Problem is that if the circuit is modified to accommodate 4-channels summed together instead of only one channel, then the value C18 for stability may have to be found by trial and error. Need a scope, and maybe some way to look at an FFT. Would suggest some careful listening too, or perhaps a more objective measurement using a biometric sensor 🙂
trying to be funny or sarcastic? sorry, have to ask explicitely as I don't seem to get it...
In another thread Jakob2 pointed out that using humans as sensors for measurements is considered objective as opposed to subjective. IOW properly performed perceptual testing results are considered to be an objective measure. One example almost nobody will claim is subjective would be a threshold of audibility. Humans are the sensors, yet the result is objective.
It makes it easier for the active filter, but it may increase the load impedance that the DAC sees, which may affect the performance of the DAC itself. Chances are that there is an optimum somewhere. You need a resistive term somewhere to damp an LC filter.
As I said with C18 in circuit there are oscillations both in practice and in simulations depending on the value of C18. To determine a stable value probably requires tuning in circuit which in turn will require proper tools. The possible benefits of C18 have not yet been confirmed in practice.
Would you be willing to share yout sim files and elaborate a bit how to detect oscillation in both the sim and in real world? where to look (which frequency band, possibly predicted by sim) etc? that would be especially useful to ME as I will have pcbs made a few weeks into the future and after assembly testing will begin...
This is what oscillations look like in LTSpice. The schematic is from post #32. In actual circuit the oscillations can be seen e.g. as elevated high order distortions in THD spectrum.
I used LT1028 as integrator to speed up the simulation. With OP161x the simulation is very slow or does not converge.
I used LT1028 as integrator to speed up the simulation. With OP161x the simulation is very slow or does not converge.
You would see this very clearly on any scope. Why would you want to infer it from distortion measurements?
So hook up a scope to the output of the i/v and look for "stady state" sinuisodal oscilltaions in the tens of MHz range?
Here is a picture of how I implemented this. To the left is the circuit of this thread. To the right is plain OP1612 I used as comparison.
Work of art! I will have the opportunity to incorporate all this into one pcb. Let's see if it will all work out in the end 🙂 Thanks again for sharing!
You should be able to see it as widening of the trace of 1kHz signal. Similar to what you see in LTSpice if you let the simulation run a bit longer.
Where can I find the loop gain probe used in your circuit?
If I replace it with a straight wire the simulation fails with ".set prb=0 Unknown control card"
Here are the asc and asy files. But it should work without these at least if you change ".set prb=0" to a comment.