Replacement RJ45 on CTB16PC V1 & V2

[lightingman117]

TL;DR

Is there a better RJ45 replacement than the TE unit?

---

I have a number of CTB16PC V1s & V2s  with failing RJ45's

I found in the forums the replacement RJ45 (Mouser PN: 571-5556416-1)

Is this really the best solution?

I'm slowly going through the mouser catalog trying to find something that is a bit better than 'Board Lock' mounting style.

I know I could send them back to LOR, but I can solder... I'm more interested in a connector that's a little bit more full-proof.

I did order some of the RJ45 waterproof extensions so that after replacement the wear location will be the coupler located on housing bulkhead.

---

For temporary solution I've been using T-taps (3x female RJ45's and a short jumper to one of the functioning RJ45's.

I know LOR frowns on splitting the RS485 signal, but I've gotten away with this multiple times and with splits ranging from 20-100+ft in each direction.

All at the standard 57k tho.

 

 

 

Edited December 8, 2022 by lightingman117
Added V2 controller having issue

[k6ccc]

What is your issue with that connector?  Would you rather have the only mechanical support be the soldered connection?  I'm not looking at a CTB16 right now, but I'm going to assume that if that is the recommended part number, I would assume that the board was designed around that mounting....

19 minutes ago, lightingman117 said:

I know LOR frowns on splitting the RS485 signal,

It's not LOR so much, it violates the RS-485 standard.  However, especially at low speed, you can get away with violating the standard pretty well.  Let's face it, prior to some of the newer controllers having on-board terminations, how many people properly terminated their networks?  Damn near none...

 

[TheDucks]

Knowing when and how to fudge the rules is the sign of knowledge.

A 'T' close wired should work fine an slower speeds.

I learned RS485 rules when working with Security Systems that would refuse to find/use devices if the networks were not terminated properly. I terminate... Period

Old tech addage: If it does not work, follow the (design) rulebook 😛

[lightingman117]

I mean, what I really want is a connector soldered to the board rather than friction fit.

I'd settle for a connector with longer 'feet' that protrude through the back of the board.
I need to get my caliper and keep making my excel list of possible connectors.

Good news is most connectors have the same footprint (pitch).
Bad news is most of the feet are shorter :(

 

Maybe some glue to hold the feet in.

 

Failure type: Pins no longer forward; requires connector to lean forward.
There are others. But this is the primary failure I see/have.
https://www.dropbox.com/s/ze4n9hkg0c58d1x/1208221028_HDR.jpg?dl=0

Feet barely peak out, the wedge shape definitely doesn't make it to the far side of the board causing flexing and eventual permanent displacement.
https://www.dropbox.com/s/relg73svc1qgr2x/1208221027a.jpg?dl=0

Underside when in 'working' mode.
https://www.dropbox.com/s/9mgoqohbh9uyo3t/1208221027.jpg?dl=0

 

I've violated every standard possible 
I've never fixed an issue via termination for RS485/DMX.
It's always been a cable, connection, or chip failure.

[TheDucks]

26 minutes ago, lightingman117 said:

 I've violated every standard possible 

 

If you got away with that, more lights to you 👍

The object is a setup that works reliably. No glitches. No foul. 😛

[k6ccc]

24 minutes ago, lightingman117 said:

I mean, what I really want is a connector soldered to the board rather than friction fit.

Well, we could double the cost of the connectors and go to metal housings rather than plastic housings...

Looking at your photos, you must really be abusing the cables / connectors.  That's not damage just from plugging in and unplugging.  Are you by chance using Cat-6 cable?  It is highly recommended NOT to use Cat-6 because it is so stiff that it strains the connectors when closing the door on controllers that have the cable leave perpendicular to the board.

1 hour ago, lightingman117 said:

I did order some of the RJ45 waterproof extensions so that after replacement the wear location will be the coupler located on housing bulkhead.

Good plan.  Make sure to zip tie (or equivalent) the pigtail so that it can't pull on the connector.

30 minutes ago, lightingman117 said:

I've violated every standard possible 
I've never fixed an issue via termination for RS485/DMX.
It's always been a cable, connection, or chip failure.

Most of us have violated the standards - and gotten away with it.  The RS-485 standard allows 32 devices on the network and a total cable length of 4,000 feet.  Most of us are no where near either one of those limits - which makes it easier to violate the standard and get away with it.  This is particularly true when operating at a low data speed.  What you can get away with at 56K is very different than what you can get away with at 1000K speed.  If you have an O-Scope, hang that across your network somewhere in the middle and compare the square waves with and without termination.  Unless you are running a really short network, it may surprise just how crappy the signal gets.  The chips do a pretty damn good job of digging the signal out of the crap.

 

[lightingman117]

7 hours ago, k6ccc said:

Well, we could double the cost of the connectors and go to metal housings rather than plastic housings...

Not seeing a huge price difference. But yeah; cost is assuredly a factor.
FWIW I don't care about the metal housing or shielding; just a better physical interface between connector & board.

7 hours ago, k6ccc said:

Looking at your photos, you must really be abusing the cables / connectors.  That's not damage just from plugging in and unplugging.  Are you by chance using Cat-6 cable?  It is highly recommended NOT to use Cat-6 because it is so stiff that it strains the connectors when closing the door on controllers that have the cable leave perpendicular to the board.

Slight mistake on the root cause analysis.
The jacks were straight; I had to bend them because the metal contact wires got to the point that they don't make contact with the 8p8c/RJ45 (why is everyone on this forum so concerned with calling it an 8p8c lol; get with the rest of the IT world and just call it the wrong thing, hahaha  )
Once a jack is bent to the correct angle everything lines up (first pic, center connector) vs the straight jack that doesn't (first pic, right connector)

7 hours ago, k6ccc said:

Good plan.  Make sure to zip tie (or equivalent) the pigtail so that it can't pull on the connector.

Most of us have violated the standards - and gotten away with it.  The RS-485 standard allows 32 devices on the network and a total cable length of 4,000 feet.  Most of us are no where near either one of those limits - which makes it easier to violate the standard and get away with it.  This is particularly true when operating at a low data speed.  What you can get away with at 56K is very different than what you can get away with at 1000K speed.  If you have an O-Scope, hang that across your network somewhere in the middle and compare the square waves with and without termination.  Unless you are running a really short network, it may surprise just how crappy the signal gets.  The chips do a pretty damn good job of digging the signal out of the crap.

 

FWIW, I totally agree with both you and TheDucks about properly adhering to a specification. I just find it hilarious how robust the protocols (yeah yeah, technically RS485 isn't a protocol it's a wiring topology; whatever) we use are.
I don't mind 'getting away' with non mission critical stuff like this. If it works all season; why bother fixing it? I have other things to do.

I'm sure I'll have to pay attention if/when I add pixels; but I plan on going straight E1.31 and skipping any ELOR/1M speed stuff. Ethernet is just easier at some point.

Edited December 9, 2022 by lightingman117
Forgot respond to connector cost