What's the purpose of this voltage noise test circuit?

Question

Ran in to this circuit from parts.io tweet.

This is my analysis of it thus far:

It is using TI's MC33078 OP-AMP. It is a dual OP-AMP, but in the example only half of it is used.

The name of the circuit is Voltage Noise Test Circuit. It suggests that it tests for noise between 0.1 Hz to 10 Hz. Checking some cut-off frequency of the filters, I see 2 low pass filters, and 1 high pass.

The low pass filters (4.3kohm/2.2uF and 2.0kohm/4.7uF) have cut-off at 16.8Hz. And the high pass (22uF/110kOhm) is at 0.1 Hz. I have used this online calculator for these values.

These number do not exactly match with the description of the circuit (0.1Hz - 10 Hz), but perhaps it's close enough.

Also, the negative feedback on MC33078 has about 5 gain factor (1+ 100K/24.3K). It is amplifying the output of DUT by that amount.

So far so good.

Now I look at DUT. I had to look it up to see what it means. DUT apparently means "Device Under Test". So this is the device I am testing the voltage noise for.

However, I am not sure why the positive feedback is there. Both inputs of the DUT are grounded. There shouldn't be any current flowing through the 10ohm resistor and the 100Kohm resistor, as there is no voltage source.

So in short, I am confused as to what the circuit is testing. Is it testing for the supply rails of DUT?

Answer 1

Both inputs of the DUT are grounded. There shouldn't be any current flowing through the 10ohm resistor and the 100Kohm resistor, as there is no voltage source.

Wrong - the internal voltage noise source of the inputs is always present and that is what is being tested by putting the amplifier into a gain of 10,000.

Here is a more general view of op-amp noises: -

The noise source (implied by the circuit in the question) that is specifically being tested for is labelled \$e_n\$ in the diagram above.

These number do not exactly match with the description of the circuit (0.1Hz - 10 Hz), but perhaps it's close enough.

They are using two filters at 16.8 Hz which together produce a 3dB point of 10.8 Hz. Having said that this will have a noise bandwidth of 12 to 13Hz (not an exact science) so it's a little too much beyond the 10 Hz limit in my book - this calls into question the guy who designed it.