| Electronics > Projects, Designs, and Technical Stuff |
| PICs with Op Amps: Non-inverting? Off-piste? |
| << < (2/2) |
| NivagSwerdna:
I'm tempted to just put a INA180B3 into the design |
| David Hess:
--- Quote from: NivagSwerdna on March 31, 2020, 10:15:55 am ---... and in inverting mode there is less CMRR error since both inputs held to ground (or virtual ground) and hence no CM dynamic voltage. --- End quote --- That is right; if the common mode voltage is not changing then there is no common mode error. And for the same reason, inverting mode has less distortion than non-inverting mode. 45 dB of common mode rejection is truly horrible, and barely better than the first modular tube operational amplifiers 50 years ago. Good precision parts are 120 dB or 1 microvolt per volt. Chopper stabilized amplifiers can be 160 dB. "High performance" operational amplifiers like the 741 set the standard at 90 dB typical. |
| fourtytwo42:
My view for what it's worth is that one should not approach microcontrollers integrating analogue electronics from a precision point of view, rather they are a means of increasing circuit density and reducing component count. If anybody want's a precision solution they should be using separate components that will always be better. |
| NivagSwerdna:
--- Quote from: fourtytwo42 on March 31, 2020, 07:04:39 pm ---My view for what it's worth is that one should not approach microcontrollers integrating analogue electronics from a precision point of view, rather they are a means of increasing circuit density and reducing component count. If anybody want's a precision solution they should be using separate components that will always be better. --- End quote --- It's a trade-off.... but one that can only be made once you understand the limitations... As was mentioned above.... CMRR is in the datasheet if you look hard enough. I'm actually tempted to put both solutions on the board and do some experimentation. |
| David Hess:
--- Quote from: NivagSwerdna on March 31, 2020, 09:08:03 pm ---It's a trade-off.... but one that can only be made once you understand the limitations... As was mentioned above.... CMRR is in the datasheet if you look hard enough. I'm actually tempted to put both solutions on the board and do some experimentation. --- End quote --- Dual slope integrating analog-to-digital converters which included the input buffer (voltage follower) faced the same problem with low CMRR on a CMOS process albeit with an analog CMOS process so not nearly as bad. The solution for them was to either require an external buffer, in which case a precision JFET or bipolar operational amplifier could be used, or cleverly do the automatic zero cycle with the buffer common mode input voltage set to the input voltage so the common mode error was included as part of the zero error and removed. I suspect Intersil got a patent on this to the detriment of competitors like Siliconix and Texas Instruments. Let's see ... they needed a minimum common mode rejection ratio of ... 86 dB for 4-1/2 digits or 20,000 counts of resolution and that is tough to do on an analog CMOS process even now. Many early multimeters which used these had linearity problems because of this and inspecting old equipment sometimes reveals that the integrated input buffer was replaced with an expensive precision JFET or low input bias current bipolar operational amplifier. |
| Navigation |
| Message Index |
| Previous page |