SMA connectors

[martafafara]

While reviewing Kasli SoC, I've noticed that the Amphenol SMA connector is used. On the other hand, in most newly designed project Delock 65848 occurs. Some discussion evolved in the pull request. In this particular project, author decided to keep Amphenol. The aim of this issue is to decide which solution should be applied:

1. Use Delock 65848 in any project
2. Creating the additional release with only changed this one part i.e 1.2 with Amphenol and 1.2.1 with Delock.
3. Including additional variants in project with SMA connector alternative part

[gkasprow]

use Delock everywhere

[dnadlinger]

Aren't the Delock ones the ones where the barrel easily comes off? We have had several Stabilizers here break in this way recently, even though users are educated on use of an appropriate torque wrench, and we have never seen this on any other of the (thousands of?) connections.

In the linked ticket, it was mentioned that an issue might be that common SMA torque wrenches are set to a higher torque than these connectors can take. If that's the case, the correct solution surely is to switch to connectors that are compatible with the assumptions of educated expert users – insisting that the user is technically at fault for doing something reasonable will only reflect badly on the quality of our hardware.

[gkasprow]

I compared both connector types and didn't see substantial difference. I broke a few pieces of both.

[gkasprow]

Were those Stabilizers equipped with plastic panel spacers?

[jordens]

We had done the same test and agree. I have seen no data that indicates the Amphenol is more robust. You'd need steel connectors to survive our expert users using 1 Nm wrenches. By definition they are not educated, otherwise they would known the right torque. Please first exclude issues with damage or stress due to the front panel/washers.

[gravelpitman]

In my opinion, the connector itself is not the issue. The Delock connector performs well and withstands repeated tightening with a torque of 0.7 Nm, which exceeds the typical range of 0.3–0.6 Nm for brass connectors.

Maybe we should simply add the recommended mating torque information to the front panels.
The user would be immediately informed about the recommended tightening torque. This way, it's clear that exceeding this torque puts the connector at risk of damage due to user error.

Sources of information regarding the recommended SMA connector tightening torque are listed below.
https://docs.keysight.com/kkbopen/is-an-sma-torque-wrench-equivalent-to-a-3-5-mm-torque-wrench-589746928.html
https://www.amphenolrf.com/rf-connectors/sma-connectors.html
https://www.taoglas.com/assets/application-notes/[SMA_Connector_Torque_Guide_(APN-25-8-001).pdf]
https://www.microwaves101.com/encyclopedias/connector-torque
https://coaxicom.com/specifications/sma-specifications/
https://www.data-alliance.net/blog/torque-ratings-of-sma-and-rpsma-antenna-cable-connectors-adapters/

[dnadlinger]

By definition they are not educated, otherwise they would known the right torque.

Many decent torque wrenches sold specifically as "SMA Torque Wrench", such as the Fairview Microwave ST-SMA-516-BO8A or Huber+Suhner 74_Z-0-0-21, have a > 8 in-lbs (0.9 Nm) nominal torque.

Other test and measurement equipment manufacturers, such as National Instruments, also recommend 5-9 in-lbs for "SMA (brass & stainless)" specifically (as opposed to their specs for 3.5 mm, K, etc.). This is just the first example I found; there are probably more, but the point here is just that our target users might reasonably use the "wrong" tools.

Sure, when dealing with a metrology-grade interface on an instrument where keeping the calibration intact is paramount, I'd fault the user if they didn't make sure to handle the connectors in exactly the right way and that tools sold as compatible actually meet the required specs.

But for a random device, connectors repeatably breaking is just a bad look. As I mentioned, we have at least several hundreds (probably thousands) of SMA connectors across the labs here from different manufacturers, and the only issue with connector breakage I can remember – apart from outright misuse, e.g. trying to jam in a cable with a broken centre pin – across ~10 years was recently with a Stabilizer v1.3.3 and a Pounder v1.2. I'm away from Oxford at the moment, so I can't check whether all torque wrenches that might have been used across their life meet the lower brass spec, but either way, it was a noticable pattern.

