Don't buy SMA cable assemblies from CentricRF

[KE5FX]

I really don't know what "appreciable" changes are or what a "typical" wrench would be.   

That seems to be the heart of the problem.  What is "appreciable" or "significant"? 

When I write assembly instructions for EMS personnel, job #1 is to get them to use a torque wrench at all.  Training the assemblers in a 40,000-square foot ISO-certified facility in the Bay Area, it was surreal to discover that none of them had ever seen an SMA torque wrench before.  Although there was no shortage of RF hardware being shipped out of that factory, my product was either the only one that specified connector torques, or the only one whose instructions explicitly called for a torque wrench to be used.

Beyond that, I'm happy if the torque applied is within, say, -25% to +50% of the rated amount.  Decent-quality SMA connectors are going to work just fine in the 6-12 in-lb range.  Just as the equipment breakage described in the university lab manual .PDF was not caused by students holding a torque wrench incorrectly, the quality issues I saw weren't caused by tolerance violations.  They were caused by connectors being left untorqued entirely. 

Ultimately the best solution was a redesign that eliminated 80% of the connectors and left the others more easily accessible.  Lots of lessons were learned on that project, some by the assemblers but mostly by me.

The NRAO guidelines posted above are pretty interesting in that regard.  They are telling people to torque each connector 3 times, and I'm kicking myself for not thinking of that.  Instead of my assemblers being unsure if they have torqued a given connector at all, the NRAO assemblers are going to fret about whether they've applied the wrench 3 times or only twice.

No matter, finger placement has been covered but rather then posting gibberish,  feel free to prove me otherwise.  Plot the six or so data points I previously asked for.

Nobody is going to do all this work for you, Joe, and if they do, it probably won't prove what you're hoping it will.  (Which is what, exactly?)  A half-assed video is about all you can expect for free.

[joeqsmith]

Wow there's a lot of gibberish in this thread.

And thank you for taking the time to add even more gibber jabber.

Gee.  I can't imagine how your interactions with manufacturers didn't go the way you wanted them to.
 

Nice, more gibber added to the pile, compliments of the person complaining about it.

Wanted?  You do realize that there is NOTHING for me to gain by running such a test.  My offering to run an independent check wasn't /isn't something I wanted to do but certainly would.   

1) You cannot reach the recommended torque for an SMA with your fingers.  Use a wrench.  If you insist on being fast and imprecise, spend $5 on a hasco SMA finger wrench.
https://www.hasco-inc.com/tools/sma-3-5mm-2-92mm-2-4mm-1-85mm-connector-finger-thumb-wrench-blue/
That will at least increase the diameter to a size where you can apply the required torque.

I have some open end wrenches I could use if the goal was to over torque them.  I assume you are suggesting that is not possible with the tool you are recommending. 

You do know what happens when you assume right?

The sarcasm slipped past you. 

That tool means that you will at least apply enough torque to mate the reference planes and compress any gasket.  The basic, bare minimum requirement. 
It will remind you that these are just SMA connectors as it starts to slip, but they's nothing to stop you from just acting like a fool and tightening your grip.

That was my point.

4) Connector gauges are generally comparators against a reference standard.  Check your reference.  Is it good?  This is easy for SMA but harder for N.  Either way, I didn't notice any reference to zeroing the gauges.

I can easily remedy that.  See attached image.  Feel better now?

Now, how good is that standard?

Quit being lazy and tell us.  What number are you zeroing to? 

To the master gage.

[joeqsmith]

No matter, finger placement has been covered but rather then posting gibberish,  feel free to prove me otherwise.  Plot the six or so data points I previously asked for.

Nobody is going to do all this work for you, Joe, and if they do, it probably won't prove what you're hoping it will.  (Which is what, exactly?)  A half-assed video is about all you can expect for free.

Referring to:

If your torque wrench appreciably changes its delived torque based on where your fingers are on the handle, it is a POS and belongs in the trash.

