Yes, if all you plan to measure are voltages going to the input jack of the 2i2. You should do this as minimum case to convince yourself how REW works.
But in practice, when measuring a power amp putting out say 28.3Vrms into a dummy resistor load (8ohm non inductive 100w heatsinked type), connect the attenuator (on -26dB setting) and feed that to REW. Apply 28.3Vrms to the dummy load measure with DMM across the load. What you want to do is adjust gain knob on 2i2 until it lights green and adjust to get about -10dB full scale of the 2i2. This way, you are using its max (intrinsic) dynamic range and leave 10dB away from clipping. Now click calibrate and enter the measured Vrms from DMM. Now, the 0dB FS on the graph corresponds to whatever max you sent at, in this case. 28.3Vrms.
Note that you cannot touch any gain knobs from here other measurements if you want that scale to be consistent.
But in practice, when measuring a power amp putting out say 28.3Vrms into a dummy resistor load (8ohm non inductive 100w heatsinked type), connect the attenuator (on -26dB setting) and feed that to REW. Apply 28.3Vrms to the dummy load measure with DMM across the load. What you want to do is adjust gain knob on 2i2 until it lights green and adjust to get about -10dB full scale of the 2i2. This way, you are using its max (intrinsic) dynamic range and leave 10dB away from clipping. Now click calibrate and enter the measured Vrms from DMM. Now, the 0dB FS on the graph corresponds to whatever max you sent at, in this case. 28.3Vrms.
Note that you cannot touch any gain knobs from here other measurements if you want that scale to be consistent.
I feel so dense right now.
Why should I use the voltage at the dummy load when the interface only sees the voltage at the output of the attenuator?
I am sorry, I am struggling.
Because you are testing what is applied to the dummy load - it’s not important what the actual voltage that’s measured at the 2i2 (except that it’s not clipping and has good dynamic range).
For example, the test condition is power in watts applied to the 8ohm load, not the voltage at the input to the ADC input.
You could do it that way and then measure the exact attenuation and post process apply it. But better to know without.
For example, the test condition is power in watts applied to the 8ohm load, not the voltage at the input to the ADC input.
You could do it that way and then measure the exact attenuation and post process apply it. But better to know without.
This morning, this seems to make a lot more sense to me. I don't know what my brain did overnight, but it worked.
Your patience and explanations were instrumental no doubt.
I did a whole bunch of measurements and I think I am getting reasonable results.
Also the magic box, headphones version is done:
If you do not find a 9K0 or 90k resistor in your stash - just build the 3 resistor divider, for instance
a 10k series resistor,
parallel with 100k resistor gives 9k09
and with 1k0 parallel resistor
the unloaded attenuation is 20.08dB.
A simple approach that obviously works with any other value...
For higher precision you might add 1MOhm, 10MOhm... - do the math
a 10k series resistor,
parallel with 100k resistor gives 9k09
and with 1k0 parallel resistor
the unloaded attenuation is 20.08dB.
A simple approach that obviously works with any other value...
For higher precision you might add 1MOhm, 10MOhm... - do the math