Is is possible to repair a hairline crack in semi-rigid coax?

[biastee]

I was motivated to move my sig-gen's (SMIQ06L) rear-mounted RF connector to the front after reading this eevblog message:
https://www.eevblog.com/forum/repair/rohde-and-schwarz-smiq06ate-to-rohde-and-schwarz-smiq06b/msg2947570/#msg2947570
The originally rear-facing N-female is connected a semi-rigid coax (L= 22cm, OD=3mm) with 4 bents (attached photo 1).
To move the said connector to the front, I needed to straighten the semi-rigid coax. Despite delicately straightening the bends, a hairline crack developed in the shield (attached photo 2). I suspect the hairline crack will degrade the shielding effectiveness at microwaves. Is there any way to 'save' this semi-rigid? I am thinking of covering the crack with self-adhesive 1cm wide copper tape (3M Scotch, attached photo 3). Is there a better way? What do I need to watch out for? Thanks.

[paul@yahrprobert.com]

Assuming that the coax outer conductor is tin plated copper and not aluminum you could wrap a piece of plain copper foil around the break and solder it on.  You'll need a pretty good soldering iron to get enough heat.

[Gyro]

I don't see any bare copper in the fracture line. The insulation looks like polyethylene too, which would melt under soldering temperature.

[PartialDischarge]

Just apply copper tape after cleaning with solvent, put a small cable tie at each side to hold it in place for ever...

[TimFox]

After applying the copper tape, can you hold everything together with heat-shrink tubing?

[joeqsmith]

... I needed to straighten the semi-rigid coax. Despite delicately straightening the bends, a hairline crack developed in the shield (attached photo 2). I suspect the hairline crack will degrade the shielding effectiveness at microwaves. Is there any way to 'save' this semi-rigid? I am thinking of covering the crack with self-adhesive 1cm wide copper tape (3M Scotch, attached photo 3). Is there a better way? What do I need to watch out for? Thanks.

Your idea and mine differ as to what a hairline is.  That appears more than 0.3mm.  The SMIQ06L is working at several GHz and I would expect that discontinuity to degrade the performance.   

Had you asked before damaging the coax, I wonder had to tried to anneal the copper first (assuming it is copper) could you have prevented the damage. 

You may be able to save the connectors but I would just replace the coax.    Even if I was only going to use the system at very low frequencies and didn't require it to meet the original specs, I would still replace it.   But hey, I use torque wrenches and am told by many hams how stupid that is.  So take it for what it's worth.   

That said, let's have a little fun with it.   I took a scrap section which I believe is a similar size to what you have.  I then bent it, straightened it, then bent it the opposite.  No cracks but has some bad wrinkles.   I then straightened it again and it cracked.  Showing the split next to my calipers with a 0.25mm gap.   

[joeqsmith]

Good point about the copper.  It could very well be silver plated aluminum which is what I am using.  Shown is our damaged section after using a razor blade to shave it. 

[joeqsmith]

So lets try and replicate your damage and see if we can somehow repair it.   First we need a way to measure it before the damage, after the damage and after repairs.  I don't own a time machine so I will attempt to recreate what you have shown. 

[TimFox]

I'm not aware of silver-plated aluminum in this application, although silver plating of aluminum is possible, but silver-plated copper is common.
I would worry about metal fatigue of aluminum in this application, compared with OFHC copper.
Correction: tin- and silver-plated aluminum are common in semi-rigid coax.

[joeqsmith]

Using my old Agilent PNA to measure the cable, using my software.   Full disclosure, the calibration algorithms that wrote for the LiteVNA are based on various papers from HP and Agilent.  These recently came into question from a member and these are the same algorithms I use with my software for the PNA.   While I took the time to try and walk the viewer through how my algorithms compare with those supplied by Agilent that run locally on the PNA, they never responded.  I could have some errors.  You can find that detail here:

https://www.eevblog.com/forum/rf-microwave/experimenting-with-waveguides-using-the-litevna/msg4796609/#msg4796609

A local member here also was able to provide me with details on how to enable the PNAs higher range.  It is now limited to 9GHz (which we will use to make these TDR comparisons).    I am using a homemade short and load cal standard along with a sorted Mini-Circuits terminator that had the best return loss of a group I purchased.   I used another members Matlab scripts to calculate the coefficents based on a set of Aglient standards.    The problem is, that data was all taken at 6GHz.     Point being, the measurements will have errors but because we are looking at a relative change, should be alright.

I started out measuring the cloned test cable that was terminated to 50 ohms and saved that as my baseline (Step1).

Next I bent the coax to form a small split (Step2).  This time I only bent it to the point where about half the shield had cracked.     

I then attempted to use some 3M copper foil to repair the damage.  There was little change so I attempted to press the foil into the gap with a sharp edge (Step3).

I then tried to solder the shield back together (Step4).

Maybe your repairs were successful and you were able to maintain the original performance but I doubt it. 

[joeqsmith]

One way to perform the repairs it is to remove the bad part of the cable and add on a bit of extra.   I saw a ham once who made a video where they connected several adapters in series and then claimed it had minimal effect. 

