) and build a test jig/pcb upon which you can test your bonding skillz. Once done and successful we can tackle the final repair. All would make a great series of videos. I second someoneelse's comment about reaching out to a uni or agilent for help sourcing a bonding machine.
28:28 That doesnt look cheap one bit.. every single bit placed on there by hand, impressive. And there are quite many scratches on the edge surface where there shouldnt have been any all from fab so this have been open some times/s before for sure, maybe the same problem who knows.
I got such PNA from an auction... But all the softwares has been lost. Only the windows xp SP2 is there with bunch of Agilent NA software. When turned on it says PNA Hardware not found DSP Driver not installed. Can anyone here help me getting it factory reset ? The CD/Floppy is not there with the VNA since I saw it in auction house. Kindly help me getting them. Agilent N5230, 10Mhz-20Ghz, PNA-L is my model to be precise. Please help.
You need a wire bonding machine in your lab anyway, but If you can get your hands on that die I'm sure it can be arranged for now at some company or university.
Wire bonding is one thing. But how to get the old die out and the new die in? Is that part obtainable atvall? Would be interesting to see how this type of circuitry is worked with, without throwing everything out of impedance matching. At the end of the day, this is not an ordinary pcb, isnt´t it. Any way of repair attempt is going to be an interesting challenge and video of course.
You are dealing with frequencies several zeros higher than I'm used to, so I apologize in advance if this is a dumb suggestion. Is it too much to hope for that the switch on the output side properly terminates the unused leg? If it did, then it seems to me that you could forego a switch on the input, and drive both paths in parallel, through a series resistor to correct the impedance. That might be an easier modification to manage than hacking in a different switch. You would lose 6db on both ranges.
6dB of loss is probably enough to get Unlevel-Errors across the entire frequency-span. The Unlevel-LED was already partially on when the device was at 0 dBm and the output was around -4.5 dBm in the video.
How are these things at 29:00 constructed? Did they cnc some metal and than clued ceramic pcbs and ic into the metal? What metal is it brass? Some aluminium with gold finish? Im not really sure how to call this and didnt found much on google. Would be nice if some can point me in the right direction.
Great vid. Is it fairly quick and easy to get that cover off the PNA? I can see in this vid, that the cover was initially on, and at 20:27, the cover is then off.
Thank you, that was fantastic! The sub-10.5 GHz path between the two switches (that path without the doubler) seems to consist of three striplines bonded together. What's the reason for that compared to one continuous piece of transmission line?
I would say that they have standard sizes of stripline-modules and use them for cost and handling-reasons. As the substrate is ceramic, I suspect them to be quite brittle. And the longer a piece, the higher the chance of them breaking during assembly.
Great video and very interesting stuff, especially inside the module. Sorry for the naive question as I don't have a lot of experience with these things, but how are the lines inside the module stripline and not microstrip? Does the cavity that the module is in provide the upper ground for the stripline?
I am facing an error which says the dsp hardware or driver not found Tried to install them, installs successfully but no hardware found. Can you help me out ?
I have similar model but front panel keypad is not working. I normal use the mouse to do the setting. Any tip to troubleshooting the fault. A3 board part no. E8361-20061 hope you can make a video how the keypad control board work.Thanks Awang snr electronics lab tech.
It is interesting that that component you showed as having problems does not appear to be obviously bad. What would be the failure mode? Can you look at it with a thermal imager on the component to see where things are heating up?
Of course I'd love to see you get a wire bonding machine. But if not would you perhaps be able to restore the specified output below 10.5 GHz by removing the switch entirely and jumpering the low path w/ fine wire & solder? At the expense of no longer having any function above 10.5 GHz.
I have SA N9320A like video: ua-cam.com/video/td9SOOgdzls/v-deo.html Sometime it arlam LO UNLOCK then I can not receive any signal I will open the packet but It looks difficult to open Does anyones has experience open this device? thanks
*Don't forget to watch to the end and let me know your thoughts on how to proceed.*
Looked up bonding machines. See my other comment
Yeah a packaged part will never work. You will ruin the module beyond repair trying what you described. Not to mention the parasitics at microwave. I don’t believe that can be done. And if you determine the part number of the bad die and purchase a replacement or identify a suitable alternative, which is probably the only viable path forward, you’d need to scrape out the bad one and silver epoxy down and cure the new one. Very delicate process. You need an industrial oven to cure it properly. Obviously not an option so you have to let it self cure by letting it air dry for a few days, which still might not work. Then wire bond it. But you don’t have the equipment needed for these modules, so unfortunately you are probably SOL.
