|
* 8 components
* +-2V to >= +-30V power supply
* 43kV/V open-loop gain @ +-10V
* >100μV input offset
* Rail-to-rail output (within 10mV)
* Slew rate 1.25μs (conditions below)
* Adjustable output impedance & current consumption
* Intentionally unrealistic
Inputs marked by switches:
* NC button = non-inverting input
* Closed SPST = inverting input
I wanted an op-amp for use in circuits where the op-amp is powered by the surrounding circuit, such as power supplies. Realistic behaviour is good, realistic design unnecessary. The output stage of this one is highly unrealistic, requiring exact precision in the BJT's gain and every parameter of the FET. The up-side of this is that the transistor parameters may be represented in text, unlike the connections of a more complex but more realistic design. This is useful in EC's Chrome app, where the parameters may be pasted to the right of the work space.
~ Component Parameters
~~~~~~~~~~~~~~~~~~~~~~
Long-tailed pair:
* 2x 3kΩ resistors
* 2x N-channel FETs
* width 300nm
* Other parameters unchanged
* 1x 1mA current source
Other:
* P-channel FET
* Width 30μm
* Length 10nm
* KP -139μA/V2
* VTO -1.49V
* Lambda -200m1/V
* NPN BJT
* Forward beta 1kA/A
* Collector resistance 10mΩ
* Base resistance 1kΩ
* Emitter resistance 10mΩ
* Other parameters unchanged
Feel fee to copy!
~ Gain and Offset
~~~~~~~~~~~~~~~~~~~~~~~~~
Gain varies with power supply, from 12kV/V at +-2V to 156kV/V at +-30V.
Input offset also varies with power supply, but is ridiculously low over the full range given. It's less than 1μV at +-10V, less than 1mV at +-2V, and less than 100μV at +-30V. It could be intentionally increased to increase gain a little. I adjust it with the PFET's KP, which increases gain a little as it's increased. To measure the offset, I watch the output offset in open-loop configuration, and then divide it by the gain.
~ Current & Output Impedance
~~~~~~~~~~~~~~~~~~~~~~~~~
Current consumption is a little high. With +-10V power supplies, it's around 100mA with the output at 0V, or over 200mA with the output pushing its negative limit. It can be reduced by reducing the power voltage, or it can be halved at the expense of doubling the output impedance. Set the output resistor to 200Ω, and set the length of the P-channel MOSFET to 20nm -- double both of them. Gain is very slightly reduced. I'm sure other values are possible.
I should have published this with a 200Ω output resistor in the first place. (I don't want to re-check everything, although it probably doesn't need it.) As-is, it's fine with low value feedback resistors: under 1kΩ but not under 100Ω. The exact limit depends on the configuration and output swing.
If the resistors are too low, the effect on a non-inverting amplifier is to clip the positive peaks. It can affect an inverting Schmitt trigger too. I'm not sure what the effect is on inverting amplifiers or non-inverting Schmitt triggers.
~ Slew Rate
~~~~~~~~~~~~~~~~~~~~~~~~~
With a +-10V supply and the output swinging the full range, the time it takes to get from 10% to 90% is 1.25μs when rising, 257ns when falling. There is a delay of about 750ns before falling begins, but the edge is very sharp.
I don't know how to measure the bandwidth in a standard way, but there seems to be no trouble at all below 100kHz, and little above. That may reduce with very low power voltage.
Just ran a frequency test with a 10x gain non-inverting setup. It started to show instability at 200kHz, but always recovered. It looked terrible at 500kHz. That fast but delayed negative-going swing makes for some fun!
|
