Calculating for OpAmp Open Loop Gain

[beaker353]

I have literally been banging my head into a table for the past 3 hours trying to figure out how to approach this lab question I have for my solid state II class.  I think that either I simply don't have enough information to do the calculation or I'm over complicating it and looking right past the simple approach (my money is on the latter). Neither the textbook nor a hammering Goggle for the past half hour is helping.

I have your typical non-inverting opamp configuration with a a 200kohm Rf and a 1kohm Ri. 50mVpp sin wave source. My measured gain is 204/46.19dB (10.2Vpp Out) my upper critical frequency (-3dB from mid) was measured at 4.9kHz. I've determined successfully that the unity-gain bandwidth is 980kHz.

I'm trying to calculate the open loop gain, and then the open loop critical frequency. All the equations and derivations I can come up with requires to know either the open loop gain to determine the open loop critical frequency or vise versea. The one that I was able to come up with doesn't seem to play nice is Vout/Vin = Aol/(Aol*B), which gives me a Aol of -13,668, which is completely just wrong.

Any suggestions of how I need approach this? I'm probably going to feel really stupid when someone shows me some easy equation I seen 2 dozen times, but I have no pride left.  I gave up pride when I decided on top of two full time jobs I was going to add pursuing a EET degree at the age of 35...

-EM

[T3sl4co1l]

Hmm, let's look at things then.

- Gain is already high.  This helps some because the error from ideal (i.e., 201 according to the resistors -- assuming their values are exact) will be larger.
- But, you're still trying to subtract a small difference out of a big measurement (i.e., the couple milivolts the op-amp needs at its input to do its job, out of a measurement of 10V).  Anywhere you see a scaling issue like this -- physically or numerically -- proceed cautiously.
- Check your expectations.  You should intuitively expect gain is slightly less than 201, because the gain is finite.  How does this compare to your measurement of 204?
- Aol = -13668 sounds pretty reasonable to me.  It's wrong, but that's a byproduct of the measurement, which is never exactly correct either (even doing this in SPICE, you are limited by finite tolerance parameters).
- Lastly, consider sources of error in both the input and output voltages.  How did you measure them?  If on a scope, you're assuming the scales are *exactly* in the proportion given (i.e., 10mV/div is exactly 200 times more sensitive than 2V/div or whatever).  What is the effect if they are not?  And how much are they allowed to be off by?  (For a scope, usually a few percent -- an error on the order of 1/200th is impossible to be sure of!)

If you have access to a very good ratio (such as a ratio transformer), you could possibly do the measurement on the same scale, so that both measurements have the same proportional error.  You could also do it differentially (let the circuit do the actual subtraction for you!), or use existing components (measure the feedback directly?).

Overall this sounds more like a lesson in tolerances and errors -- and statistics -- than electronics.  Hopefully you'll have some courses soon to round that out?

Tim