How steady is your hand with a reeally fine tipped needle with a lil dab of conductive epoxy on the tip?
What about making up an external box with amplifier, switches, doubler, filter and whatnot... then modifying the module to always run through the low band path? Chip out the first amp and switches and replace with some easier to work with substrate, use conductive epoxy and copper ribbon to bond over all the gaps?
If the instrument isn't worth much money it would make for an interesting video for you to try to bodge the repair
The 5th channel on the Quintbrick mixer and IF digitizer is for use in the 4 port PNA-L. As a low cost 4 port, one incident channel is shared for all 4 ports with the splitter before the port switch. It makes TRL family calibrations impossible without another method to determine switch terms. PNA-X and probably the regular PNA have 8 receivers on the 4 port models.
Excellent! Thank you.
So this is the difference between TRL and TRL* calibration ?
Such a privilege to witness the genius of electronic engineers in 2018
I had the power supply fail on our N5230A. The replacement was quoted at 2K, but instead I just took it apart and replaced the two bootstrap capacitors, fixing it for about $2. Thanks for your very helpful videos.
People seem to forget that the journey might be nicer than the "destination" - What a great video. Thank you very much.
There is a 7 second offset of the audio from the video during the block diagram portion. At least on my computer.
Same, I've noticed a small offset on a couple of the other recent videos during screen recording sections.
Same here too
Same here, but seems good from around 13:00 or so.
Reminds me of the point to point wiring seen in the Vacuum Tube era, just a bit smaller and all gold plated. Great content and details, looking forward to the next video!
Just awesome. I'm definitely ill-equipped to tell you how to proceed, but I appreciate that you split the video and taught us so much!
20:47 A directional coupler ranging from DC to 65GHz, yea probably the cost of a small countries GDP..
Epic video. Jamming in a QFN switch would be mighty impressive but i would be equally interested in wire bonding. I eagerly await the conclusion.
There are those darn probe power connectors we were never able to find a couple of years ago!
Hey Shahir, great video.
I suggest you attempt to buy the die and perhaps go to one of the unis near you and see if they have a wire bond machine? might make a good video as well.
Wishing I had a few spare wire bonding machines....I would send one over right away.
I'd say get one of those household switches, the ones to turn on the ceiling bulbs, jam it in there, and call it a day.
Beautiful man, even if this doesn't work, I really learned so much watching this video. first time vie ever seen someone go this deep on an RF repair! using a small QFN swicth will be interesting!
Mikeselectricstuff's talking speed is contagious.
I am dying for the second part and want to see how you Replace/fix those microscopic compnents..
Excellent analysis as usual. Wire bonding machine on alibaba for a bit less than $2000 would be the trick. No point buying it unless you're absolutely positive a die is available for purchase.
I'm a bit conflicted - On the one hand, I would like to see a repair with a QFN-Chip to see how well a packaged chip that is 21 years younger can hold up against a bare die. And also to see if such a device can be repaired in this way (if so, ebay-prices for defective network-analyzers will spike again, I think :D).
But on the other hand, I would also like to see how a wirebonding-machine works and how the dies and ceramic modules are being put into the metal casing. As well as how to remove the die without having debris flying all over the place, possibly causing further damage.
Either way, it'll be a very interesting 2nd part. :)
Nice video Shahriar, thanks. Interesting - never seen that kind of mesh bond before. Know of any test equipment breakers that have any used complete modules? There area few on ePay that come up. On the other hand I'd like to see a 'hobbyist'-style best-efforts fix with a packaged SPDT switch.. but that's a bit of a one-way mission.
Your videos are so damn clear !
And one has to marvel of the beautie of the engineering of the whole thing. Then you say 'jam it in there', I nearly got hernia :D
If you totally destroy the switch in the repair can you still bodge the pass through so that it still works under 10.5GHz.? If yes you have nothing to lose.
Probably, but there's also the possibility that the switch has some attenuation that is required in the signal-chain to be within an area that can be regulated by the ALC-Circuits.
