Hello,
I would like to kindly ask for an advice regarding the accuracy of the impedance measurement on AD5941ELCZ when using frequencies above ~50kHz. I'm using the modified version of example AD5940_Impedance, i.e. the ratiometric approach. Even if the high power mode is enabled, the accuracy deteriorates with the increasing frequency independently on chosen Ctia/Rtia.
I'm using non-isolated 4-wire configuration. The unknown impedance is a resistor having 165 kOhm connected between CE0/RE0 and DE0/SE0. I have tried the connection via P6 as well as directly via P7 but there is no difference in the results.
Below are the results of the measurement for the range from 100Hz to 100kHz. A little bit worse accuracy of the measurement at 100Hz is caused by the insufficient number of periods of the measured signal. But for 50kHz, 70kHz and 100kHz a quite huge magnitude as well as phase error is observed.
| I (Rcal) | I (Rz) | Rz | ||||||
| Freq | Re | Im | Mag | Re | Im | Mag | Mag | Phase |
| 100 | 119 | -24557 | 24557,29 | 41 | -1483 | 1483,567 | 165528,7 | 1,305987 |
| 141,25 | 143 | -24501 | 24501,42 | -13 | -1478 | 1478,057 | 165767,7 | -0,83834 |
| 199,53 | 235 | -24533 | 24534,13 | 19 | -1485 | 1485,122 | 165199,4 | 0,184224 |
| 281,84 | 315 | -24523 | 24525,02 | 14 | -1483 | 1483,066 | 165367 | -0,19506 |
| 398,11 | 448 | -24526 | 24530,09 | 25 | -1486 | 1486,21 | 165051,3 | -0,08263 |
| 562,34 | 628 | -24521 | 24529,04 | 34 | -1485 | 1485,389 | 165135,4 | -0,15548 |
| 794,33 | 888 | -24514 | 24530,08 | 49 | -1486 | 1486,808 | 164984,9 | -0,18597 |
| 1122,02 | 1254 | -24497 | 24529,08 | 69 | -1485 | 1486,602 | 165000,9 | -0,27009 |
| 1584,89 | 1770 | -24465 | 24528,95 | 97 | -1484 | 1487,167 | 164937,4 | -0,39829 |
| 2238,72 | 2498 | -24400 | 24527,54 | 137 | -1480 | 1486,327 | 165021,1 | -0,55675 |
| 3162,28 | 3522 | -24271 | 24525,21 | 193 | -1474 | 1486,582 | 164977,2 | -0,79699 |
| 4466,84 | 4956 | -24015 | 24521,06 | 271 | -1462 | 1486,905 | 164913,5 | -1,15919 |
| 6309,57 | 6951 | -23506 | 24512,21 | 382 | -1437 | 1486,907 | 164853,7 | -1,58683 |
| 8912,51 | 9677 | -22500 | 24492,74 | 533 | -1388 | 1486,82 | 164732,4 | -2,26493 |
| 12589,25 | 13274 | -20541 | 24456,73 | 736 | -1291 | 1486,061 | 164574,2 | -3,18385 |
| 17782,79 | 17667 | -16807 | 24384,38 | 993 | -1103 | 1484,135 | 164300,3 | -4,4332 |
| 25118,87 | 22070 | -10024 | 24239,75 | 1274 | -756 | 1481,422 | 163624,9 | -6,25804 |
| 35481,34 | 23933 | 1085 | 23957,58 | 1466 | -158 | 1474,49 | 162480,5 | -8,7471 |
| 50118,72 | 17742 | 15266 | 23405,75 | 1282 | 696 | 1458,746 | 160451,2 | -12,2125 |
| 70794,58 | -1711 | 22290 | 22355,57 | 309 | 1394 | 1427,836 | 156569,6 | -16,8878 |
| 100000 | -20111 | 3613 | 20432,96 | -1141 | 747 | 1363,778 | 149826,2 | -23,0276 |
Configuration of the switches: Dswitch: CE0, Pswitch RE0, Nswitch SE0, Tswitch DE0 (tried also SE0LOAD with the same results).
