Coupling cap 360deg phase shift??

[MondyT]

Hi All
I am toying with the below circuit for an instrument preamp and I wonder if you experts can help me with an issue that I noticed when simulating the circuit in LTspice. The circuit seems to work well in sim, but why do I seem to be getting a 360 degree phase shift in the signal if I use 1uF as coupling cap values for C1, C3 and C6? If I change these to say 4.7uf I get an expected phase shift response of 0 degrees for the majority of the bandwidth. I wanted to use 1uf here so that I can limit the low end response somewhat and also use non polar plastic film caps in for coupling rather than unbiased electrolytics etc. Can anyone explain why this should happen?



Cheers
Ray

[Conrad Hoffman]

If you limit the low end, you're making a filter and thus get the phase shift associated with whatever response curve you manage to make. You can't have one without the other. BTW, your supply will be very poor (if it works at all) in practice because the DC return path is through 100K resistors. You need a proper bipolar supply. With one battery, either get a TI "rail splitter" or build the equivalent. That's probably also messing up the sim. [you can also bias all the inputs up to halfway, but IMO that's a pita.]

[godfrey]

Originally Posted by MondyT

...why do I seem to be getting a 360 degree phase shift in the signal if I use 1uF as coupling cap values for C1, C3 and C6? If I change these to say 4.7uf I get an expected phase shift response of 0 degrees...

That's just LTspice changing it's mind about how to display the result. 360 degrees is the same as 0 degrees.

[jcx]

0 degrees = 360 degrees, the sim software can have some ambiguity choosing reference branch for the phase unwrapping - you can search for "phase unwrapping"

it can matter how low you start the AC analysis relative to the corner frequency as to what the sw decides "0 phase" is

what matters is the phase difference in the audio band, more exactly the linearity of the group delay - in Ltspice try plotting tg(V(out)/V(in)) (change to Cartesian plot)

[MondyT]

Thanks for your help! JCX and Godfrey were right, I changed the frequency plot range from 3Hz - 300Khz and the plot was back referenced to 0 degrees shift. Phew!! LTspice experts for sure!!

With the coupling caps as shown I am getting around +/- 12 degrees phase shift at the lowest and highest frequencies of interest (around 80Hz and 30Khz) with double this at around 20Hz. Do you think this OK or could this cause problems further down the signal line with say EQ circuits or further signal processing? Should I aim for less phase shift?

Cheers
Ray