Next I had to find out how the OLED noise was getting into the rest of the receiver. It could have been through the SCL SDA or even the ground lines. It could have been just through capacitive or inductive coupling from the display board itself. A big clue came when I tried powering the display from a completely separate power supply: BINGO! The noise disappeared. So I knew the noise was going into the rest of the receiver through the Vc line that powered the OLED.
I had been powering the OLED from the 5V regulator on the Arduino Uno. In an effort to isolate the noise, I put a separate 5V regulator in the circuit for the OLED. No joy -- noise still there. I then tried putting an RC low pass filter between the OLED and the 5V regulator. Still had the noise. Finally I remembered something from the AF AMP circuits of Roy Lewallen, Rick Campbell and Roger Hayward. ( I think Roy was the pioneer on this one.) They all used an "active decoupler" between the first AF amp and the power supply line. I confirmed that it was my first AF amp that was picking up the OLED noise. I built the active decoupler (just three parts!) and the noise disappeared. GONE!
There are only three parts, but the way this circuit works is kind of complicated and not very intuitive. There is a good discussion of how it works here:
www.facstaff.bucknell.edu/dkelley/elec351/Lab/elec351lab5_sp04.doc
Roy, Rick and Roger were using this circuit to knock down 60 Hz AC hum, but I found that my OLED noise was at around 200 Hz -- I figured (correctly) that the active decoupler would take care of this as well. I think this little circuit can be useful in dealing with the kind of noise generated by the digi displays that many of us are now using.
David Rowe has a really interesting analysis of this circuit here:
http://www.rowetel.com/?p=4781
This sort of thing wasn't uncommon decades ago, described as a "capacitance multiplier". Back when low voltage power supplies were often about adding more capacitance even though the hum never went away. The concept seemed to fade as three terminal regulators became common. But obviously there's still use.
ReplyDeleteMichael
I also looked into the circuit about a year ago here
ReplyDeleteWow David, I was really pleased to find out that there is someone else out there interested in how this active decoupling circuit really works.
ReplyDeleteWhen I looked at your write up -- especially the LTSPICE simulations -- it was quite eerie. I have the same simulations here.
I think we might have slightly different takes on this.
I think you can see the circuit as TWO different circuits. The 47K resistor and the 10uF cap form a standard RC low pass filter with a very low Fc. That obviously will make a big contribution to knocking the noise down, but the voltage drop across that resistor could be problematic, right?
The second element is the transistor. I think that it too -- by itself -- would contribute to the filtering. This is because the emitter current is FAR MORE of a function of base current than it is of the collector voltage (where the noise is).
With the LTSPICE model, disconnect the 47K resistor and the 10uF cap. Cut the line from the collector to the base circuit. Set up a separate 11 volt source and feed that through another 47k resistor to the base. Watch the noise output. It disappears.
So the RC circuit in the base can be seen as a way of getting the needed 11 volts DC without the noise. The transistor itself contributes to the noise reduction and -- of course -- also makes a big contribution by carrying the current to the load. This avoids the voltage drop you'd get by trying to pull that current through just the RC filter.
I'm not quite sure of my analysis. What do you think?
Bill
Hi Bill,
ReplyDeleteYes I was riding my bike yesterday here (on a 38C Adelaide day!) listening to SolderSmoke and heard you talking about this circuit and it rang a few bells! Fine Business - really getting our head around these circuits is one joy of the hobby.
I like your analysis - the step of holding the base at a fixed voltage is a very useful thought experiment (or simulation).
Yes I agree the the resistor selection is a trade off between DC voltage drop and filtering. The gain of the transistor makes it possible to decouple the two circuit elements, at the expense of a little voltage drop to keep the transistor away from saturation.
I also went a little nuts analysing an oscillator that wouldn't start and again trying to find the source of a notch in a double tuned filter. I fear I have "the Knack" !!
73, David VK5DGR