While you can see we did remove the hump that was caused when we damage the cable, we also picked up a new disturbance from our added connectors.  What worked for our friend the ham may not work for your setup.   My advice is still replacement and learn from it.   

[EE4all]

Replace it. Remember, the signal is traveling along the INSIDE of the shield, so even if you have repaired it outside, you will in most all cases still have an internal impedance bump.

[biastee]

I don't see any bare copper in the fracture line. The insulation looks like polyethylene too, which would melt under soldering temperature.

Yes, there isn't bare copper. It's aluminium.

After applying the copper tape, can you hold everything together with heat-shrink tubing?

A heat shrink tube would be best, but the toroid rings stand in the way. As the 2nd best solution, I will tightly wrap electrical tape around the Cu tape.

A local member here also was able to provide me with details on how to enable the PNAs higher range.  It is now limited to 9GHz (which we will use to make these TDR comparisons).    I am using a homemade short and load cal standard along with a sorted Mini-Circuits terminator that had the best return loss of a group I purchased.   I used another members Matlab scripts to calculate the coefficents based on a set of Aglient standards.    The problem is, that data was all taken at 6GHz.     Point being, the measurements will have errors but because we are looking at a relative change, should be alright.

I started out measuring the cloned test cable that was terminated to 50 ohms and saved that as my baseline (Step1).

Next I bent the coax to form a small split (Step2).  This time I only bent it to the point where about half the shield had cracked.     

I then attempted to use some 3M copper foil to repair the damage.  There was little change so I attempted to press the foil into the gap with a sharp edge (Step3).

:-) Using a torque wrench is a reasonable behaviour in my book. I will only ostracise those who brandish the depth gauge!

I am amazed by the length you have gone to replicate my cable - down to the toroid rings. Thank you very much for the time, effort & your considerable expertise. Your TDR measurement really opened my eyes to another error mechanism. I initially thought the problem will be confined to signal leakage when measuring receiver MDS / sensitivity.

Pertaining to your TDR measurement, I can't help but notice that the undamaged cable has deviations that are almost as large as the cracked one. Assuming 49.8ohm as the nominal Zo, the max. deviations in both directions are +1.4 ohm and -1.8 ohm. I guess the deviations are caused by a bent which has the effect of compressing the dielectric on the inner side, while stretching it at the outside.

Going back to the taped cable, what are the potential effect/s of the Zo changes on the sig-gen's performances? Degradation in the amplitude flatness over frequency? Referring to the TDR graph, the taped cable's Zo variations don't look significantly worse than the undamaged's.

On replacing the semi-rigid, one problem is I live in a Third World country where microwave stuff are hard to come by. 

73, Leong, 9W2LC.

------------
[Update]
I am trying to figure out how the deviation from nominal Zo can affect sig-gen's performance. According to Joeqsmith’s TDR measurement, the Zo rises to 51.8 ohm at the crack location (L = ~0.38m). 
Eyeballing the gap, its widest point is ~ a quarter mm and the crack covers ~1/3 of of the coax circumference. Assume after taping, the discontinuity’s equivalent length is 0.1 mm. The wavelength Λ at 1 GHz in coax with 0.7 Vp = 300 x 0.7 = 210 mm. So, the 0.1 mm gap translates to electrical degree at 1 GHz, (0.1 / 210) x 360 = 0.2 deg.

The equivalent discontinuity is modeled using the TLIN component in a linear simulator (attached image). Yes, the simulation predicts a degradation in insertion loss compared to the good unblemished coax. However, the loss is still insignificantly small at the sig-gen's max freq. Likewise, the degradation in return loss is also too insignificant to matter.

To all commenters: Thank you for the useful suggestions & words of advice.

[joeqsmith]

:-) Using a torque wrench is a reasonable behaviour in my book. I will only ostracise those who brandish the depth gauge!

Ostracise away. 

Thank you very much for the time, effort & ...

Not sure how helpful it was.  My VNA is not nearly good enough to look at these small gaps with any resolution.  I have a homemade 70mm air-line Beatty standard and even that is pushing it.   Looking at < 1mm damaged are, the best we can see if that it did cause a reflection.    I did have a ham once tell me how I could increase the resolution by decreasing the number of points!  So there's that...   

I can't help but notice that the undamaged cable has deviations that are almost as large as the cracked one.

I want to be sure that you understand we are not only looking at the cable but where the cable solders to the pin, where the pin fits into the connector, where the pin mates .....  There are several discontinuities all smeared together.   If the old PNA wasn't limited to 9GHz, we could get a much better view of what is going on. 

So, for fun here is another bit of the same coax.  This time I am using a better connectors and terminator.  Now that I understand that you are interested in how the discontinuities of the connectors compares with those caused from the damage, I have taken the time to properly torque everything. 

I once again took a baseline measurement of the cable after assembly (Red).   I then bent the cable 90 degrees (Violet).   After, I straightened the coax and bent it one more time.  When straightening it a second time it once again cracked (Blue).   

