Microstrip repair - expert ideas to repair PCB microstrip

[Ick]

Dear most valued EEVblog community,
reaching out to collect some expert ideas and experiences for a repair.

I recently got an Agilent E7495B basestation tester from a surplus seller I purchase from here and then for a quite fair price.
I repaired the display gasket (known issue with these devices) and the unit was in very good optical condition overall. Considering these are heavy use service devices, this one almost lookes in mint condition.
While testing the device I could notice two problems a) software problem -  this shall not be the topic here, updating the firmware, probably broken flash memory. and b) an attenuation on the specturm analyser of ~34dB  at lowest frquency and 28dB at 2.2GHz (my R+S SMIQ does not go higher)

For the firmware upgrade I needed to open the unit, flipping dip switches on the Processor Board Assembly,  to force booting from a external media (PC Card / PCMCIA) as the flash image is corrupt (and maybe the flash chip itself, we will see..)
So I actually wanted to focus on the software upgrade to recalibrate the unti via GPS etc., but conincidently I discoverd probably the cause for issue b), the bad attenuation:
While removing the RF board I could see something that looked like a tiny surface scratch on the PCP stripline below the RF In N connector. After I fully removed the RF board, super carfully, I could see a similar "scratch" on the backside of this stripline.
See photos attached. Nothing too good I thought.
I suspect that besides the unit seems very rugged, the outer N connector must have gotten some impact or was bent with some lever to the side and probably has cracked the PCB and the inner stripline. This would perfectly explain the high attenuation getting smaller with higher frquency.
I have no clue why Aglient directly attaches a PCB to an N connector, instead of using rigid coax. I have never seen a connector directly screwed to a PCB...anyways.

To validate, I attached a simple Nano VNA and did a TDR measurement. This shows a clear open, so a broken conductor in the stripline, at exactly the "scratch" position. See photos also attached.
From the optics I am quite certain this is a stripline PCB, two ground planes on the outside with a center conductor in the middle of the multilayer board.

Question to people with experience: Is there any way you would recommend to repair this? (I know the option throw it away and I also know it will impact the non existent calibration as well)
* Attempting to grinding up the PCB down to the center conductor and solder on a coax and try to sheld that?
* Cut the outer PCB and try attach a coax somewhere in side the RF housing?
* try halve-drill the PCB down to the center conductor of the microstrip and solder on an MMXC connector on the PCB (might be better shielding, if at all)
*Any other good idea?


Any hints from people with repair experience would be greatly apreciated!
BR ICK

[bob91343]

I would go the heavy handed route and just flow solder over the break.  If that doesn't do it, I'd probably try to find a used assembly that works.

[Miti]

I would go the heavy handed route and just flow solder over the break.  If that doesn't do it, I'd probably try to find a used assembly that works.

There’s no way that the solder would go into the crack and close the circuit. On the other hand, probably that part of the board is just an appendix from an expensive RF board. If it was my instrument I would cut the board before the crack, grind the top ground plane and board on one side of the stripline for about 2 - 3 mm to reach the copper, then solder a piece of semi rigid coax cable with an SMA connector at the other end.

[KrudyZ]

Looks like a fully shielded stripline, not a micorstrip.
Have you opened the shielded case to take a look inside?
There might be a better place to access the traces you're interested in.
You could probably cut it where the stripline resurfaces and add a rigid coax there.
Of course you would have to do the same to the other channel for symmetry.

[perieanuo]

my 2 cents, you have metal residue shorting to ground the signal. try eliminate this with precision knife, acetone, proxxon milling, whatever works for you

[Ick]

my 2 cents, you have metal residue shorting to ground the signal. try eliminate this with precision knife, acetone, proxxon milling, whatever works for you

Thanks a lot @perieanuo. The TDR analysis relatively clearly indicates an open, not a short. This would also be intuitive to me, because where should the short come from with ~750um of dielectric on both sides of the center conductor, when a crack goes throgh this sandwich..

BR ICK

[Ick]

Dear all,
many thanks for your fast replies and providing me your view and ideas.
Overall not the simplest task probably,  but based on your recommendations and ideas I think I am decided on how to proceed.

I would first try to fix the PCB:
- cut a 2x4 mm opening into the copper top layer, centered on the strip line, over the crack position
- work carefully down with a Dremel with a small milling tool ~400-500um
- then using a 2mm blade soldering tip and heat, soften and excavate the epoxy until I hopefully lay open the broken trace
- clean it, cut out 2mm of the broken trace, insert a new copper strip of equal width bridging the crack and solder it on
- then fillig up the excavation with 5 min epoxy with some ground PCB material and leveling the surface again using the Dremel grinder
- put 2x4mm coper clad on top again and solder on
- add some protective coating (solder mask or PU spray)
- finally try reenforce the PCB with some extra epoxy casted around the 90 degree angled area to increase stiffness

Will require proper light, magnification, paitience... (ideally I would hire a dentist for this  )

If that salvage attempt would not work out, I could then still move on to the more coarse technics as proposed here: Cutting the board, soldering a Coax or find alternate contact points on the RF board inside the shielding can.

Thanks to all. I will try to shoot some pics and report success or failure and learnings here in a while, I will dedicate one of the next weekends to this.

BR ICK

[perieanuo]

my bad, i supposed it's signal attenuation

[Ick]

Which it is . Pretty bad, 34dB of attenuation as described to be exact.
The crack would work like a pretty small series capacitor in the RF line. Two sides of the broken trace with air in the middle forming the cap. And that equals to a DC open.
If math does not trick me the crack should introduce a 0.64pF capacitance (= 2.4kOhm reactance @100MHz) in series (34dB attenuation is 10^1.7= almost exactly factor 50 in field quantity; )

[perieanuo]

don't forget something: if they put some termination resistor or some amplifier inside that shielded box near the rf connector and you have short to GND between the box and the device, the device may say also the "connector side " isn't there, cause it doesn't "see" the resistor; i'm no expert in radio stuff, but this may happen