Instrument Control (GPIB, Serial, VISA, IVI)

cancel
Showing results for 
Search instead for 
Did you mean: 

Getting 100-200ms latency from data rec'd to program action. Any ideas?

We are trying to isolate an apparent 100-200ms latency from data xmitted to NI485/4 and program response. Using win2k with virtually no CPU being used. Code appears to be well written and has system priority. We have Intel D845GBV2 board and have three PCI cards, NI485/4, Data xlation 3153, and I/O card.

I suspect PCI problems. The NI 485/4 works fine except for latency and installation shows no problems except for memory address conflict between each of the four ports, however I suspect this is not a problem as all four ports are on the same card.

Any suggestions?
0 Kudos
Message 1 of 10
(3,593 Views)
Hmm - the 100-200ms latency is unexpected and should not occur with the PCI-485/4. I'm a little concerned about the memory address conflict. The PCI-485/4 requires 4 separate IO windows even though the ports are on the same card. If there is a conflict, and two ports share the same address, the hardware will not function properly.

Any way you can tweak your memory settings to resolve the conflict?
Message 2 of 10
(3,593 Views)
Chris,

Thanks for the fast reply. I reloaded the driver (but did not physically remove card as of yet; system is installed in cabinet and not easily accessed). System shows card working properly under "Device Manager"; no exclamation points or other symbols show up on the ports. However, when I right-click and check properties| resources tab for any one of the used ports (three in my case)it shows the memory address as also being used by the other two active ports. I appear to be unable to change the memory I/O at this screen as settings are grayed out.
If you can confirm this is not a good situation, I will investigate further. Thanks again
0 Kudos
Message 3 of 10
(3,593 Views)
This may be ok - what are the I/O settings assigned to your card?
0 Kudos
Message 4 of 10
(3,593 Views)
What are some of the parameters of your setup?:
1. Baud rate
2. How many bytes are requested in the "read" api call to the NI485/4?
3. What RS485 cable are you using - 4-wire, 2-wire?
4. What Transceiver/Wire mode is the serial port configured to?
5. What is the serial data Flow Control mode being used?
6. What is the speed of the computer processor?
0 Kudos
Message 5 of 10
(3,593 Views)
1. 19200
2. I will check but data rate is exceedingly low.
3. 4 wire
4. 4 wire
5. none
6. celeron 2 GHZ, Intel 845GBV board
0 Kudos
Message 6 of 10
(3,593 Views)
FF8FEC00-FF8FEC7F

same for each port
0 Kudos
Message 7 of 10
(3,593 Views)
OK - that's the memory window for the PCI interface and its ok (actually expected) to duplicate across the ports. Does each port also have an I/O window (0x3f8 or something along those lines)?
0 Kudos
Message 8 of 10
(3,593 Views)
Yes, each does with no conflict showing.
0 Kudos
Message 9 of 10
(3,593 Views)
So, one byte time is about 10/19200 = 0.521ms which could be significant considering that your measurement magnitude is about 100 - 200ms. That raises a question whether the measurement was accurate, or whether some multiple of byte times crept into it.
0 Kudos
Message 10 of 10
(3,593 Views)