08-14-2012 12:49 PM
I recently purchased two Prologix ethernet to GPIB interfaces. I have struggled through the debug process with these and have learned the following:
1) The interfaces do not behave well when in dhcp mode. We have a local controller we set up that handles the dhcp on a private network. I used WireShark and looked at the dhcp transactions and found that the device would request an address, the controller would return an address, but the interface just kept requesting over and over, generating a lot of extraneous net traffic.
2) If you ping either a dhcp generated or static ip for the device, it actually returns duplicate data. Windows filters this out and you won't see it from windows but if you use Linux, it traps multiple returns with the same sequence number but different flight times.
Configuration requires putting the box on you local net and running netfinder.exe. Using that you can configure either the dhcp or set up your static IP.
Once that is done, you need to go to NI Measurement and Automation. Make sure VISA is installed. Go to the Devices and Interfaces, then down to Network Devices. Select Create New. Then select VISA TCP/IP Resource, then Manual entry of Raw Socket. You type in the IP address and port 1234.
Visa can then supposedly see it. It worked that way for me for several days of heavy usage, then quit.
I have had luck talking directly to the port using TCP instead of going through the VISA layer.
I have a huge need to make sure that the net traffic is correct and there is a zero error rate, so I have decided to abandon these devices. For messing around they may work OK but I can't afford to have things crash in the middle of a one week confirmation run.
08-15-2012
09:42 AM
- last edited on
03-12-2024
07:27 PM
by
Content Cleaner
Hello t_houston,
You may be interesting in taking a look to one of out GPIB ENET devices, as his is our HW you should face no issues communicating with it using MAX or LabVIEW
Regards
Mart G
08-15-2012 02:53 PM
I would love to use the NI product but it costs 6 times as much. My employer will not go for that price.
08-15-2012 04:24 PM
Time spent debugging = $$. Opportunity cost of that = 2*($$)
Lost data/failed confirmation run = $$$$
My $0.02. 😉
-AK2DM
08-16-2012 08:10 AM
Here is a response from ProLogix
Prologix GPIB-ETH transmits all packets twice. This is by design and all standard TCP stacks can handle it. The reason is as follows:
Due to limited memory in the microcontroller, the IP stack in GPIB-ETH has a TCP window size of one. This means the client must acknowledge each TCP packet before GPIB-ETH can send the next packet. Unfortunately, IP stacks on Windows, Mac and Linux use "delayed acknowledgements" -- an ACK is send only after two TCP packets are received or after 200 milliseconds, whichever is earlier [See RFC 1122, Section 4.2.3.2]. To avoid the 200 ms delay for each ACK, every packet is transmitted twice by GPIB-ETH.
For the above reason you will see multiple DHCP request and ack packets as well.