I'm really not asking or expecting anyone to as it should be obvious that once you choke up beyond the pivot, you have bypassed the mechanism.  Your acknowledgement     

Yep, I was out of line to suggest that it didn't matter at all.  It clearly does, in the sense that grasping the wrench well away from the end of the handle can apply significantly more torque at breakover.

was all I was looking for.   

[joeqsmith]

The NRAO guidelines posted above are pretty interesting in that regard.  They are telling people to torque each connector 3 times, and I'm kicking myself for not thinking of that.  Instead of my assemblers being unsure if they have torqued a given connector at all, the NRAO assemblers are going to fret about whether they've applied the wrench 3 times or only twice.

I was surprised when I read that comment by Galen Watts.  I've never paid any attention to if the connector would move on a second  or third application.  They talk about temperature causing the connectors to loosen. 

I play a bit with motorcycles and assemble my own motors for them.   To torque the head, you work the pattern at a low level of torque, then the final level.   Then repeat at the final level.   Then you let the motor run and repeat it again at the final level.   Much higher torque levels, and wider temperatures.  Plus compressing copper gaskets.   

It's a different game than a little SMA but I guess they can have similar problems.   

***
Just to add,  most of the engine parts are aluminum.  Tempco is really poor.   I am surprised NRAO wouldn't use stainless to get that temperature stability.   Maybe they are. 

[KE5FX]

I play a bit with motorcycles and assemble my own motors for them.   To torque the head, you work the pattern at a low level of torque, then the final level.   Then repeat at the final level.   Then you let the motor run and repeat it again at the final level.   Much higher torque levels, and wider temperatures.  Plus compressing copper gaskets.   

It's a different game than a little SMA but I guess they can have similar problems.   

Yeah, ultimately we're arguing about how best to do a robot's job.  High-precision repeatability requires specifying all of the conditions including how to account for operating temperature variations (hence the NRAO notes.)

I will say that one problem with the six-point test is that there is no way to apply all of the force to one point or even two adjacent points without holding the wrench in a VERY unnatural way.  Such as by hooking one finger around it and pulling at 90 degrees. 

Doing that at the very end of the handle results in the correct torque.  Doing it below the halfway point results in substantial overtorque, meaning more than my personal abs-max limit of 12 in-lb.  But again, it proves nothing other than that the wrench can be misused if you try hard enough.

The other point you raised, regarding applying smooth force over about 2 seconds, is more important than I would have guessed.  The longer I spent trying to get a clear reading on the digital torque wrench, the less reproducible the result... or rather, the less confidence I had in it.

[joeqsmith]

...

I was expecting you to make a fully automatic test fixture with a simulated finger.   Your idea of using a robot would be fine.   EEVBLOG recently lost the worlds leading expert in robotics.  I bet he could have pulled it off with some fishing line, old socks along with his AI. 

***

...proves nothing other than that the wrench can be misused if you try hard enough...

Suggesting that it doesn't matter (hand finger placement) at all, invites such misuse.  We don't put warning labels on poison telling people not to drink it because we are intelligent.      

I have had friends who have worked in the medical and nuclear industries.  I am always amazed about how detailed their procedures are.   There's no room for cowboys.

***
Cowboy as a verb. 

a person who is reckless or careless, especially when driving an automobile.

An engineer running in a vacuum, no peer review, doing their own thing.  A prodigy, perhaps Mr Bigs kid.... 

Mr Big?  The man, boss, owner of small company, who ever you work for.... 

[KE5FX]

I'm sure he'll be back.  They always come back.

[mark432]

5) If your torque wrench appreciably changes its delived torque based on where your fingers are on the handle, it is a POS and belongs in the trash.

It has been proven mathematically that torque wrenches based on a click mechanism are sensitive to position of where the force is applied due to the difference in arm length between the rotation point of the jaw and the pivot point of the click mechanism. I'll admit it is something that takes some time & thought to wrap your head around but this is basic physics.

Here's a link to what I believe you are referring to


 I haven't checked his work but it seems right. 
I'll give you credit there.   

