I have this circuit using TL064IN OPAMP to amplify a 50Hz signal two times with different gains, the two output signals has to be in phase with each other. The problem is when I did the simulation to figure out the phase shift between two signals and then try to compensate for this phase shift by using capacitors, the LT spice simulation gave me a 38 mili degree when using the capacitors shown on the schematic. But in reality the phase is more than that. Is there any way to know the real phase shift and find the exact values for the capacitors other than try and error?
the graph of Gain VS frequency vs phase is not very useful for accurate calculation.
-
\$\begingroup\$ Use a better or more complete model of the op-amp. Newer op-amps are likely to have had more time spent characterizing the performance than old dinosaurs like the TL064! \$\endgroup\$– Andy akaJun 19, 2016 at 12:17
-
\$\begingroup\$ I will try to find one similar to TL064. But since it was made, they didn't refine their spice model to be very close to real? \$\endgroup\$– Ashraf AlmubarakJun 19, 2016 at 12:43
-
\$\begingroup\$ Spice models are compromises and most op-amp macromodels use the typical values from the datasheet. Newer devices may be better characterised, but a real circuit will inevitably be slightly different. \$\endgroup\$– Peter SmithJun 19, 2016 at 12:53
-
1\$\begingroup\$ There is an old (but still accurate) application note on OPAMP macromodels and the compromises made: linear.com/docs/4139 \$\endgroup\$– Peter SmithJun 19, 2016 at 13:37
-
1\$\begingroup\$ In addition to Neil_UK's suggestion, I notice that 50Hz is right in the middle of the internal compensation filter transition region (for the ST part - TI shows a different frequency range); I would use all 4 of the amplifiers - put a unity gain buffer behind the low gain path. The total delay is likely to be well matched within the four parts and you may not need the external capacitors. \$\endgroup\$– Peter SmithJun 19, 2016 at 15:19
1 Answer
First things first. You need to stop asking for insane levels of accuracy. Sure, your Spice package gives phase measurements with a resolution of 0.9 micro-degree, and group delay resolution of 1 psec. Do not pay attention to this. When measuring 50 Hz phase, for instance, your 38 mdeg corresponds to 2 usec. Do you really care about microsecond differences? If so, why? And consider that your real-life components will probably be 1% resistors and 10% capacitors, so the equations which produced the numbers really cannot be used at this level of accuracy.
With that in mind, tell us a) what accuracy you need, and b) what accuracy you're getting. "more than that" is not remotely useful.
-
\$\begingroup\$ I may not state this very clear , the 38 mdeg is very good with me if I have had on real . But unfortunately the real measurements gave a phase shift about 1 degree or slightly less than 1. And yes this 1 degree is very sensitive to the application. \$\endgroup\$ Jun 19, 2016 at 19:07
-
\$\begingroup\$ And I had try too many values of capacitors, so the tolerance have no much effect on phase. The most dominant thing on my opinion is the phase vs frequency vs gain characteristics of the opamp. And that's what I am trying to deal with it \$\endgroup\$ Jun 19, 2016 at 19:11