On 10/1/2020 9:55 PM, Jasen Betts wrote:> On 2020-10-01, Don Y <blockedofcourse@foo.invalid> wrote: >> I'm currently using punchdown blocks to terminate network cables >> on my devices. But, they are considerably larger (volumetric) >> than I'd like. Coupled with the bend radius of the cables, >> it requires a larger enclosure volume than I'd like (I'm aiming >> for really small devices -- on the order of a couple of cubic >> inches). >> >> I'd ruled out the more traditional 8P8C's as they're even more >> bulky (and require additional labor to install). >> >> Are there other connector/termination options that might be >> leaner? > > Do you have any bandwidth requirements?I need to run 100Mb/Gbe over the link. But, I don't need to be able to "unmate" the connection (which tends to be a requirement for "typical" network connections).>> I'm currently looking at designing a custom connector along the lines >> of the IDCs used for ribbon cable. But, I'd like to avoid those tooling >> charges and reliance on a single supplier. > > what's wrong with the availalble IDC terminations?Aside from 8P8C plugs, I don't see any IDC connectors/terminations that are suited for the "4 discrete pairs" that you encounter in a CAT5/6 cable. I imagine you can repurpose a connector for that sort of application but can't imagine it being a reliable undertaking (installer errors, etc.) given that these aren't really just "8 generic conductors" What I'd like to see is 8 closely spaced holes/tubes/channels into which the individual conductors could be inserted. Then, a "cinching" mechanism to pierce their insulation and physically secure them to the board. (there's very little mechanical force exerted on the cable/connection after mating; mate-and-forget) The imperative being minimizing the required volume of an encapsulating enclosure (without having to place special constraints on that enclosure). So, the "connector" has to consume minimal volume and offer flexibility in how the cable exits wrt the PCB itself.
CAT5e/6 connector/termination alternatives
Started by ●October 1, 2020
Reply by ●October 2, 20202020-10-02
Reply by ●October 3, 20202020-10-03
On 10/2/20 8:49 PM, Don Y wrote:> What I'd like to see is 8 closely spaced holes/tubes/channels into > which the individual conductors could be inserted. Then, a "cinching" > mechanism to pierce their insulation and physically secure them to > the board.What about the screw terminals that are common in alarm systems and PLCs. I see them in all shapes, sizes, and number of connectors. http://www.cnkeco.com/keco-terminal-block/71-screw-terminals.html -- Grant. . . . unix || die
Reply by ●October 3, 20202020-10-03
Grant Taylor <gtaylor@tnetconsulting.net> wrote in news:rl8smi$hqm$1@tncsrv09.home.tnetconsulting.net:> On 10/2/20 8:49 PM, Don Y wrote: >> What I'd like to see is 8 closely spaced holes/tubes/channels >> into which the individual conductors could be inserted. Then, a >> "cinching" mechanism to pierce their insulation and physically >> secure them to the board. > > What about the screw terminals that are common in alarm systems > and PLCs. > > I see them in all shapes, sizes, and number of connectors. >snipped link You really do not know how modern high speed comms work then, Grant. Look at a true CAT6 connector. It has: A: A shield/metal shroud that goes right up to the socket insertion section. That shield is because the twisted pairs get 'opened up', as in 'untwisted' for their termination into the connector. No shield == not CAT6. exposed untwisted pairs... not CAT6. Even that short little segment less than a half inch long matters. So yeah, if you want to dumb down your comm link to 15 years ago's speeds then splay those motherfuckers out uncaringly. Stick to 'net consulting'. Your hardware acumen is lacking. Modern PLC wiring is fully shielded. At least when one is talking about automated assembly machines containing hundreds of connections and pneumatic actuators and machine vision cameras, etc. Especially automotive assembly machines where a failure results in high dollar law suits by victims of said failures.
