I used to do +100 Amp connections. Crimps were never good enough. Soldering was always better. Solder is not a good mechanical connection. We always crimped first for mechanical then soldered for electrical. Dwight

On Sat, 13 Sep 2014, dwight wrote: Done correctly, using the right compression dies and tools, crimped connections are always far more secure and reliable both mechanically and electrically than soldered connections. However, the correct tools (not one of those cheap multifunction plier type strip/crimp/bolt cutter tools, or worse, a pair of vice grips or needle nose pliers) are expensive and more often than not, outside of a large company and/or factory, people don't have access to proper compression tools. In the automotive industry for example, at one time it was common for even "factory authorized" dealer operated repair shops to crimp and then solder replacement terminals for vehicle wiring harnesses. That practice has since changed with better access to the correct tools and with better training, and technicians are now told specifically /not/ to solder crimp type terminals.

This topic come up often on the Amateur Radio lists I am a member of, especially as the UK Intermediate Licence includes fitting a connector to coax cable, but doesn't specify if it should be soldered or crimped, and many folks feel crimping is in some way cheating. Whilst I prefer solder, many folks say crimp is best. The problem with solder is that it creates a stress point where the solder "ends" so if the cable moves the cable can fracture. The heat can also alter the temper of the cable and make it brittle. (Unlike steel some copper alloys harden with slow cooling). This probably isn't an issue with most classic computer projects.... A properly crimped connector forms a cold weld and so is no less conductive that a soldered joint, should be mechanically sound, and flexes better than a soldered joint, but the tools can be expensive.... .. and I guess we have all had poorly made UTP/RJ45 (I know there is no such standard as RJ45) that has stopped working because the plug is no longer connected to the cable... Dave G4UGM -----Original Message----- From: cctalk-bounces at classiccmp.org [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Tony Duell Sent: 14 September 2014 22:06 To: cctalk at classiccmp.org Subject: Re: Problematic spade lug in power supply I might question the 'far' here, but I have heard that crimp connections are very relaible _if done properly_. The last is the importnat point. I feel that a well done solder or crimp conneciton will beat a poorly done one of the others. And a well done crimp conenction means the right size crimp conencotrr, the right crimping dies, the correct tool to hold them, and so on. It does not mean the average DIY shiop crimping tool. And I am darn suroe of one thing. Even if a soldered connection is less reliable on paper than a crimped connection, a well-soldered connection in your classic computer is not going to be a major source of unreliability. -tony

On Sun, 14 Sep 2014, Chuck Guzis wrote: You want the terminal less malleable. Friction fit terminals and contacts have a certain "springiness" to them and that is how they achieve a low resistance connection. Most terminals are made of plated brass or copper alloy. That reminds me of the time I made a bunch of 50 ohm terminators. I didn't have any terminators on hand at the time, but I did have a bunch of screw-on BNC connectors, fillister head screws with the same thread as the BNC connectors, and a package of 51 ohm 1% resistors. I cut the screws to length, drilled a hole, and soldered one end of the resistor into the screw. Then I simply screwed the screw+resistor into the BNC connector. They were still working fine when I also replaced the last of my thinnet runs with UTP. They'd probably still work fine today ;)

On Sun, 14 Sep 2014, Dave G4UGM wrote: To expand on your points some, flux and solder also tend to wick up into the conductor underneath the insulation. With most RMA fluxes, corrosion isn't going to be an issue, but with all these "modern" water-clean and citrus based fluxes, this could be a major problem. Solder wicking up into the wire is likely to be more of a problem as it creates a point where fatigue failure can occur out beyond the connector housing or insulation crimp. In the case of friction fit terminals and contacts, the heat of soldering can also affect the temper of the terminal. This is especially a problem with the really tiny discrete contacts used in modern automotive wiring, and another reason the industry has done a complete 180 on the former practice of crimping+soldering for harness repair as connections have become smaller and more numerous. With coax specifically, soldering also can melt and distort the dielectric. With lower frequency RF (RG58 with a BNC or RG8 with a PL259 for example), it isn't going to cause much of an issue, but with higher frequency stuff, it absolutely can cause problems. Soldering coax connectors without damaging the coax is a bit of an art, and really is something a radio operator should learn and be able to do well, even if they mainly use compression connectors. The bigger issue with coax connectors is sealing them from the elements, and despite what some people may say about the compression variety, both soldered and compression types need supplemental protection from water ingress when used outdoors. WRT to modular connectors (8P8C "RJ45"), one major problem with those are plugs intended for use with stranded wire being crimped onto solid wire. If you look at the plugs sold by a typical computer store, they are more often than not intended for use with stranded wire for patch cables. The bulk of the UTP (cat5/cat5e) network cabling sold in these stores is solid conductor, which is intended to be installed in walls and used with IDC connections. The results of this are easy to predict when you think about how many people who buy supplies from these stores are going to crimp plugs onto bulk UTP cable instead of properly terminating the cable with jacks and using patch cables at each end. The conductivity of a soldered-only connection is also still less than that of a properly crimped connection. This has been well tested in the industry and is just basic physics. The conductivity of tin/lead solder is simply higher than that of the wire and terminal alone, so adding a layer of solder between them will result in a higher resistance connection than if the connection were crimped. With gold plated contacts, you have yet another potential issue, and that is gold embrittlement of the soldered connection. Many smaller discrete contacts are fully plated, and if you solder those, the gold in the crimp area will dissolve into the solder. Depending on the thickness of the plating, there may or may not be enough dissolved gold in the solder to be a problem, but this is still a potential gotcha that many people never consider. The real question people should ask though, is how are these type of connections done in aerospace and full-sized aviation? In those industries, terminals and contacts must be crimped and not soldered. At one time I used to solder or crimp+solder terminals too. I've not done this in decades though.

I feel like a bit of a fraud now. I have a 'full' UK licence but have never shown I can fit a connector to a bit of cable (needless to say I have done that many, nmany, times, just never had to prove it). Incidnetlaly, the trick with fitting BNC and TNC solder plugs is to ignore the ridiculous instructios often given, of measuring so much sinsualtion to strip back, then strip so much off the inner conductor, etc. The conenctors are designed to be easy to fit, every cut/strip is to the end of a part you've fitted. I beelive the solder forms an alloy wit hthe copper (it virtually has to for it to 'wet' at all, and this alloy is somewhat brittle. However well-designed connectors deliberatly prevent the wire flexing at this point. Agreed My feeling, as I said last night is that a well-made solder connection or a well-made crimp connection will be reliable. Even if the latter is more reliable on paper, the former is 'good enough' for any classic computing application THere is an RJ45 witing scheme, and it uses 8p8c connectors. It has nothign to do with ethernet, though. -tony