- From: John Conner <conner_at_empiredi.com>
- Date: Thu, 30 Oct 2003 13:18:17 -0700
Is anyone using the MCC PCI-DAS1602/16 board at moderate to high sample rates - greater than 16000 channels per second? I keep comming back to the board and hitting a fifo overflow problem. I have not been able to get it to work properly at any rate greater than 1000 scans per second of 16 channels - 16000 channels per second. And that is a bit questionable. The cb_pcidas driver does not seem to properly handle the burst mode scanning - converting all channels as quickly as possible at each sample peroid. The driver works somewhat better if the sampling is evenly spaced. I'm trying to get the board to sample 16 channels at 4000 scans per second - thats 64000 channels per second. MCC's specs say it should do 200k channels per second. I'm able to do it easily with a PC104-DAS16jr/16. For a simple test bed I'm using the cmd.c demo from comedilib. Changing the number of channels to 16 and the frequency to 64000 and the number of scans to say 640,000 (160 seconds). About one in 6 runs gets a fifo overflow on the DAS1602 board. If I let the system set for a while (20 min or so) it almost always fails on the first try. Then works for a number of runs. I have not seen the PC104 board fail. The system is a 450MHz K6, 384Meg RAM. Debian 'testing'. Kernel is 2.4.22 compiled with current gcc-3.3.2. Comedi 0.7.67 and comedilib-0.7.19. I'm not getting the oops that I saw in early September with the current setup. If you've got a setup that works with the PCI-DAS1602/16, I'd like to see the comedi_cmd you are using. Thanks, John Conner
Received on 2003-10-30Z20:18:17