Reply by ●October 3, 20202020-10-03
On 10/2/20 10:30 PM, DecadentLinuxUserNumeroUno@decadence.org wrote:> You really do not know how modern high speed comms work then, Grant. > > Look at a true CAT6 connector. It has:Per the OP's subject, CAT 5e is sufficient. CAT 6 is not required. Depending on the distance that the OP wants to run cables, CAT 5e is perfectly fine for the desired 1 Gbps speed.> A: A shield ...Is not required for 1 Gbps speed.> So yeah, if you want to dumb down your comm link to 15 years ago's > speeds then splay those motherfuckers out uncaringly.The OP said that they did not want to fall back to 100 Mbps, which is not required. CAT 5e will quite happily handle 1 Gbps.> Stick to 'net consulting'. Your hardware acumen is lacking.Part of my net consulting is figuring out what clients actually need to accomplish their goal. Part of that process is identifying potential solutions. Potential solutions include things that don't actually provide the answer but do provide information as to why they don't provide the answer. Thereby providing more information to make it easier to hone in on the solution. There's no need for a personal attack. If you don't like the suggestion, ignore it and move on. I've got multiple colleagues running full Gbps traffic across connectors that fall into the family of what I suggested. If the OP wanted 10 Gbps, that would be a different discussion. It might even be different for 2.5 Gbps. But 1 Gbps is still fairly easy to get and has a lot of options. -- Grant. . . . unix || die
Reply by ●October 3, 20202020-10-03
Grant Taylor <gtaylor@tnetconsulting.net> wrote in news:rl90u4$sg0$1 @tncsrv09.home.tnetconsulting.net:>> A: A shield ... > > Is not required for 1 Gbps speed. >Nice snip. I did not attack you. I said you lack hardware acumen. That is a fact. Your snip, however, while you rambled further, IS tantamount to an attack. Again... A TRUE CAT6 connector IS shielded, turkey turd. Not the entire cable, it is not about the entire cable, JUST THE TERMINATION. The cable is twisted pair and the termination IS important And crosstalk and EMISSION occours without it and handshakes result in SLOWER than 1Gb/s speeds. Just like when you owned a "56k" modem but the damn thing never hooks up faster than 14,400. It FALLS BACK to slower speeds. So, IF too much untwisted pair segments are exposed, said fallback DOES occur. Reading what HE wrote, he DOES know that and wrote about closely spaced PCB mounting.
Reply by ●October 3, 20202020-10-03
On Fri, 2 Oct 2020 17:14:30 -0700, Don Y <blockedofcourse@foo.invalid> wrote:>On 10/2/2020 12:07 AM, Jeff Liebermann wrote: >> On Thu, 1 Oct 2020 22:03:56 -0700, Don Y <blockedofcourse@foo.invalid> >> wrote: >> >>> I don't need more than one cable termination. >> >> How about an M8 or M12 ethernet connector? >> <https://www.google.com/search?q=m8+ethernet+connector&tbm=isch> >> <https://www.google.com/search?q=m12+ethernet+connector&tbm=isch> > >Even worse. The "connector PAIR" has to be small.The smallest connector I know of are "closed bottom pin sockets". <https://www.te.com/usa-en/product-50863-4.html> <https://www.te.com/usa-en/product-5050863-6.html> Line 8 of these in a row on the surface of a PCB. Sweat solder the sides of these pins to pads on the component side of the PCB. To make it look good, use plastic separators or spacers, possibly 3D printed. Or, just use a fixture that looks like an 8 pin "fork" to align and hold the pins until they can be soldered to the PCB. Conjure some kind of cable clamp to hold the CAT5 cable to the PCB, or just hot melt glue it in place. Solid conductor CAT5 is usually 24 AWG. The other CAT's can be different gauges. <https://www.cablexpress.com/blog/cat6-cabling-what-is-the-big-deal-about-awg-american-wire-gauge/> Stranded conductor CAT5 will need the wire ends tinned and a larger pin sockets to compensate for the added solder. 6.6 mm might be a little too short to get a decent grip on the CAT5 wires. Maybe something longer if you spare the PCB real estate. Drivel: No new fires in the area so far. It's been very hot for the last 3 days and probably will continue to be hot for the next two or three days. I'm holding my breath. -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
