Two Pole DC Rail Voltage Decoupler - John Dunn, Consultant, Ambertec, P.E., P.C.

The following sketch is of a two-pole rail voltage decoupling circuit.  

 
The circuit works by doing a current division between capacitor C1 and the remainder of the circuit whose impedance is examined as follows:   

The absolute values of the above circuit's series impedance versus frequency vary with the circuit elements. On the left in the above sketch, the effect of the variations in the emitter resistance R1 is seen. On the right, we see the effects of varying resistor R3.

When we feed power in from a DC source having a zero (negligibly low) source impedance, the load sees a rail-voltage source impedance consisting of the parallel combination of this series element and the aforementoned shunt capacitance, C1.

You want this filter to be either over damped or critically damped. If you let it be underdamped, there can be transient response voltage excursions undershooting or overshooting the final settling voltage which would not be a good thing.

The following table shows some examples of damping for value combinations of R1 and C1:   

 

Comments

[david pacholok]

John:
You come up with the most arcane BUT USEFUL circuits and great analyses thereof!
Back at Motorola Comm div, about 1978 I remember a similiar circuit used to isolate the VCO in an 800 MHz PLL from +12 VDC supply noise corruption... it worked great!

[Jerry Meyerhoff]

John & David (Hello, yes we're still alive !),
At MOTO many called the VCO application a "super filter" and cloaked it in mystery & reverence. I'm trying to recall its initiator. Could have been Bob Loving who had a knack for making truly effective circuits. Others might have called it a "Beta Multiplier" because crudely the effective capacitor value was roughly multiplied by the transistor Beta or current gain. At one point I think the circuit was integrated into a "wrap-around" ASIC customized to do the PLL frequency phase detector and/or some quasi-analog functions of a typical PLL frequency synthesizer ? But I think it's useful bandwidth was mostly in the base station AC power supply hum region. There was still 1-over-f and flicker noise to contend with from multiple sources within the VCO itself. Ah, strolling down memory lane... jdm 2/1/2012

[John Dunn]

Hi, Jerry.

The origin of this thing seemed to be shrounded in mystery when I first came across it back around 1994. Nobody could remember who created it, but it's arrangement because quite widely known and widely used. It didn't behave as a capacitive multiplier though. Rather, it's a Sallen-Key filter element turned sideways to be used as a series impedance versus the shunt impedance of the capacitor going to ground. The two impedances form a current divider.

This thing can be extended to three poles and more too, at least according to theory. I ran some promising looking simulations, but I never tried to build the hardware for that.

[Dave Peter]

John,
Thanks for posting this stuff, it's interesting and makes me think. Is there a reason this would be preferable to something like a LDO regulator with decent bandwidth?