If you apply force in a different place than expected you can change the result. 

It would certainly be possible to design a clicker wrench that is super sensitive to hand position.

Given all this:
You need to keep in mind the expected and required level of precision and accuracy for various tasks.
Torquing an RF connector is more of a within 20% with decent repeatability is fine type of activity.  You don't need 1%.

Still, it's a bit surprising manufacturers don't just have the 5/16 open end exit the handle at a 90 degree angle to eliminate this effect entirely, as is standard practice with use of torque adapters.

But I suppose it still comes down to being " good enough"

[mark432]

Wow there's a lot of gibberish in this thread.

And thank you for taking the time to add even more gibber jabber.

Gee.  I can't imagine how your interactions with manufacturers didn't go the way you wanted them to.
 

Nice, more gibber added to the pile, compliments of the person complaining about it.

Wanted?  You do realize that there is NOTHING for me to gain by running such a test.  My offering to run an independent check wasn't /isn't something I wanted to do but certainly would.   

Obviously you want something.  Look at the title of the thread.

Boohoo!   Don't buy from this company because I couldn't bully them into doing what I want!

What could you gain?   Oh I don't know. Maybe you were looking for free product.

To the master gage.

Which is what?  Quit being lazy.  You did check it right?  Is it calibrated.  Do you have something else to check it with?


You're stating measurements down in the fractions of mils remember. 

[G0HZU]

My advice to anyone unfamiliar with this stuff is to read the manual for the VNA cal kits that have the Keysight 8710-1765 torque wrench in it
The manual explains what can go wrong if you don't use the torque tool correctly. It tells you that you can't get consistent readings if you hold it differently.

If you experiment with different grip types you can also achieve under-torque even though it feels (to the operator) like the tool is being held nearer to the break point.

The golf club grip should result in a lower break point because the handle can act as a lever that pivots around the thumb. I could easily see a -20% error with this grip type.
Other grip types can introduce more subtle errors but the repeatability will suffer because the forces applied by the fingers and thumb and maybe even the palm of the hand will be distributed differently each time the tool is used.

Hold it and use it correctly and the accuracy and repeatability is much improved.

[joeqsmith]

Wanted?  You do realize that there is NOTHING for me to gain by running such a test.  My offering to run an independent check wasn't /isn't something I wanted to do but certainly would.   

Obviously you want something.  Look at the title of the thread.

Boohoo!   Don't buy from this company because I couldn't bully them into doing what I want!

What could you gain?   Oh I don't know. Maybe you were looking for free product.

ROFLMAO!   I'm not so hard up that I can't afford cables.  I actually work for a living.  Also, my current selection of RF cables are fine (or I would buy what I needed).   Your idea of free products is false.  It's rare I will accept products, but there have been a few exceptions.   When I consider my own investment in these so called free products, I am still in the negative.     

To the master gage.

Which is what?  Quit being lazy.  You did check it right?  Is it calibrated.  Do you have something else to check it with?

You're stating measurements down in the fractions of mils remember.

Similar to suggesting I would have a way to calibrate any other equipment I have.   You would have to be very naive to suggest anyone would have a NIST traceable calibration equipment.   Like the kids suggesting I check my meter meters with one of those cheap "standards", or worse, try to align to them... 

I have no way to check a master gage outside of sending them in for calibration.  Because the masters are not used as punch to drive nails,  I am confident in their original factory specs.   Maybe you do not understand that the dial indicator gages are not a master gage.   In that photo, the master can be seen at the business end of the dial indicator (see attached).

***
A quick search of NIST for  gauge, block, calibration &  certification,  revealed the following paper detailing what is involved.   Welcome to the world of metrology... 

https://www.nist.gov/system/files/documents/calibrations/mono180.pdf

[joeqsmith]

I used both the Maury A027 SMA push-on gages and the (much more accurate) 3.5mm thread-on gages from an HP 85052B calibration kit. Both of my gages are designed to measure pin protrusion/recession.  I do not have the Maury kit to measure dielectric protrusion/recession and (of course) this would not be applicable for 3.5mm connectors.

