Speed up acquisition time by removing multiple delay() calls... #9
Conversation
Removed all delay() calls, as the AD9850 DDS has microscopic settling time for frequency changes and serial communications. Changes reduce acquisition time for a 1-30 MHz scan from 11.74 seconds to 1.65 seconds.
|
Thanks for making the contribution! Based on a HR360 post I think the LCD delays might be required for stable operation, but I'll test it out and can merge at least the tuning related changes. |
|
Mine seems to be rolling fine, but your mileage may vary and only testing will tell. I will add that if you look at the code for the LiquidCrystal library, it is full of calls to delayMicroseconds() to account for the slow LCD interfaces. |
- added automated calibration routines to setup(). Measures the FWD/REV ADC values when no input is appled to compensate for idle noise. These offsets are subtracted from the later FWD/REV reads. - added diode mismatch compensation values, empirically read at the ADC when a shorted antenna port is being drive. The ratio of measured value (REV/FWD) is used to compensate the FWD value.
- The two compensation ratios (Gain and Diode) turned out to be frequency dependent, requiring equations to determine the compenstation ratio during the sweep (left in the mean value and linear methods of compensation as comments). - Changed display routines to provide a more readable output with fixed sizes and scaling/ - Moved the ADC readings more stable by adding an averaging routing. - Removed the "throwaway readings" routine, as the problem seems to have solved itself. - removed unused code and variables.
|
I didn't realize that my later edits to the local copy were going to show up here as commits in the pull request (warned you all I was a Git newbie). For the purposes of this pull request, I only want you to consider the first commit I offered, where I removed the delay() calls. |
|
@etchorner no worries. I can easily separate them out. If you create a different branch for your own work then it won't be added to this pull request. Sorry it's taking a while to get this merged, didn't have much free time this weekend.
|
Removed all delay() calls, as the AD9850 DDS has microscopic settling time for frequency changes (~43ns) and serial communications (~30ns). Further, the ADCs on the Pro Micro acquisition time is only 13 - 25 µs, so there should be no need for any delay() calls.
Proposed changes reduce acquisition time for a 1-30 MHz scan from 11.74 seconds to 1.65 seconds or less. Reduces code size 142 bytes.
related to #8