Hi All

I have a phase shift oscillator, and would like to find the minimum overall gain required from component tolerances given, I am using LTSpice, I would like to ask what kind of method should I take? Since I believe it's possible to do this just using LT Spice and I think I don't need to derive any equations but I may be wrong
thanks
Art

 

If you don't write the equations you are guessing. You can plot the equations to look for component interaction.

 

If you don't write the equations you are guessing. You can plot the equations to look for component interaction.

I see I have used equations to find the theoretical gain can should I just plug in the uncertainties into this and find the minimum that way?

 

If you don't write the equations you are guessing. You can plot the equations to look for component interaction.

would you mind explaining what you mean plotting the equations to look for component interaction

 

Here's some info you may find useful.

 

Here's some info you may find useful.

That's perfect thank you

 

Lets assume that four components can change value enough to change frequency. Hold three component values constant while cycling one component value through its possible range, then plot the worst case transfer function values (gains) versus the component value. After four tries you can see the effect of one variable on gain four different times. Assuming that a range of component values exists, then a worst case value also exists, but where in the range does it exist--- finding this point is the object of finding worst case solution. Finding the worst case solution involves crafty thinking, good circuit knowledge, 256 tries with graph matching, or a big fast computer. Circuit analysis programs are approximations, so it is chancy to depend on their solutions unless 3 or so programs agree.

Unless you are doing this for a very high volume production job it is usually easier to add feedback that keeps the gain constant.

The phase shift of the amplifier(s) can be critical in this design, so the amplifier 3dB BW >> 50 signal frequency.

 

Lets assume that four components can change value enough to change frequency. Hold three component values constant while cycling one component value through its possible range, then plot the worst case transfer function values (gains) versus the component value. After four tries you can see the effect of one variable on gain four different times. Assuming that a range of component values exists, then a worst case value also exists, but where in the range does it exist--- finding this point is the object of finding worst case solution. Finding the worst case solution involves crafty thinking, good circuit knowledge, 256 tries with graph matching, or a big fast computer. Circuit analysis programs are approximations, so it is chancy to depend on their solutions unless 3 or so programs agree.

Unless you are doing this for a very high volume production job it is usually easier to add feedback that keeps the gain constant.

The phase shift of the amplifier(s) can be critical in this design, so the amplifier 3dB BW >> 50 signal frequency.

Yes thank you so much I feel so inexperienced with designing circuits can you give me any recommendations/advice to become more proficient?