Hardware?

[Peter King]

This may sound a little general, but I'm just doing some investigating
and I don't have much experience in this particular field, but any
help is greatly appreciated.

I work at a local manufacturing company that specializes in vehicle
instrument panels and who also do some Circuit board manufacturing.
With any company such as this testing is a key element. What I've
been looking into is the possibility of implementing automated testing
via software (LabView) and required Hardware interfaces (this is where
I need help).

For such a setup what kind of hardware should I be looking into to
allow the use of LabView to handle the testing of various instrument
setups, digital boards, analog boards, cable harnesses, and such.

This may be too vague but I'm only doi
ng an initial investigation to
get a sense of what would need to be purchased to get started and
possibilities for future expansion.

Thanks for any help,
Peter
pking@newtech.ca

[Dennis_Knutson]

You've got a lot of possibilites. How are you testing now and what instruments are you using. To weed out 90-95% of manufacturing defects, you might want to look at first having in-circuit testing done on the boards. If additional functional testing is required at the board or system level, then you'll need to evaluate what measurements need to be taken, what kind of accuracy is required, volume requirements, and many other factors. LabVIEW (and especially along with NI's TestStand) is very well suited to automatic testing. LabVIEW can communicate to instruments inside of a PC, over serial cable, GBPIB bus, VXI bus, and PXI. GPIB is probably the most common and you'll find the widest variety of instruments available with it. PXI instruments are pci cards wi
th a different physical form and additional triggering. They're usually a little more expensive but are a lot faster. If you can define the measurements, I'm sure a lot of people here can give more specific advice.

[Peter King]

For now we are testing our panels using custom made test boxes, which
provide power, input signals (controlled by manual switches), LED's
for output, points for "Scope and Meter", etc. We'd like to move to
a more automated software controlled system for a couple of reasons:
to eliminate human error, to allow more flexibility as designs are
updated and changed. The equiptment in this system would have to be
able to provide input signals (waveforms, dc voltages, control
signals) and be able to gather output (waveforms, dc voltages,etc) the
instruments we deal with and need to test are things like
speedometers, tachs., switch and circuit breaker clusters, controls
for throttle and vehicle operations, proximity sensors, etc.

For the circuit cards, I have to admit I don't have much dealing with
the boards we produce, the ability to test them with this system is
not of primary concern, but would most likely be a future
implementation. The only thing I can see is the ability to handle
Digital I/O for fuctionality testing of the completed board.

Thanks in advance for any responses.

Peter K

Dennis Knutson wrote in message news:<506500000005000000DD520000-1005954886000@exchange.ni.com>...
> You've got a lot of possibilites. How are you testing now and what
> instruments are you using. To weed out 90-95% of manufacturing
> defects, you might want to look at first having in-circuit testing
> done on the boards. If additional functional testing is required at
> the board or system level, then you'll need to evaluate what
> measurements need to be taken, what kind of accuracy is required,
> volume requirements, and many other factors. LabVIEW (and especially
> along with NI's TestStand) is very well suited to automatic testing.
> LabVIEW can communicate to instruments inside of a PC, over serial
> cable, GBPIB bus, VXI bus, and PXI. GPIB is probably the most common
> and you'll find the widest variety of instruments available with it.
> PXI instruments are pci cards with a different physical form and
> additional triggering. They're usually a little more expensive but are
> a lot faster. If you can define the measurements, I'm sure a lot of
> people here can give more specific advice.

[&lt;Mike Ross&gt; mross@smpcorp.com]

Peter,

I am just past being where you are...I had to purchase a general setup
for a lab to work on more or less unknown projects.

Dennis' advise to try and qualify and quantify the types of
measurements you need to make is the best you will get. You have to
be able to resolve the events you are interested in (requiring
hardware speed), you have to be able to process the multitude of
signals (bandwidth) and you will need aquired data of sufficient
quality to be useful (signal conditioning). This is your homework
assignment.

If you economize now, you will be at best dissatisfied later, and at
worst, forced to buy what you need later after some suffering trying
to get insufficient hardware to make do.

I think you should get an NI field rep in to help. That said, I can
explain roughly how I decided to go:

I see that there are three general areas of speed/bandwidth in the NI
line:

Fast - PXI is needed for simultaneous aquisition of many inputs at
high resolution and higher numbers of control tasks. PXI has a high
speed bus for big data dumps to the hard drive(s). I don't need this.

Medium - PCI. I opted for PCI (bought a Dell in a tower to get 5
expansion slots) because I don't see an simultaneous triggering or
huge data transfers in the future. I bought the fastest DAQ card NI
makes. (You can have more than one...if you have slots to put it in.)
So I can stream to the hard drive, but the DAQ card has finite
bandwith which is fairly easy to use up (1.25Msamples /second). We
have GPIB interface for connecting to stand alone instrumentation.
(I thought this was optional, but now think it is mandatory.)

Slow - Fieldpoint. I opted for a Fieldpoint setup to manage
millisecond and slower tasks (saves bandwidth on th DAQ card). I
picked RS485 for better noise suppression, long cable runs, and for
expandability into the plant environment.
I use a USB to RS485 converter to keep a slot free.

The other topic is signal conditioning. I am still not far enough
into this to have a good feel for what hardware is offered and the
economies inherent in the various choices. But you need to condition
your DAQ. I picked the NI SC-2345 as a start. It is a glorified
breakout box with lots of off the shelf modules you can buy rather
than rolling you own conditioning circuitry. The upscale option here
is SCXI. It is the signal conditioning counterpart of PXI. It gets
you speed and bandwidth.

I have not looked much outside of NI. I don't have time. NI seems to
have all the bases covered.

One other thing. You need a training budget. At least get the course
kits for LabVIEW. They seem quite good.

Good luck,

Mike Ross

Dennis Knutson wrote in message news:<506500000005000000DD520000-

then you'll need to evaluate what measurements need to be taken, what
kind of accuracy is required, volume requirements, and many other
factors.