How much more expensive are connectors that can properly withstand amounts of torque that are reasonable expected – if outside the recommended range for brass – in this application (e.g. 1 Nm)? If this is only a small amount (say some 10€ per module), I'd argue the obvious choice is to go for something that doesn't lead to a bad user impression.

The Delock connector performs well and withstands repeated tightening with a torque of 0.7 Nm, which exceeds the typical range of 0.3–0.6 Nm for brass connectors.

Is the 0.7 Nm you mentioned the limit above which they start to break, or the recommended value? Presumably, many connectors with a specified 0.3–0.6 Nm range still have some safety margin until the point where the barrel comes loose.

Please first exclude issues with damage or stress due to the front panel/washers.

The broken connectors had loose insulation inserts around the connector barrels that looked 3D-printed. It didn't seem like they would cause any undue strain.

[dnadlinger]

Note also that the https://docs.keysight.com/kkbopen/is-an-sma-torque-wrench-equivalent-to-a-3-5-mm-torque-wrench-589746928.html article linked above states: "To connect a 3.5 mm male connector to a SMA female connector, use a 3.5 mm torque wrench." Following this advice would lead to a 0.9 Nm wrench being used on the Sinara modules – of course, there would be little reason to use a 3.5 mm on any of our modules, but it underscores my point about connectors outright breaking with < 1 Nm torque being unexpected.

[dnadlinger]

In summary, I don't have a strong opinion on which particular connector model to use on the Sinara modules, and from my current state of knowledge, certainly can't blame the Delock connectors specifically over other alternatives. What I wanted to make sure isn't missed is just that (a) we've seen several instances of broken connectors on recent Stabilizer/Pounder modules that we didn't before across many years/vendors, and (b) the exact torque specs for SMA are sufficently unclear across manufacturers that users might reasonably end up using the "wrong" torque even if they are aware that SMA connections require attention in that regard.

(a) and (b) are not necessarily related, but even if they turn out to be, this seems like a problem best by going (back to?) less marginal connectors/assembly techniques/… rather than blaming our users.

[dnadlinger]

(A small front panel marking with a caution sign saying "0.6 Nm max." or something like that could work if there really is no economic alternative to easily-breaking connectors.)

[gravelpitman]

I've tested additional SMA connectors and created a spreadsheet listing their advantages and disadvantages from my perspective.
Additional note: I’ve established internal TS guidelines stating that the SMA connector should protrude 8.5 mm from the front panel. This helps resolve several issues (e.g., SMA_IDC_Adapter#2 and Kasli#74). That’s why I refer to some connectors as being too short, etc.

SMA type	Advantages	Disadvantages
Delock 65848	-Already in use -No design changes required -Cost-effective -Long threaded section solves the previous issue of insufficient thread engagement with SMA cables -Compact footprint (6 × 6 mm), suitable for space-constrained modules like DIOSMA	-Low safety margin for overtightening (only ~20% above the recommended torque)
Amphenol 132203	-Very high safety margin (I couldn’t break it, even when applying 2 Nm+ of torque) -Already used in some modules	-Shorter threaded section than in the Delock version (2.5 mm less thread, without changes to the PCB layout) -Questionable compatibility with current front panels (connector axis sits 6.35 mm above the PCB surface, compared to 6 mm in the Delock) -Relatively expensive
Wurth 60311042114502	-Very high safety margin (I couldn’t break it, even when applying 2 Nm+ of torque) -Average price	-Shorter threaded section than in the Delock version (2.5 mm less thread, without changes to the PCB layout) -Incompatible with current front panels (the connector axis sits 7.06 mm above the PCB surface, compared to 6 mm in the Delock version)
Adam Tech RF2-49B-T-00-50-G-HDW	-Cost-effective -High safety margin (withstood approximately 1.4 Nm in my tests)	-Shorter threaded section than in the Delock version (2.5 mm less thread, without changes to the PCB layout) -Questionable compatibility with current front panels (connector axis sits 6.35 mm above the PCB surface, compared to 6 mm in the Delock)