Here at ITER, a driver has been developed to interface to the channels of the NI PXI 6259 board, which has
binary input and output, analog input and output and waveform data capabilities. In order for the analog
input records to provide data, several parameters must be initialized:
“num_ai_chans”, “sample_rate”, “num_samples”.
Also, the “channel_type” must be set up for each record.
In the ITER implementation, each of these parameters is initialized by using an EPICS record. I am not familiar with another driver that works that way. Is this a common practice of which I'm not aware? Are there advantages or disadvantages?
My own feeling on this is:
1. My preferred approach would be to set these parameters in an initialization call. (There already is one, anyhow.)
2. Sometimes it might be useful to have quick and easy ways to set some of these parameters experimentally, but I expect that operationally they would almost always be set up once and left that way forever. In any case, records for setting parameters could be allowed, but shouldn't be required.
3. I think it unlikely that one will want to set "num_ai_chans" (controls the number of ai channels that are updated by the driver) after initialization, at least until we have an EPICS version that allows dynamic record allocation. (Would it be a good idea to set this based on the records that are found? Just an idea.)
Thanks for your input.
Pam.
Pamela Gurd
External Contractor
CODAC Division
ITER Organization, Building 519/026, DIP, Directorate for CODAC, Heating and Diagnostics
Route de Vinon sur Verdon – 13115 St Paul Lez Durance – France
Phone: +33 4 42 17 64 88