Measuring slew rate

[sahko123]

I've made a discrete opamp and am now characterising it. I've fed it some square wave and it seems to be stable and happy. Should I measure slew rate at unity gain or open loop? Also the output waveform has a delay before the square wave starts to rise. Do I measure from the beginning of the rise or from when the input goes high?

[TimFox]

The convention is to measure "rise time" from 10% to 90% of the full transition.  If that is slew-rate limited, then that would be a good region to measure the time derivative (slew rate).
Note that the slope might not be limited by the slew rate:  it could be limited by the slope of the input waveform, or due to the frequency response of the unit under test.  This can be checked by changing the amplitude of the waveform:  a true slew-rate limited wavform will have a constant slope (and therefore a longer rise time) as the amplitude increases.  Of course, at lower amplitude, the slew-rate may no longer be the limiting factor and then the rise time will be constant and the slope will reduce as the amplitude is reduced.
When measuring slew rate on an audio-frequency amplifier, with gain determined by feedback, I drive the input with a sine wave and increase the frequency until I can see a ramp (constant time derivative) section in the output waveform and measure its slope.

[magic]

https://www.eevblog.com/forum/metrology/measuring-slew-rate/
 

[David Hess]

I've made a discrete opamp and am now characterising it. I've fed it some square wave and it seems to be stable and happy. Should I measure slew rate at unity gain or open loop? Also the output waveform has a delay before the square wave starts to rise. Do I measure from the beginning of the rise or from when the input goes high?

The delay is caused by limited bandwidth and slew rate is completely independent of it so measure from the 10% to 90% points while ignoring the delay.

Bandwidth and associated delay are a small signal phenomenon and linear.  Slew rate is a large signal phenomenon and non-linear.

[sahko123]

I love how you linked to my own post. I just wasn't able to access it through phone

[garrettm]

Read Jim Williams' Taming of the Slew (LT Application Note 94).

Slew rate is related to the rise time of the step function applied to the OPA's input. So to test a very fast slew rate, you would need a very fast rise time signal source.