Would like to operate the AD8302 Phase/Gain detector at sub-Hertz frequencies. How low can it go?
What components need addressing (cap values and type), and what performance should be expected?
AN-691 addresses low frequency operation of AD8302 and some other detectors.
To help get part of the way there - let's start with audio frequencies.
CAN THE AD8302 WORK DOWN TO AUDIO FREQUENCIES?
It is possible to use the AD8302 down to audio frequencies. Such an applicationrequires different values for some of the external capacitors. The inputcoupling capacitors (C1 and C5 in Figure 14 of the data sheet), offsetcompensation loop capacitors (C4 and C6) should be 0.1 microfarads or larger.It will also be necessary to use filter capacitors for the output bufferamplifiers. These are shown as C2 (MFLT) and C8 (PFLT) in Figure 14, both ofwhich should be 1 microfarad or larger.
Still need to answer the VLF question.
In theory, yes, provided that the cutoff of the dc offset compensation loop is low enough, which means huge capacitors. The input capacitors would also be huge. At some point you'd transition to electrolytic capacitors... The settling time of the system would be huge and there might be ripple on the outputs. Some applications are a can of worms best left closed!
Why not use a dual audio codec and calculate the phase and magnitude in DSP?
I use AD8302 as a detector at frequencies near 4 MHz. When I measured my detector I have found that the output phase noise was too big. I have tried to explain this fact. My explanation is written in the attached file. Please, answer me: am I right or not?
The phase error in and around 0 degrees of phase difference (and pretty much to +/- 30 degrees of phase difference) is very large, please reference TPC 28. You are probably seeing this inherent phase error. What I recomend is putting a known phase offset on one of your signals, say 90 degrees and then measuring your phase difference with the 90 degrees calibrated out, i.e., if your two signal were already 10 degrees out of phase and one of them went through the known phase delay of 90 degrees, the signals would be 100 degrees out of phase but the phase differenence error at 100 degrees is much less than at 10 degrees. So the Vphs output should be about .8V at 100 degrees and we have a known 90 degrees of phase difference which is about .9V, so really only have about 0.1 V of phase error.
Hope this helps,
I already use 90 degrees phase shift by means of two phasers, one channel has shift +45 and anouther -45 degrees (it gives zero first derivative d(phase)/d(frequency)). Thus the zero input phase difference correspondes to the middle output voltage range. There is interesting fact that output noise is minimal for 0 or 180 input signal phase shift (without external phasers). I think that it is connected with zero derevation d(output voltage)/d(phase diffrence) at these phase differences. Also I want to note that the mesured output noise depends on input signal amplitudes nonmonotonic. I think that the origin of this dependence is similar with VMAG output error - internal amplifires saturation.
Excuse me for my bad English, I am a foreign man.
Retrieving data ...