Maker Pro
Maker Pro

!! Testing a panel.

My company needs to test several panels. The panels have various looms
, connectors, fuse boxes , motor controllers and contactors etc. What
is the best way to develop a test solution for these which must do a
functional test but also check continuity between crimped wires etc.
Is a PC with an A-D card the best way or a cheap PLC or developing a
relay board that operates via a PIC chip with transistor outputs?

please replay to [email protected]

http://myweb.tiscali.co.uk/xymalf
 
C

Chris

Jan 1, 1970
0
My company needs to test several panels. The panels have various looms
, connectors, fuse boxes , motor controllers and contactors etc. What
is the best way to develop a test solution for these which must do a
functional test but also check continuity between crimped wires etc.
Is a PC with an A-D card the best way or a cheap PLC or developing a
relay board that operates via a PIC chip with transistor outputs?

please replay to [email protected]

Hi, Adrian. You might not know that newsgroups don't work that way.
You post, answer and discuss on the newsgroup. If you're looking to
hire somebody, I guess the rules are different.

Unless you're talking about dozens or hundreds of one type of panel per
month, using a fully automated test for control panel wiring isn't
going to be cost-effective. And it won't test everything you want to
test, either. You would be much better off doing a manual check with a
Shortsqueeker and an ohmmeter (possibly assisted by a PC), then testing
with a simulated machine on the test bench if necessary and the price
of failure is sufficiently high.

- - - - - - - - - - - - - - -

First off, you should be aware that most panel wiring defects are
actually the result of poor management, engineering and training. You
don't "test in" quality on a panel any more than you would any
manufactured product. I've had opportunities to work with several
manufacturers that made control panels for machines, and have some
experience in this.

Your first level of effort should be with understandable, usable
documentation. In addition to the physical panel layout (all panel
holes and cutouts indicated on print as necesary), you should also have
a wiring list with wire color and gauge, wire marker number, start and
end points for each wire, and any specific wire routing instructions.
That would be referred to as the netlist. If you have sufficient
volume to be manufacturing wiring harnesses independently of the
machine, you still require the netlist for the final assembly.

You also need to ensure the wiring technicians have the tools necessary
to do a good job. That includes good wire strippers which cut
insulation without nicking, and ratcheting crimpers which will apply a
set pressure to the crimp to ensure a gas-tight connection.
Unfortunately, determining that you have a gas-tight crimp is a
destructive test. That has to be engineered into the process by
specifying, purchasing and using the right tool properly.

Once you're done with that, you need to ensure a level of training such
that the person doing the wiring cares about doing a good job, and is
able to understand and interpret the documentation for the job, as well
as perform the necessary tasks in the way you specify, rather than the
way he's always done it. This will require an effort to develop
documentation and training materials on your company's procedures, from
how far to strip wire insulation to how to handle panels. The first
qualification (caring) is obviously the most difficult to teach.

When the tech is wiring, he should check off each wire as it is
completed on the netlist. Doing them one at a time, in numerical
order, is helpful in applying wire markers, too. For a low or medium
volume production shop, a copy of the netlist can be Xeroxed and given
to the tech with the job to pencil up as necessary.

Well, we're at that point in the game where the panel is wired. What
to do now? In a small or medium volume shop, the first thing would be
to "wiggle the wires". It isn't a bad idea to use a needlenose pliers
and just gently pull on the wires to see that they're in the screw
terminals firmly (remember, we're not testing wire crimps here --
that's assumed to be OK because of your ratcheting crimper). You might
want to occasionally get out a torque driver to ensure the tech is
cranking screws and things down properly. If the tech has access to
it, he can spot check for himself. This is also a good point to
visually inspect the connections and terminals, looking for stray wire
strands and other foolishness that can gum up the works. Just having a
second chance to look at the connection and wiggling it around to look
underneath the wire can show plenty.