I've used my 3.5mm gauges to measure some SMA adapters/cables purchased from eBay and the results where concerning.  Quite a few male center pins extended too far out beyond specification.  This can clearly damage a quality 3.5mm connector and maybe even another mating SMA connector.  Until you have an indicator/gauge you're kinda flying blind.

https://www.eevblog.com/forum/rf-microwave/testing-rf-connectors-and-cables/msg2733738/#msg2733738

[joeqsmith]

Sadly, I never did hear back from CentricRF.  Maybe they are closed for the holidays.   I thought for fun I could look at a few lower grade cables I have.   This set includes one from Digi-key, LiteVNA, V2Plus4 VNA, a very low cost one from Amazon, and the ones I used for my 10GHz experiments with the EPR spectrometer.

[joeqsmith]

Remember how I stated that the pins vary a lot.  SMAs are not precision connectors.  So before I do anything else, I'll take some photos of what we are dealing with.  Maybe we can't even measure these accurately.   

This first two photos are the cables I use for my high frequency experiments.  Notice the more common thin outer conductor and the  stepped 
(conventional) male pin.   Notice the OD of the step and how flat the surface appears.  Also note that the dielectric is not interfering with the pin's step in anyway. 

You can also see how the dielectric is recessed from the reference plane and the step appear basically level with the dielectric.  I have no doubt we could accurately measure the reference plane to the step, dielectric and pins tip.

[joeqsmith]

I recently picked up some very low cost cables off Amazon from China.  These are very interesting as these don't use a stepped pin or a thru.  I would place it into the angled category. 

Notice how the pin is machined in a stair step pattern.  While the dielectric doesn't interfere with the pin,  the gage is designed to measure a flat surface.  Who knows which stair step it will catch on and measure to. 

So while I could easily measure the reference plane to the dielectric and pin height, there is no way I could reliably measure where the step is.

[joeqsmith]

Looking at a custom cable from Digi-key.  Note the wider outside conductor's mating surface.  More important to this discussion, there is no step.   Again, I could measure the dielectric and pin positions but we can't measure a step that doesn't exist.

[joeqsmith]

The LiteVNA is rated up to 6GHz and is supplied with cables.  I think we paid about $120 for the first Lites.  These last ones were $150 or so. The cables have not appeared to have changed over the three hardware revisions I have looked at. 

Notice how it appears to use a stepped type pin, but the outside diameter of the step is very small.  The gage's pin also has a thickness.  While the steps OD may be large enough for the gage to sit on it, the dielectric may interfere with the measurement. 

[joeqsmith]

I like the rubberish feel to the cables supplied with the V2Plus4 VNA.  They are very flexible.   They appear to made of brass rather than stainless like the LiteVNA's cables.   

Notice again how the pin appears to have a step for how the dielectric is again interfering with the surface of the step.   

If I were to blindly gage this connector without knowing what the pin looked like,  I may be measuring the raised part of the dielectric.   

Maybe you see what I am driving at.  Of all of these cables, really the only one I would have any confidence in measuring the pin step's position is the first one. 

[joeqsmith]

For a comparison, let's look at an SMA used on a cable produced by Gore.   I'm sure I could easily measure this connector.   

[shabaz]

It feels too much to expect SMA cables to be very precise as you say. I'm no expert in this though, and don't have anything other than some instinct to back that up, from visual inspection of the pieces and how they are put together.

Slightly off-topic, but I got curious to see how center-contact depth is set on SMB connectors.

The only few products I've worked on that used similarly-sized connectors didn't have SMA but rather used right-angle SMB connectors internally, and so, for a while, I thought I'd standardize on SMB for my home lab too, but SMA simply had more general traction, so I aborted that idea pretty quickly.