You could look at the input/output loss of the good (identical) switch to determine RF loss. Touch a probe on both sides to look at delta of relative RF strength. You could then put a SMD attenuator to replace the bad switch to accomplish what George suggests while addressing your concerns.
I have to admit I'm really interested in seeing what sollution you'll eventually come up with. Thanks for the awesome video
If you short the signal to both the straight path and the x2 path all the time i.e. bypass the input side switch, would the working output side switch provide enough rejection of the x2 signal when it is not wanted? If not, could the power to the x2 amp be disabled by external logic when it is not required? Would it wake up quickly enough? If not, could the mystery circuit mentioned as an attenuator at 34:31 be brought in to play to reduce the x2 signal level when below 10.5GHz? I assume the input side switch must be there for a reason, but I don't know exactly why.
A special saturday night treat! A repair video. Most excellent!
Followed the block diagram along with you, I liked this way- laggy pc or not, still learned a great deal
Test the mesh for open (moderately) separated-waveguide RF transmission attenuation too! (Along with the other suggestions last videos!)
40:41, the damage, hypothesis to what caused that? Was that more of an "etch" damage or "crusty" damage? Probably irrelevant, but, curious none the less
As far as proceeding on part 2, it sounds like your plan of attack is the best choice (considering the equipment, or lack there of, at the moment -it would be VERY cool if you got a wire bond-er though!)
..bout the only other thing that may be best, is just finding a new module -maybe the none "L" versions would have a comparable one that could be found as "parts/scrap" and used just as well
Anyways, great video!
As usual
Learned a lot and loved the look into the high tech side of electronics. Absolutely beautiful.
*Thank you, again and again
Wire bonding is an art and even with all the equipment to remove the old die (typically achieved by scraping the die away with a scalpel), vacuuming out the cavities, applying new epoxy and placing the new die (taking care not to damage any airbridges, scratch any tracks/components and not chip the edges of a die).
THEN you can think about wire bonding (which will require lots of time setting up the machine (ultrasonics, flame off, pressure, timings etc).
Trading favours with an established company that has the equipment and trained personnel in this field soon becomes a lot more attractive!
Perhaps they could trade wirebonding for a “sponsored video”?
Thanks for the great video. Started watching late last night but just watched the final 10 minutes.
I'd look into getting a wire bonding machine anyway. If we've nothing to lose and can't obtain a replacement die *AND* you have a suitable candidate in qfn for why not depot the switch in it (that would be interesting in its own right) and build a test jig/pcb upon which you can test your bonding skillz. Once done and successful we can tackle the final repair. All would make a great series of videos. I second someoneelse's comment about reaching out to a uni or agilent for help sourcing a bonding machine.
28:28 That doesnt look cheap one bit.. every single bit placed on there by hand, impressive.
And there are quite many scratches on the edge surface where there shouldnt have been any all from fab so this have been open some times/s before for sure, maybe the same problem who knows.
I got such PNA from an auction... But all the softwares has been lost. Only the windows xp SP2 is there with bunch of Agilent NA software. When turned on it says PNA Hardware not found DSP Driver not installed.
Can anyone here help me getting it factory reset ?
The CD/Floppy is not there with the VNA since I saw it in auction house. Kindly help me getting them.
Agilent N5230, 10Mhz-20Ghz, PNA-L is my model to be precise.
Please help.
Useful video
Is that a hair I see sitting near the doubler in the microwave module at 33:00?
I saw that too.
That is a spec of dust. It is tiny.
You need a wire bonding machine in your lab anyway, but If you can get your hands on that die I'm sure it can be arranged for now at some company or university.
You may be able to sand off some package material to make more room.
The wire bonded dies are absolutely fascinating. Where can I learn more about their design?
Wow that's a bugger! No idea how to proceed. Maybe do a saved search in Ebay for a spares machine?
Wire bonding is one thing. But how to get the old die out and the new die in? Is that part obtainable atvall? Would be interesting to see how this type of circuitry is worked with, without throwing everything out of impedance matching. At the end of the day, this is not an ordinary pcb, isnt´t it. Any way of repair attempt is going to be an interesting challenge and video of course.
You are dealing with frequencies several zeros higher than I'm used to, so I apologize in advance if this is a dumb suggestion. Is it too much to hope for that the switch on the output side properly terminates the unused leg? If it did, then it seems to me that you could forego a switch on the input, and drive both paths in parallel, through a series resistor to correct the impedance. That might be an easier modification to manage than hacking in a different switch. You would lose 6db on both ranges.