***
Also, if not obvious, the disturbance to the right is from our connector, adapter and terminator all lumped together. 

[joeqsmith]

Just for completeness,  I snipped off a small bit of 3M tape and wrapped it around the split an took another measurement.   I zoomed into that area to give you a better idea of the effect is has.  Sure, there was some slight improvement but not what I would ever consider for a repair.

[biastee]

@joeqsmith
I noticed the y-axis scale (impedance) has changed in the new graphs. Did you normalize to 50 ohm? What is the peak impedance value now?

[joeqsmith]

@joeqsmith
I noticed the y-axis scale (impedance) has changed in the new graphs. Did you normalize to 50 ohm? What is the peak impedance value now?

If you look at the two plots, the initial was showing the step response.   The last plot shows the impulse response.   The step response is calculated by taking the integral of the impulse.  Z is then Z0(1+S11)/(1-S11).   

There are a lot of very good papers on TDR that the main players have published.   

Almost all of my math was derived from various HP papers which are readily available on-line.   I have ran into mistakes in some of these papers and have posted about some of them on this site.   So readers beware.   It's also why I wasn't too surprised when a viewer suggested their was an error in my algorithms. 

****
This short article describes the basics:
https://incompliancemag.com/article/time-domain-measurement-tdr-or-vna/

[biastee]

If you look at the two plots, the initial was showing the step response.   The last plot shows the impulse response.   The step response is calculated by taking the integral of the impulse.  Z is then Z0(1+S11)/(1-S11).   

I don't know how to make sense of the impulse response's y-axis (impedance). I can understand Zo = 50 ohm, but can't wrap my head around a baseline Z = 0 ohm.

Anyway, gimme your best guess: what sig-gen's parameter will be impacted? And how much? Thanks.

[joeqsmith]

Anyway, gimme your best guess: what sig-gen's parameter will be impacted? And how much? Thanks.

If you like guessing, maybe try finding one of these:
https://en.wikipedia.org/wiki/Magic_8_Ball

If you want data, return it to the manufacture and have them calibrate it.  They can provide you with a report.  Beyond the damaged cable, they may identify other problems.

[joeqsmith]

One thing I did try was to take some very fine copper wire and wrapped it around the damaged area making sure it was laid into the cracked area.  I then soldered the entire area and filed the rough spots.   Looked nice but was much worse than the copper foil.  I suspect the copper wire may have melted into the dielectric insulation. 

[syrnyca]

One way to perform the repairs it is to remove the bad part of the cable and add on a bit of extra.   I saw a ham once who made a video where they connected several adapters in series and then claimed it had minimal effect. 

While you can see we did remove the hump that was caused when we damage the cable, we also picked up a new disturbance from our added connectors.  What worked for our friend the ham may not work for your setup.   My advice is still replacement and learn from it.   

Regarding the video you posted of that ham  - sure those principles don't apply when it comes to higher frequencies and, as we've seen in this case, coax repairs, but I have to ask what exactly is incorrect in the ham's analysis?
For lower frequencies HF and maybe some VHF, adapters don't introduce much in terms of insertion and reflection losses. Provided they are good quality and properly mechanically fastened adapters, of course. Or is this statement untrue?

[joeqsmith]

One way to perform the repairs it is to remove the bad part of the cable and add on a bit of extra.   I saw a ham once who made a video where they connected several adapters in series and then claimed it had minimal effect. 

While you can see we did remove the hump that was caused when we damage the cable, we also picked up a new disturbance from our added connectors.  What worked for our friend the ham may not work for your setup.   My advice is still replacement and learn from it.   

Regarding the video you posted of that ham  - sure those principles don't apply when it comes to higher frequencies and, as we've seen in this case, coax repairs, but I have to ask what exactly is incorrect in the ham's analysis?
For lower frequencies HF and maybe some VHF, adapters don't introduce much in terms of insertion and reflection losses. Provided they are good quality and properly mechanically fastened adapters, of course. Or is this statement untrue?

I did not make any statements about their analysis. Seems you've answered your own questions.  It depends on your requirements. 

[coppercone2]

I think you can plate fill these things the challenge would be not getting electrolyte under the skin.


Theoretically if you sensitize the plastic, and micro grind the crack into a V shape, it should fill 100% with plating and be as good as new if you have clean electrolyte that makes low stress copper.. the process is called electroplate joining or electroplate welding or electrochemical joining


what I would be curious about though is how good using the highest conductivity MG chemicals silver epoxy to fill the crack

[Bud]

Fill the crack with SMT paste and reflow? Molten paste will fill the gap down to the inner side of the shield.

[coppercone2]

the problem with all temperature things is that its clad over a low max temperature substance. if it was teflon filler your fine probobly but I would still recommend just pressing a big soldering iron into it instead of using air so its faster heat transfer

if its poly ethylene you need to be cold


foam is mega unstable with temperature increase


I don't unders tand how your supposed to terminate this stuff if you cant solder the outer bit though. usually hard line is soldered on the outside for the connector. I guess its like swaged or some shit