I don't think I've ever assembled a straight SMB connector (I bought ready-made SMB cables for my home lab at that time, and maybe assembled a few right-angle ones where the center pin is pre-fitted). Anyway I took a look at the assembly instructions for some straight Amphenol SMB connectors, and it's so much more simplified than straight SMA assembly. No need for dielectric insertion tools, nor pin depth setting tools. Still not a precision solution, but for that sort of sized board-to-board connections inside a product, it is attractive.

[joeqsmith]

It feels too much to expect SMA cables to be very precise as you say. I'm no expert in this though, and don't have anything other than some instinct to back that up, from visual inspection of the pieces and how they are put together.

I did provide that Maury paper that you could read where they discuss the SMA and the 3.5mm precision connectors.  I'm sure Google would find a lots of hits as well. 

I took a photo for you of the cables I put together for my HP8754A VNA. 
https://www.eevblog.com/forum/testgear/dont-buy-sma-cable-assemblies-from-centricrf/msg5746121/#msg5746121

Attached showing a closeup of the male pin.  Again, notice the step's wide OD and no interference with the dielectric.  An easy one to measure.   

[Smoky]

To add to the collection, here's a picture of the SMA connector on Amphenols ATC Phase Stable series of cables:



Part number 095-902-466-001

Looking closer, notice how loose the center contact is within the dielectric:



Amphenol claims that these SMA connectors are rated for 5000+ mating cycles.

[joeqsmith]

To add to the collection, here's a picture of the SMA connector on Amphenols ATC Phase Stable series of cables:

Part number 095-902-466-001

Looking closer, notice how loose the center contact is within the dielectric:

Amphenol claims that these SMA connectors are rated for 5000+ mating cycles.

It looks like the step protrudes beyond the dielectric a fair amount, and the dielectric is recessed from the reference plane.   Do you have a  way to measure it?

I bought some used cables for my HP3589A that appear to be custom made using Suhner connectors.  These get a fair amount of use and it looks like I purchased them in 2011.  What is interesting about the SMA is that double step feature.    I wonder again if the gage would touch off on that first step or the second.  Imagine if it measures the lower step and the upper one protrudes beyond the reference.   

I think it is safe to say if you want to measure them, it is not as simple as buying a gage, zeroing it to a standard, cleaning all the parts, press on the gage and bingo... Personally, I always start with a visual inspection.  I'm surprised the big players don't have any training videos that cover the SMA in particular because there are so many variations.   

[joeqsmith]

As that paper discusses, one way to avoid the variations with SMAs is to use a different connector.  Cost goes up and you are presented with new problems, like how to measure your $2 ebay Tees from China without risking your precision connectors. 

https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg5665155/#msg5665155

[G0HZU]

For fairly serious VNA use, I think that it's best to simply abandon low and mid range cables and connectors.

For VNA and (e)cal kit and test fixture stuff, I use 18GHz rated Gore and Suhner cables and I use decent 18 GHz rated SMA end launch connectors.
There shouldn't be any need to measure pin depths with any of these because they are high quality items.

I do sometimes use MiniCircuits 18 GHz rated cables like the CBL-3FT-SMSM+ cables. These are quite well made with good shielding but I tend to use these for 4 port measurements below about 1GHz.

For less serious stuff, it's OK to use lower budget (but still decent) cables but I wouldn't recommend connecting them to any precision connectors that also get used with precision cables and connectors because of the risk of damage to the precision connectors. I don't see the measurement of pin depth as a means to reduce the risk as there are various ways that cheaper cables can damage cal kits or other precision connectors.

For casual stuff with a budget VNA, I use a dedicated homebrew cal kit with my nanovnaH for example and I use dedicated cables with it.
To minimise risk, I either use a dedicated connector saver at the end of the lower cost cables or something like a 3dB attenuator from Minicircuits. This isolates the homebrew cal kit and test fixture from the lower cost cables that I use with the nanovnaH.

There's so much to go wrong with the various dimensions and machined surfaces of budget SMA connectors that it won't be cost effective to measure them thoroughly.