[sdiy] Spice vs. real life noise problem

[Rutger Vlek]

Hi guys,

I recently wrote about making some progress with a variation on the "discrete OTA" idea, well...now it seems I'm stuck. I've done a few weeks of debugging and simulating, and I just can't figure out 
what's going on. I have a VCA running, and that works pretty well (given known limitations of OTAs and discretes) but has a problem in the output stage. There's a constant level of white noise on the output, that is quite a bit higher than I was expecting. It is totally uncorrelated with the VCA's control voltage, so it's even there (at the exact same level) when the VCA is maximally closed. I have several of these VCAs on a board, and as far as I can measure, each circuit behaves exactly the same. Because of this, and because of the "white" nature of the noise, I excluded layout-problems and interference as a possible source (am I right to do so?). Other possible sources I suspected were: opamp-noise (there's one 1/4 of an TL074 opamp in the output stage) or resistor noise. For this, I've moved to my Spice simulation (first in B2 Spice, later in TINA-TI with an opamp noise-model by Art Kay: ftp://ftp.ti.com/pub/linear_apps/noise_article_series/article11-noise4-jan2007.pdf). However, the noise doesn't show up in Spice. The noise that is seen in the Spice model is several factors smaller (and much, much more acceptable) than what happens in real life. This has made me really confused..... what noise source can be white, and doesn't show up in Spice? Is anyone experienced with such issues and interested in helping out? I would prefer talking privately over mail with a few people about this, rather than having large public discussions on the list. Although if we do any shocking discoveries or gain insights, I will of course share this.

Best,

Rutger