The next thing would be to ohm out the direct point-to-point
connections shown on the netlist. I've found that a relatively high
speed kelvin connection (4-wire) ohmmeter such as the Fluke 8842A,
along with a GPIB-capable junker DOS PC with a footswitch attached to
the serial port and a trivial QuickBASIC program helps this part of the
process along nicely, if you do enough of these. You can also use a
serial output instrument with a PC that has a spare serial port, if you
want. It can be helpful to have the tester apply the kelvin probes to
the terminals specified on the screen, then tap the footswitch to
accept the data once the meter stabilizes and move on to the next test.
By having an output tone when the resistance value is within limits,
you minimize the amount of headshaking back and forth, and the wiring
tech doesn't have to do math in his head or have to erase the data in
the event of a miswire -- he can just fix it and continue with the test
at the beginning of that node. It also replaces the Xeroxed netlist
with a screen display and a disk file that can be copied and examined
later. Second choice would be an audible Shortsqueeker with
presettable cutoff resistance, following the netlist. Third choice is
a beeping continuity tester. And don't forget to check a
representative sample of wire nodes for shorts to the panel GND where
appropriate.

Once the operator has run through the netlist for zero ohm connections,
it's desirable to insert the relays and fuses, and then sprinkle in
some tests for ohmic connections to see that things are wired up
properly. Your coils will have known DC resistances (usually +/- 20%),
so you can check point-to-point here with an ohmmeter, too. Only put
in the relays and fuses at this point of the test, otherwise the N.C.
contacts will drive you nuts. Nominal values and/or limits for these
tests can actually be written down at the end of the netlist for
convenience if you're not using a PC.

Having done these things, you've got to wonder whether just wiring up
the panel into the machine, saying a quick prayer for intercession to
St. Liebowitz, and turning on the breaker through a GFCI is good
enough. Actually, it frequently is. But that's a personal evaluation
you have to make. It's based on price of failure (ego + machinery +
electronics + time), versus test cost in time and money, both to make
the test and to set it up each time, and how much test stuff you can
afford to have cluttering things up. And how satifying it would be to
flog the wiring technician if he bollixes it up. I've occasionally
found it to be best to build a small machine simulator on a piece of
plywood, and hook up the panel terminal block to that breadboard with a
built-in wiring harness to test the panel before wiring to the machine.
If you make a good enough simulation, just operating the piece of
simulated equipment with the control switches on the plywood will give
plenty of confidence that the panel is OK. And many things don't
really need to be tested, do they? If the wires are in the right
places, they're either go/nogo, or there's no penalty if they're wired
wrong. The possibility may also exist that it's important, it can be
wired wrong, and that part of the machine I/O is either to expensive or
too large to simulate. St Liebowitz comes in handy here.

Again, unless you've got massive quantities of panels to cover, doing
the testing manually gives the test technician a chance to look things
over and gives ownership, as well as doing things an automated test
fixture couldn't do (tidying up, looking for loose strands and scrap
wires, &c).

The whole concept of doing wiring continuity checks, followed by
functional simulation, for any non-trivial industrial control panel,
sounds like gilding the lilly here. Far more trouble than it's worth
unless you're making thousands of panels.

By the way, don't ever let anyone use shop air to blow out wire scrap
from panels. Flogging isn't good enough for that. Having them drawn
and quartered is barely sufficient. On our side of the pond, we used
to tar & feather 'em, which is much better. You see, the tar was
boiled first, then poured over 'em, then they had chicken feathers
poured on 'em. Ghastly way to go.

Good luck
Chris
 
Thanks you very much for that Chris. This has given us a road map of
where we can go from here - we only make a few panels for each customer
so I don't think the PLC route is viable. Take your point about swarf
- seen idiots doing this all the time.
 
C

Chris

Jan 1, 1970
0
Thanks you very much for that Chris. This has given us a road map of
where we can go from here - we only make a few panels for each customer
so I don't think the PLC route is viable. Take your point about swarf
- seen idiots doing this all the time.

Good to know the above was of help.

Take the time to do the netlist on the prototype and proof it before
you ship. It will increase your confidence in your work, and will make
any repeat orders a breeze.

There's software that can help here, but the ones I've seen and used
have a fairly steep learning curve. If you're a small shop or just
starting out, just do it manually, and put off the investment in the
software if you must.

Good luck
Chris
 
Top