Problems with PC/104 GPIB

[Rod B. Nussbaumer]

Greetings:

I have been trying to get a NI PC/104 GPIB board working, unsuccessfully.
My first attempt was to use the NI-4882-DDK to create a functioning driver
under vXWorks. I was able to compile and build the driver, and it seems to
work to some extent. Whenever I try to get it to communicate with an instrument
or to get it to execute bus commands (UNL, UNT, etc.), it fails with a timeout.
Through inspection of the code and enabling debugging output, I was able to
discern that it should be generating interrupts as a way of responding to
writes to various registers. These interrupts seem not to be happening, as
evidenced by the driver's debugging output.

My second strategy was to test the board under MS-DOS, using a set of DOS
based NI tools. Using the 'ibdiag' tool, I was able to see the board pass the
diagnostic tests using a variety of settings of IO address, DMA addres, and
IRQ selection. Also, the board would fail the tests in ways that are expected
under certain other jumper settings. The version of ibdiag used was 1.0.
Having confidence that the board is functionally configured, I tried to
communicate with a simple instrument (HP5384A Freq. Counter). I used the
'ibic' tool (ver 3.01) to try to control and read the instrument. Whenever
I try to send data or bus commands, the commands fail with a timeout error.
Essentially, this seems to be the same behavior as my VxWorks driver.

The cables, instrument, etc. have all worked in an arrangement using a NI
GPIB-Enet10 interface, so I am confident that the fault is not related to
that part of the setup.

I am thinking that the board (never has been working, was purchased new
about 6 months ago) may be faulty. Is there any basic software tests that
can be used to confirm this? I don't want to send it back to the factory
until I can be quite certain of a fault. Even better, any suggestion on what
I can do to get it working? I have not yet tried the board under any other
OS, but if it would be useful, I could attempt to build & run Linux and
the open source drivers for the board. My end goal is to use the board under
VxWorks ver 5.4. The PC/104 motherboard is a ZFmicro-based (486 clone),
MZ104+, built by TRI-M Systems.

Thanks in advance for any help.

Rod Nussbaumer.
Controls Software (ISAC), TRIUMF.
Canada's National Laboratory for Particle and Nuclear Physics

[DiegoF]

Hi,

It definitely sounds like the board is not generating interrupts. The only other test I can think of is to use the Windows driver, which is the lastest driver for this interface: NI-488.2 1.7.

You could also check the interrupt line by connecting a scope to the jumper. The jumper connects the interrupt line of the TNT4882C to the ISA bus, so that line will reflect if the chip is interrupting and the PC is not detecting the interrupt.

DiegoF
National Instruments.

[Rod B. Nussbaumer]

DiegoF wrote:
> Hi,
>
> It definitely sounds like the board is not generating interrupts. The
> only other test I can think of is to use the Windows driver, which is
> the lastest driver for this interface: > "http://www.ni.com/support/gpib/versions.htm#embindus">NI-488.2
> 1.7.
>
> You could also check the interrupt line by connecting a scope to the
> jumper. The jumper connects the interrupt line of the TNT4882C to the
> ISA bus, so that line will reflect if the chip is interrupting and the
> PC is not detecting the interrupt.
>
> DiegoF
> National Instruments.

Thanks for your reply.

I have pretty much convinced myself that the board is faulty. I have
now tried it on a conventional desktop PC, using a PC/104-to-ISA adapter.

The results were identical to th PC/104 hosted tests. Also, I have
monitored the GPIB using an IOtech GPIB Analyzer. No commands that
assert the GPIB data lines are getting onto the bus. A few passive-read
type commands (IBIC commands, that is) seem to work, but most commands
fail with a timeout error.

I will try to run the board under the Windows drivers, but getting a PC
running Windows on an ISA based computer will take a little more effort.

Should this not work, who at NI would I contact, either by phone or e-mail,
regarding a RMA for test/repair/replacement?

Thanks for your help.


Rod Nussbaumer.
Controls Software (ISAC), TRIUMF.
Canada's National Laboratory for Particle and Nuclear Physics

[DiegoF]

Hi,

You would contact us by phone. You can go to www.ni.com/ask where you'll be given a toll free number to call and a service request number.

DiegoF
National Instruments.