"Bank Switching" inputs and outputs

mikeexplorer

Supporting Member
Join Date
Nov 2018
Location
Scranton,PA
Posts
178
I am not sure if that is the right term, but I am looking at an idea to be able to test cables that are in production to insure they are assembled correctly. Say for example one cable has 24 conductors and what I want to be able to do is energize one wire and check on the other side that it is wired correctly (not crossed to the wrong pin) Building a PLC with enough I/O to do that can be done, however there are some cables that might have many more conductors and I would like to be able to take a slot of Inputs and Outputs and bank switch them to the conductors. So as an example, energize one output, then say take a bank of 16 inputs, connect them to the first 16 pins, check them. Then switch them to a second set of 16 wires, and so forth. The only way I could think of doing this is with a hell of a lot of small relays, but I am wondering if there is some other way. The outputs would also be "bank switched" in a case where there are many conductors.



The idea is to build the base PLC controls, then have a breakout box with the connectors specific to the cable being built. Then it could be an input of how many conductors the cable has so it knows how many to check.



Mike
 
Bank switching means you are only testing 16 pairs at a time and cannot detect miswirings between banks e.g. pin 0 in bank 0 to pin 18 in bank 1.

That being the case, perhaps hard wiring all per-bank pin 0s, 1s, etc., would be good enough. You would still not detect pin 0 in bank 0 wired to pin 16 in bank 1 or to pin 32 in bank 2 or to pin 3200 in bank 100, but maybe this approach drops the false negatives to an acceptable rate.
 
A way to detect that is for example, select bank 0 of outputs and energize output 1 (pin 1 for the cable) then on the input side, select bank 0 of inputs (pins 1-16) and scan them. Then select bank 2 of inputs (pins 17-32) and so forth until the maximum number of conductors is reached. Then repeat the process by energizing output 2, bank 0, then repeat the scan.



Mike






Bank switching means you are only testing 16 pairs at a time and cannot detect miswirings between banks e.g. pin 0 in bank 0 to pin 18 in bank 1.

That being the case, perhaps hard wiring all per-bank pin 0s, 1s, etc., would be good enough. You would still not detect pin 0 in bank 0 wired to pin 16 in bank 1 or to pin 32 in bank 2 or to pin 3200 in bank 100, but maybe this approach drops the false negatives to an acceptable rate.
 
A way to detect that is for example, select bank 0 of outputs and energize output 1 (pin 1 for the cable) then on the input side, select bank 0 of inputs (pins 1-16) and scan them. Then select bank 2 of inputs (pins 17-32) and so forth until the maximum number of conductors is reached. Then repeat the process by energizing output 2, bank 0, then repeat the scan.



Mike




good point, I forgot that there are two banks.


Would a multi-pole rotary switch do the job?
 
The reason why I am looking at a "bank switching" method is I could be looking at the possibility of testing a cable with 100 conductors or more. These cables are custom built for a customer for various machinery.



The example I give for one cable is 24 conductors which would be fine with expansion I/O. However if the conductor count is higer, the amount of expansion I/O would be many.



With bank switching, I could take two output modules of 16 outputs, normally that would be enough for 32 conductors. But if I use the second output module as a bank switch toggle, now that allows me to test 256 conductors (16 * 16) instead of 32.


This idea may not be practical at all. one way to bank switch is a series of 16 relays for each bank, but the count of the relays would be huge. I thought maybe there was a practical or something already out there that could do it.



I looked at a Click PLC and using Automationdirects PLC build tool, added the modules to the max limit which would allow the PLC to test 70 conductors. Each input and output module would cost $47 each, so that is $94 for a set of 16 modules. So if I wanted to use the bank switching method, the parts would have to cost less then that to be practical. Total cost for the PLC and modules is under $500 and adding a 3 inch monochrome LCD screen as an interface adds about $110 for a total cost around $600 for the hardware. This would be on the cheaper side as far as a PLC option goes.



The idea is to have this as part of the building process to catch a crossed wire or other wire problem before the cable is completed and final tested. This would reduce the cost of re-work considerably. These are custom built cables that can cost in the range of thousands of dollars each. They are built to Class 3 specifications, they are not a home user type cable :)


Mike






Why 16? Use 32 channel IO cards, what is your max cable density?
 
These are custom built cables that can cost in the range of thousands of dollars each. They are built to Class 3 specifications, they are not a home user type cable :)


What kind of connector?


I assume the alternative it somebody with a VOM manually checking N*(n+1)/2 possibilities with N=100, so 5k tests.


Going from that to a manual "switch" - two 16-pin connectors - drops N to 7, so 28 tests, and then press a button to initiate each N=16 automated sub-test, saves around three orders of magnitude steps. Considering that a relay-driven test would require up to 14 connections be made (up to 100 conductors), I think perfect maximized automation with relays might be the enemy of good partial automation.


(I hope I got that right, I did get the other approach wrong)
 
Connectors vary depending on what the cable is used for. I can't go into details about these cables because of security reasons.



I was figuring that the concept of bank switching was used for some application and I wanted to see if it is a cost effective idea to test these cables during assembly to catch a miswired pin before the assembly is complete.



All cables go through a rigorous testing before they are completed and shipped but that equipment is expensive and is tied up testing the final products before shipping. I was looking at a way to perhaps build or find something less expensive to use in the intermediate assembly process to simply detect miswired pins or broken wires before the cable is sent to the next step. Once it reaches the final step and if a miswired connector is found, it is expensive and time consuming to tear it down to repair the problem.



Mike








What kind of connector?


I assume the alternative it somebody with a VOM manually checking N*(n+1)/2 possibilities with N=100, so 5k tests.


Going from that to a manual "switch" - two 16-pin connectors - drops N to 7, so 28 tests, and then press a button to initiate each N=16 automated sub-test, saves around three orders of magnitude steps. Considering that a relay-driven test would require up to 14 connections be made (up to 100 conductors), I think perfect maximized automation with relays might be the enemy of good partial automation.


(I hope I got that right, I did get the other approach wrong)
 

Similar Topics

GenerL question. I dont have the part number infront of me. But i am looking at a valve with an M12 connection on it. Pin1 - not used, pin 2 -...
Replies
13
Views
3,173
I work on a 4 roll steel wire calender for a tire company. I have tasked with trying to automate the rubber supply bank for the upper and lower...
Replies
2
Views
560
Hello, Currently working to integrate a Micro820 PLC (2080-LC20-20QWB) with a Numatics G32 valvebank. (240-325)...
Replies
14
Views
2,197
Hi All, Solenoid bank controls air to 8 cylinders. We used to able to move cylinder up and down by hand no problem but now there is a small amount...
Replies
5
Views
2,358
Hi, I am trying to understand in theory how many I/O module I can place in the local bank for the compactlogix 1769-L33ER . the bulletin says max...
Replies
5
Views
1,615
Back
Top Bottom