Help - getting GPIB timeout w/over 32 bytes per read

[SynergyMike]

Hi Folks,

Well, I _did_ have all my timeout issues under control... until now.  After a PC quit, we got a refurb system with Win XP/SP3 on it.  IT installed a GPIB card, and I installed the NI 488 drivers and LV runtimes (development is done on another system).  However, I couldn't get any of the software I had written to work on it.  Using Measurement and Automation Explorer, I tried communicating directly with the instruments using ibwrt/ibrd commands in the interactive window.  Writing to the instruments works fine.  Reading works for data lengths up to 32 bytes at a time.  At 33+ bytes, the read always times out, regardless of the timing settings under the board's properties under M&AE.  I really don't want to re-write all my reads to 32 byte chunks, but I can't figure out why it always times out with longer reads.  The other PCs in house running the same software does not do this.  I think this is the only XP/SP3 system though - could it be an OS related issue?

Regards,

Michael Tracy

Synergy Microwave

[SynergyMike · Accepted solution]

I found the answer via Google search - apparently the system has a DMA conflict.  Here's a link that explains it:

http://digital.ni.com/public.nsf/allkb/862567530005F09C8625633B0064BCA7

Regards,

Michael

[billko]

---

SynergyMike wrote:

I found the answer via Google search - apparently the system has a DMA conflict.  Here's a link that explains it:

http://digital.ni.com/public.nsf/allkb/862567530005F09C8625633B0064BCA7

Regards,

Michael

---

Interesting, Mike.  Sorry we couldn't help you.  I was monitoring this thread as someone interested in the solution, not as a solution provider.  I'm guess that because you marked it as "solved" that following the prescribed solution worked?

Thanks!

[SynergyMike]

Actually, I have to confess that I jumped the gun, but I did find a solution anyway.  DMA was not an issue.  I was playing around with "MAX" some more and I tried NI SPY just for the fun of it.  There was a setting for buffer size (it said small was 64 bytes and large was something much higher - not near the system now) and it was set to small; setting it to large cured the problem.  I would have assumed that setting was specific to NI SPY, but apparently not?  Also, the break was definitely at 32 bytes not 64, so I'm still confused.  But it works now, so no complaint. 

In my searching, I found that the GPIB card apparently has a 32-byte buffer, so it looks like there was some sort of buffer reset issue.  Also interesting is that when I Googled 'GPIB "32 bytes" ' this morning, my post from this forum was the very first hit - I would have thought Google would have taken longer than that to pick it up.  🙂

Regards,

Michael

[billko]

Very cool.  Hey thanks for the update!