Instrument cluster calibration

[nizamani]

I am doing my final year project on machine vision based automotive instrument cluster inspection, and I am using virtual ports CAN 256 and 257, to stimulate the front panel objects,since I don't have any actual CAN Hardware, I will be using NI 6009 USB to transmit the CAN Signals over USB. What I am really confused about is how should I go about the callibration phase once an instrument cluster is inspected for an error, let me rephrase that: Where exactly should I do the calibration, since I would be the one to induce a calaculated error (and than later correcting it). 

would it be more efficient

a) if the error is compared with say scaling information stored in .ncd db, everytime the vi is run  and corrected in an inline fashion ? 

b) or should I induce an error in scaling information in .ncd file and than correct it in an offline mode, and importing the database again [would it be effective in anyway?]

c) or hijacking the CAN Frames and corrupting it in any fashion would help ? that being should I move from channel to frame API, for more practical means?

is there any better way?

Can anyone help ?

Thanks.

[Gibson]

lets start with some basic clearifications first:

you are saying that you do not use a NI CAN hardware at all and want to transmit/receive CAN data on your NI 6009 ?

[nizamani]

Thank you for replying:

yes i am not using any NI CAN Hardware, and the transmit/receieve part over usb is done. 

The problem is I want to be able to inspect instrument cluster for errors, that error or its consequence has to be physical in nature for machine vision algorithms to detect [on gauge needles and lights].My question is where exactly should I induce an error that I would be calibrating later using vision?

My understanding is: any unexpected value in scaling information stored in ncd file, for speedometer signals can lead to improper angle measure on speedometer, I can always detect that angle measure and compare it with any standard angle, so changing the scaling information in ncd db file, is one way to induce an error.

or Do i make a standard ncd file,[and not make any changes to it]. rather use any other way to make instrument cluster respond differently and compare the results with the scaling information stored in the standard ncd file and any differece is compensated to make IC respond in a correct fashion?

or is there any other, better way ?