Reply by ●October 3, 20202020-10-03
On 10/2/2020 8:54 PM, Grant Taylor wrote:> On 10/2/20 8:49 PM, Don Y wrote: >> What I'd like to see is 8 closely spaced holes/tubes/channels into >> which the individual conductors could be inserted. Then, a "cinching" >> mechanism to pierce their insulation and physically secure them to >> the board. > > What about the screw terminals that are common in alarm systems and PLCs.I had considered the "spring loaded" sort of connections that you'd encounter (in a grossly overlarge implementation) for speaker connections. E.g., the Nest thermostat's field connections are of this form -- though as INDIVIDUAL connections (I'd prefer "align set of conductors, depress spring release, insert conductors, release spring"). But, I'm not sure this would be a reliable connection. One of the issues is that the connection is unlikely to be made "at a bench" but, rather, atop a ladder, wandering around a mezzanine, in the bowels of a piece of industrial equipment, etc. So, you want it to be as simple and fool-proof as possible: insert wires (in some particular order), then "secure" (crimp?) (This is why I'd suggested individual tubes/channels for each conductor so the installer only has to get each conductor aligned with appropriate channel and then "drive it home" -- while moving on to the next conductor.)> I see them in all shapes, sizes, and number of connectors. > > http://www.cnkeco.com/keco-terminal-block/71-screw-terminals.html > > >
Reply by ●October 3, 20202020-10-03
Jeff Liebermann wrote:> > The smallest connector I know of are "closed bottom pin sockets". > <https://www.te.com/usa-en/product-50863-4.html> > <https://www.te.com/usa-en/product-5050863-6.html> > Line 8 of these in a row on the surface of a PCB. Sweat solder the > sides of these pins to pads on the component side of the PCB. To make > it look good, use plastic separators or spacers, possibly 3D printed. > Or, just use a fixture that looks like an 8 pin "fork" to align and > hold the pins until they can be soldered to the PCB. Conjure some > kind of cable clamp to hold the CAT5 cable to the PCB, or just hot > melt glue it in place.It might be useful to test whatever method being considered by retrofitting it onto the board of an 8-port gigabit switch. For $30 you could test 8 methods for data errors.
Reply by ●October 3, 20202020-10-03
On Sat, 3 Oct 2020 15:50:40 -0400, "Tom Del Rosso" <fizzbintuesday@that-google-mail-domain.com> wrote:>Jeff Liebermann wrote: >> >> The smallest connector I know of are "closed bottom pin sockets". >> <https://www.te.com/usa-en/product-50863-4.html> >> <https://www.te.com/usa-en/product-5050863-6.html> >> Line 8 of these in a row on the surface of a PCB. Sweat solder the >> sides of these pins to pads on the component side of the PCB. To make >> it look good, use plastic separators or spacers, possibly 3D printed. >> Or, just use a fixture that looks like an 8 pin "fork" to align and >> hold the pins until they can be soldered to the PCB. Conjure some >> kind of cable clamp to hold the CAT5 cable to the PCB, or just hot >> melt glue it in place.