6dB of loss is probably enough to get Unlevel-Errors across the entire frequency-span. The Unlevel-LED was already partially on when the device was at 0 dBm and the output was around -4.5 dBm in the video.
How are these things at 29:00 constructed?
Did they cnc some metal and than clued ceramic pcbs and ic into the metal? What metal is it brass? Some aluminium with gold finish?
Im not really sure how to call this and didnt found much on google. Would be nice if some can point me in the right direction.
The fact that you've gone as far as to repair the actual die of the component is just ridiculous
Great vid. Is it fairly quick and easy to get that cover off the PNA? I can see in this vid, that the cover was initially on, and at 20:27, the cover is then off.
Yes, it is pretty easy. Just a few screws and slides off. It is heavy, so you may need a little help.
@@Thesignalpath Thanks very much TSP! Much appreciated and best regards.
Was there any visible damage to the switch?
I think you should go with a replacement, with micro wire connections. you might wait forever the other way.
Excellent video..... Recent subscriber!....where do you locate the block diagrams?
Thank you, that was fantastic! The sub-10.5 GHz path between the two switches (that path without the doubler) seems to consist of three striplines bonded together. What's the reason for that compared to one continuous piece of transmission line?
I would say that they have standard sizes of stripline-modules and use them for cost and handling-reasons. As the substrate is ceramic, I suspect them to be quite brittle. And the longer a piece, the higher the chance of them breaking during assembly.
Put an amplifier after the module and modify the alc loop to look at the signal after the amplifier?
Great video and very interesting stuff, especially inside the module. Sorry for the naive question as I don't have a lot of experience with these things, but how are the lines inside the module stripline and not microstrip? Does the cavity that the module is in provide the upper ground for the stripline?
Wow I'm so far behind I think I'm ahead. The file cabinets that your knowledge would fill could sink a ship.
Do you need a wire bonder? What about taking it to someone who has one.
Certainly a tricky thing to work on, its just so damn small !
I am facing an error which says the dsp hardware or driver not found
Tried to install them, installs successfully but no hardware found.
Can you help me out ?
Can we fix the *phase lost* problem??
I have a E5071B in need of repair, any suggestion where to send it to?
Did you ever post a part 2?
Yes: ua-cam.com/video/uirC7tABrFU/v-deo.html
My lab is the same, it grows by me buying faulty equipment to make repair videos with, and then I usually keep it for use on my bench.
still I did not finish video, I am in part where you mention the hard drive. what do you think in replace with ssd in this kind of instruments?
Are there any labs or companies you could team up with to access a wire bonder? Or perhaps an electronics department at a well funded University?
This must be revenge for the last couple easier repairs...
Just a couple of weeks ago I saw a wirebonding machine on ebay...
Would there be any other use than repair? These things take up so much space...
This video is truly R rated for RF-pr0n!
I have similar model but front panel keypad is not working. I normal use the mouse to do the setting. Any tip to troubleshooting the fault. A3 board part no. E8361-20061 hope you can make a video how the keypad control board work.Thanks Awang snr electronics lab tech.
It is interesting that that component you showed as having problems does not appear to be obviously bad. What would be the failure mode? Can you look at it with a thermal imager on the component to see where things are heating up?
Those are very low current from an OP-AMP output. There will be no heat dissipation at all even if shortened, leave alone with 90-130 Ohm.
Two tear downs in a week 😭 yusssss
So the LO (comes) from the other side.
Get a sharpie, cross over the 13.5GHz label, write 10.5 -6dBm, done ;)
👍👍
Cool
Of course I'd love to see you get a wire bonding machine. But if not would you perhaps be able to restore the specified output below 10.5 GHz by removing the switch entirely and jumpering the low path w/ fine wire & solder? At the expense of no longer having any function above 10.5 GHz.
What cpu
I hate bond wire... :( I try to avoid in all of my design at all cost. but thanks for the nerdgasm.
I have SA N9320A
like video: ua-cam.com/video/td9SOOgdzls/v-deo.html
Sometime it arlam LO UNLOCK then I can not receive any signal
I will open the packet but It looks difficult to open
Does anyones has experience open this device?
thanks