====== fft freq mismatch ======

   * fsampling = 20 Msps
   * fin ~= 3Mhz (2459 peak)
   * FFT len 16384
   * fbase = 20e6/16384 ~=  1220.703125 Hz

^ diff [hz] ^ % fbase ^ left [dB] ^ right [dB] ^ SINAD [dB] ^ THD [dB] ^ SNHR [dB] ^ SFDR [dB] ^
| -207.52  | -17.00  |  -13.8  |  -16.8  |   10.0  |  -68.8  |   10.0  |   13.8  | 
| -109.86  |  -9.00  |  -20.1  |  -21.7  |   15.7  |  -74.2  |   15.7  |   20.1  | 
| -61.04  |  -5.00  |  -25.6  |  -26.4  |   20.8  |  -79.3  |   20.8  |   25.6  | 
| -36.62  |  -3.00  |  -30.2  |  -30.7  |   25.3  |  -83.7  |   25.3  |   30.2  | 

^ diff [hz] ^ % fbase ^ left [dB] ^ right [dB] ^ SINAD [dB] ^ THD [dB] ^ SNHR [dB] ^ SFDR [dB] ^
| 36.62  |   3.00  |  -30.7  |  -30.2  |   25.3  |  -83.6  |   25.3  |   30.2  |
| 61.04  |   5.00  |  -26.4  |  -25.6  |   20.8  |  -79.1  |   20.8  |   25.6  |
| 109.86 |   9.00  |  -21.7  |  -20.1  |   15.7  |  -74.0  |   15.7  |   20.1  |
| 207.52 |  17.00  |  -16.8  |  -13.8  |   10.0  |  -68.5  |   10.0  |   13.8  |

{{:adc2:simulations:fsim_diff_coarse.png|}}

{{:adc2:simulations:fsim_fft_coarse.png|}}

^ diff [hz] ^ % fbase ^ left [dB] ^ right [dB] ^ SINAD [dB] ^ THD [dB] ^ SNHR [dB] ^ SFDR [dB] ^
| -37.84  |  -3.10  |  -29.9  |  -30.4  |   25.0  |  -83.4  |   25.0  |   29.9  | 
| -20.75  |  -1.70  |  -35.2  |  -35.5  |   30.2  |  -88.6  |   30.2  |   35.2  | 
| -10.99  |  -0.90  |  -40.8  |  -41.0  |   35.7  |  -94.1  |   35.7  |   40.8  | 
| -6.10  |  -0.50  |  -46.0  |  -46.1  |   40.8  |  -99.2  |   40.8  |   46.0  | 
| -3.66  |  -0.30  |  -50.4  |  -50.5  |   45.3  | -103.6  |   45.3  |   50.4  | 
| -2.44  |  -0.20  |  -54.0  |  -54.0  |   48.8  | -107.2  |   48.8  |   54.0  | 
| -1.22  |  -0.10  |  -60.0  |  -60.0  |   54.8  | -113.2  |   54.8  |   60.0  | 

{{:adc2:simulations:fsim_diff_fine.png|}}
{{:adc2:simulations:fsim_fft_fine.png|}}

^ diff [hz] ^ % fbase ^ left [dB] ^ right [dB] ^ SINAD [dB] ^ THD [dB] ^ SNHR [dB] ^ SFDR [dB] ^
| -3.78  |  -0.31  |  -50.1  |  -50.2  |   45.0  | -103.4  |   45.0  |   50.1  | 
| -2.08  |  -0.17  |  -55.4  |  -55.4  |   50.2  | -108.6  |   50.2  |   55.4  | 
| -1.22  |  -0.10  |  -60.0  |  -60.0  |   54.8  | -113.2  |   54.8  |   60.0  | 
| -0.61  |  -0.05  |  -66.0  |  -66.0  |   60.8  | -119.2  |   60.8  |   66.0  | 
| -0.38  |  -0.03  |  -70.2  |  -70.2  |   65.0  | -123.4  |   65.0  |   70.2  | 
| -0.21  |  -0.02  |  -75.4  |  -75.4  |   70.2  | -128.6  |   70.2  |   75.4  | 
| -0.12  |  -0.01  |  -80.0  |  -80.0  |   74.8  | -133.2  |   74.8  |   80.0  | 
| -0.06  |  -0.01  |  -86.0  |  -86.0  |   80.8  | -139.2  |   80.8  |   86.0  | 

{{:adc2:simulations:fsim_diff_log.png|}}

**remark** : 1%@20Mhz@16kp = 12.7 Hz

**conclusion** : if we want to achive resonable measurement resutls for 10-bit ADC working @ 20 Mhz (and with 16k FFT len) input signal freq should not differ from ideal one more then ~1 Hz. 

**conclusion 2 ** : reducing lenght of FFT (fbase is becoming larger) should relax this requirement (but also degrade precision of measurement)