>It might be useful to test whatever method being considered by >retrofitting it onto the board of an 8-port gigabit switch. For $30 you >could test 8 methods for data errors.Good idea. However, it should probably be done using a managed 8 port gigabit switch using SNMP to accumulate error counts by type of error such as some of these: <http://net-snmp.sourceforge.net/docs/mibs/EtherLike-MIB.txt> dot3StatsIndex InterfaceIndex, dot3StatsAlignmentErrors Counter32, dot3StatsFCSErrors Counter32, dot3StatsSingleCollisionFrames Counter32, dot3StatsMultipleCollisionFrames Counter32, dot3StatsSQETestErrors Counter32, dot3StatsDeferredTransmissions Counter32, dot3StatsLateCollisions Counter32, dot3StatsExcessiveCollisions Counter32, dot3StatsInternalMacTransmitErrors Counter32, dot3StatsCarrierSenseErrors Counter32, dot3StatsFrameTooLongs Counter32, dot3StatsInternalMacReceiveErrors Counter32, dot3StatsEtherChipSet OBJECT IDENTIFIER, dot3StatsSymbolErrors Counter32, dot3StatsDuplexStatus INTEGER, dot3StatsRateControlAbility TruthValue, dot3StatsRateControlStatus INTEGER etc... Managed gigabit switches cost somewhat more than unmanaged gigabit switches: <https://www.amazon.com/NETGEAR-PROSAFE-8-PORT-GIGABIT-SWITCH/dp/B003GOFIC2> Unfortunately, managed switches usually don't test for important cable parameters such as low signal levels, reflections, NEXT (near end crosstalk), and excessive attenuation. Therefore, a gigabit network cable tester should also be used along with a TDR (time domain reflectometer) to look for impedance discontinuities. Some cable testers and certifiers have a built in TDR, but external is usually good enough: <https://www.ebay.com/itm/TDR-Time-Domain-Reflectometer-Fast-Clock-USB-Power-Detect-cable-faults-More/153549738452> -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
Reply by ●October 4, 20202020-10-04
On 10/2/2020 10:34 PM, Jeff Liebermann wrote:> On Fri, 2 Oct 2020 17:14:30 -0700, Don Y <blockedofcourse@foo.invalid> > wrote: > >> On 10/2/2020 12:07 AM, Jeff Liebermann wrote: >>> On Thu, 1 Oct 2020 22:03:56 -0700, Don Y <blockedofcourse@foo.invalid> >>> wrote: >>> >>>> I don't need more than one cable termination. >>> >>> How about an M8 or M12 ethernet connector? >>> <https://www.google.com/search?q=m8+ethernet+connector&tbm=isch> >>> <https://www.google.com/search?q=m12+ethernet+connector&tbm=isch> >> >> Even worse. The "connector PAIR" has to be small. > > The smallest connector I know of are "closed bottom pin sockets". > <https://www.te.com/usa-en/product-50863-4.html> > <https://www.te.com/usa-en/product-5050863-6.html>These look to just be a plane hollow cylinder -- no "spring tines" (like you would find on a machined pin IC socket) to cling to the "mating" conductor.> Line 8 of these in a row on the surface of a PCB. Sweat solder the > sides of these pins to pads on the component side of the PCB. To make > it look good, use plastic separators or spacers, possibly 3D printed. > Or, just use a fixture that looks like an 8 pin "fork" to align and > hold the pins until they can be soldered to the PCB. Conjure some > kind of cable clamp to hold the CAT5 cable to the PCB, or just hot > melt glue it in place. > > Solid conductor CAT5 is usually 24 AWG. The other CAT's can be > different gauges. > <https://www.cablexpress.com/blog/cat6-cabling-what-is-the-big-deal-about-awg-american-wire-gauge/> > > Stranded conductor CAT5 will need the wire ends tinned and a larger > pin sockets to compensate for the added solder. > > 6.6 mm might be a little too short to get a decent grip on the CAT5 > wires. Maybe something longer if you spare the PCB real estate.From my (uneducated) observations, it appears that the way connectors grab wires has a fair bit of thought involved. I'd rather trust to an ME (or Materials Scientist?) to sort that out, properly, than to hope an /ad hoc/ approach will work scaled to production quantities. Perhaps if tapered pins were affixed to the ends of the wires to ensure a more reliable fit...?> Drivel: No new fires in the area so far. It's been very hot for the > last 3 days and probably will continue to be hot for the next two or > three days. I'm holding my breath.God: "Noah?" Noah: "Yes, Lord?" God: "How long can you tread water?"
