I continue working on my micro boost effect. I wasn’t entirely happy with the noise performance from the first take. I wanted to investigate a bit more. How did it go? See the video. Links can be found in section where I’m trying to make sense out of all of this ?.
As a reminder here’s the schematic:
In the previous post I used OPA379 (OPA379AIDBVT) op-amp and for this post I used TLV313 (TLV313IDBVT) both in tiny SOT23-5 package. The pin-out is interchangeable but make sure to choose SOT23-5 package, there’s also SC70 package that might look similar but it is smaller and with different pinout.
Comparing TLV313 to OPA379 – it’s less noisy, has wider GBW, but also consumes a bit more. Whereas this circuit with OPA379 might run idle for years, with TLV313 in place that drops to 6-9 months. Quiescent current is 5uA vs 65uA + whatever other components are using.
In terms of noise, the main contributor is broadband input voltage noise density (I think) which is 26nV/√Hz for TLV313 (comparable to TL072) vs 80nV/√Hz. Flicker noise and current noise contribute little compared to the other noise contributes.
Let me extend the comparison table from previous post a bit:
|Input voltage noise density (nV/√Hz)
|Input current noise density (fA/√Hz)
|Gain Bandwidth Product (GBW)
|Quiescent current per amp (typical/max) (uA)
|Output Noise (mVpp)
Hmmm, that output noise figure does not make sense … or does it?
Making Sense Out of It
First off, I’m happy that my calculation from the previous article. Looks pretty close. It should be since I use Texas Instruments educational videos on op-amps (TI Precision Labs Op-Amps).
One thing I did not spot immediately is reference to Analog Engineer’s Calculator. It’s free to use, and I used it in the video. The figures above are coming directly out of it.
Turns out that GBW affects the noise output greatly. OPA379, even though noisier, has more limited gain bandwidth and so produces lower noise Vpp output. I have to remember that noise still exists outside of audio spectrum too and is contributing to the overall noise level. I’m still figuring out what this means for audio signal to be honest, but I’m getting there ?.
If I limit the gain for frequencies that are over the spectrum we’re interested in, I can lower this noise figure a bit. To do this is simple, and something like this we’ve seen in many pedals before (or other audio circuits):
How the above works is that at higher frequencies the capacitor acts as a short circuit and limits the gain to 1 (0dB) instead of 4 (12dB). Cut-off point is calculated like for low-pass filter, and in this case that’s about 15KHz. I had 100pF SMD capacitor handy, if this is too low and you want to keep gain for frequencies higher than this, use lower cap value.
For my cut-off frequency calculations I use this handy online tool. Anyway, the idea is that noise at frequencies that we’re not interested in is not amplified.
I’m skeptical whether my calculations are correct though, and how to use the Analog Engineer’s Calculator for this. When I use SPICE (not very reliable – depends on the quality of SPICE models), I see noise level improvement of 30% for OPA379. For TLV313 improvement is 55%. TL072 model is rubbish for noise simulation.
In the video I also went through testing OPA379 circuit, TLV313 circuit, breadboarded TL072 circuit and I also tried out MXR Micro Amp pedal. I don’t think much can be gained out of this comparison except that 12dB gain might be a bit too high for what I’m doing ?.
Too many wires, this will always add noise, and I can’t really compare this to a breadboarded version of a circuit and get anything meaningful. Also, my perception of noise is not very reliable.
Other than that, I go through some thinking about the circuit, and I also go through advice given in datasheets and contemplate what I could do to improve my circuit for the version 2.