Re: Sweep Response Results After Completing All Calibrations

Nick Kennedy

Let me give it a shot.  Sometimes I don't remember how software I wrote works, but maybe I can answer these.


On Wed, Dec 17, 2014 at 3:46 PM, lh1907@... [PHSNA] <PHSNA@...> wrote:

To try another tack on this Freq Calibration issue, based on various test results, please comment if any of the following assumptions are not valid.

1. Windows program only uses polynomial coefficients in parameters.txt file. TRUE

 2. Polynomial coefficients in .ino file remain zero and never need to be updated. 

YES - at least, in the Windows version of PHSNA, the polynomial coefficients in the .ino file aren't used.  Those in EEPROM aren't either. I can see that the Arduino source code could be confusing because it was converted from the stand-alone version and still contains some things not used in the Windows version.

 3. All polynomial coefficients in parameters.txt file must be zero when Slope/Intercept Calibration is performed and uncompensated Response Sweep data (next item) is collected. 

NO?  Well, I don't think so.  I think the calibration routine uses uncompensated values and ignores the poly values.  If you've seen something to suggest otherwise let me know so I can investigate.

4. Valid Slope/Intercept constants are inserted in parameters.txt file and windows program restarted before a set of uncompensated Response Sweep data (in .csv type file) is collected to determine polynomial coefficients.  


YES -- The starting point is having an accurate slope / intercept set so you can make accurate "raw (true)" measurements.


5. After using the documented Excel Charting procedure to determine polynomial coefficients, the resulting six polynomial coefficients are placed in the parameters.txt file. YES


6. The compensated Response Sweep is only valid over the frequency range used to determine the six polynomial coefficients in previous item.  YES

The FreqCal_4-Results.gif graph in W2HHV PHSNA Type II Checkout folder shows the results obtained over the 2-27 MHz frequency range. Step size was .1 MHz. The compensated Response sweep is above the uncompensated Response sweep by about 1.6 dBm which equals the amplitude range of uncompensated Response Sweep. Is this a program processing constraint or just a coincidence? Was my previous assumption that the compensated Response sweep should be plotted within the amplitude range of uncompensated Response Sweep invalid?


I had to run one on mine to remember what it should look like.  The setup is, output of DDS into AD8307 power meter.  Choose "Response Sweep".  Set high and low limits of scan, generally to correspond to the expected range of your DDS.  I chose 1 to 50 MHz as I have a DDS60.

The compensated plot should look flat across the entire range and have a value of 0 dBm.  Yours has a value of about -8 dBm but it's fairly flat.  So I suspect your "zero" constant of your polynomial set is wrong.

My compensated plot is fairly flat at +0.1 dBm. So not too bad.

The uncompensated plot should show the actual output. "It is what it is."  Mine starts at +0.5 dBm at 1 MHz and slopes linearly down to -3 dBm at 50 MHz.  About what you'd expect, I guess.

The standard build would result in about -10 dBm from the DDS, so your value of -8 dBm is in line with that.  Your roll off with increasing frequency looks normal.

" Was my previous assumption that the compensated Response sweep should be plotted within the amplitude range of uncompensated Response Sweep invalid?"  

I GUESS SO.  The compensated sweep should be flat at 0 dBm over the full range.  The uncompensated should be the actual value of the DDS' output and may be greater or smaller than 0 dBm and would generally fall off with increasing frequency.


Nick, WA5BDU

Larry, W2HHV

Join to automatically receive all group messages.