DFT: SINC3, OSR 2, 16384 samples
Rtia: 5k
Value of the registers (IMPCTRL_START): REG_AFE_PMBW: 000421FD REG_AFECON_CLKSEL: 00000000 REG_AFE_HPOSCCON: 00000030 REG_AFE_HSDACCON: 0000000E
The same degradation of the accuracy is observable also for 10 kOhm resistance chosen on the Z TEST board.
| I (Rcal) | I (Rz) | Rz | ||||||
| Freq | Re | Im | Mag | Re | Im | Mag | Mag | Phase |
| 100 | 118 | -24571 | 24571,28 | 168 | -24544 | 24544,57 | 10010,88 | 0,117015 |
| 141,25 | 143 | -24514 | 24514,42 | 142 | -24486 | 24486,41 | 10011,44 | -0,00196 |
| 199,53 | 235 | -24547 | 24548,13 | 233 | -24519 | 24520,11 | 10011,43 | -0,00404 |
| 281,84 | 315 | -24536 | 24538,02 | 311 | -24509 | 24510,97 | 10011,04 | -0,00854 |
| 398,11 | 448 | -24540 | 24544,09 | 443 | -24512 | 24516 | 10011,46 | -0,01049 |
| 562,34 | 629 | -24535 | 24543,06 | 621 | -24507 | 24514,87 | 10011,5 | -0,01701 |
| 794,33 | 889 | -24528 | 24544,11 | 879 | -24501 | 24516,76 | 10011,15 | -0,02106 |
| 1122,02 | 1254 | -24511 | 24543,06 | 1240 | -24484 | 24515,38 | 10011,29 | -0,02946 |
| 1584,89 | 1771 | -24479 | 24542,98 | 1752 | -24453 | 24515,68 | 10011,14 | -0,03991 |
| 2238,72 | 2499 | -24414 | 24541,56 | 2473 | -24389 | 24514,06 | 10011,22 | -0,0545 |
| 3162,28 | 3524 | -24285 | 24539,35 | 3486 | -24262 | 24511,16 | 10011,5 | -0,08019 |
| 4466,84 | 4960 | -24028 | 24534,6 | 4907 | -24011 | 24507,28 | 10011,15 | -0,11331 |
| 6309,57 | 6955 | -23518 | 24524,85 | 6883 | -23512 | 24498,77 | 10010,65 | -0,1575 |
| 8912,51 | 9683 | -22513 | 24507,05 | 9585 | -22526 | 24480,46 | 10010,86 | -0,22273 |
| 12589,25 | 13282 | -20552 | 24470,31 | 13156 | -20604 | 24445,96 | 10009,96 | -0,31418 |
| 17782,79 | 17677 | -16816 | 24397,83 | 17530 | -16935 | 24374,07 | 10009,75 | -0,44084 |
| 25118,87 | 22083 | -10028 | 24253,24 | 21955 | -10259 | 24233,64 | 10008,09 | -0,62243 |
| 35481,34 | 23946 | 1087 | 23970,66 | 23945 | 719 | 23955,79 | 10006,21 | -0,87918 |
| 50118,72 | 17751 | 15276 | 23419,1 | 18071 | 14887 | 23413,33 | 10002,47 | -1,23244 |
| 70794,58 | -1717 | 22301 | 22367 | -1041 | 22354 | 22378,23 | 9994,983 | -1,73636 |
| 100000 | -20121 | 3610 | 20442,28 | -19986 | 4471 | 20479,99 | 9981,586 | -2,43834 |
Questions:
Q1) Is is possible to explain the reason of this behavior and suggest, how to improve the accuracy for the measurements above 25kHz?
I tried also a different approach (measurement of the voltage between RE0 and SE0, current through Zunknown via HSTIA) but the same results were obtained. The RTIA / HSDAC / ADC calibration does not have any significant effect on this issue.
Q2) Looking at the oscilloscope, it seems that the amplitude of the excitation signal generated by the excitation buffer decreases with increasing the frequency (confirmed also when I tried directly measure the voltage between P and N node using ADC). Is this an expected behavior? According to the documentation, BW should be 250kHz.
Thank you
Edit Notes
Results for 10kOhm on Z-test board added.[edited by: zdva at 8:17 AM (GMT -4) on 7 Sep 2020]
