Minimal Reactance Power Supply

The Line Stage ended up being a parafeed design using a constant current source (CCS). Thinking about power supply topologies, I began to wonder if a choke was needed if the load drew a constant current. This thought opened the door and allowed me to get out of the box of conventional power supply thinking. The thoughts poured in – In my experience a CCS was better than a choke as a plate load. Was the CCS an ideal choke? Of course a CCS doesn’t store energy like a choke does, but so what? What if I were to use a CCS in the place of a choke in a power supply? Figure 19 shows the concept with CCS1 in the power supply and CCS2 in the Line Stage. In this example CCS1 provides a constant 35mA, CCS2 draws a constant 30mA, with the remaining 5mA flowing through the resistor.  The resistor is chosen to produce the desired voltage on the output of the power supply. I built a model of the design shown in Figure 20 and it worked just as I expected. My next question was “how big does the input capacitor need to be?” One of the important characteristics of a CCS is its power supply ripple rejection (PSRR). With a very high PSRR the CCS can tolerate a large ripple on its input, in fact the amount of ripple is not important as long as the instantaneous voltage stays above a certain value. In order for the CCS to provide a constant current there must be sufficient voltage on its input. If the voltage were to drop below this point, the CCS would act like a large resistor and the current would drop as the voltage dropped. It doesn’t matter if the voltage rises above the critical value; the output current remains constant. Although the DN2450 high voltage MOSFET that I use here needs only a few volts across it to work properly, I conservatively allow 20 Volts. So, say I wanted 200V on the output all I would need to do is keep the input above 220V. I could have a hundred Volts of ripple and as long as the instantaneous voltage stayed above 220V the CCS would reject all of the ripple. This means that the input capacitor does not need to be very large at all, perhaps only a couple of microfarads. But don’t you need a large capacitor to store lots of energy to provide for powerful bass and a large crescendo? Nope! Isn’t “the more the better” a good rule for capacitors? Nope! As long as the CCS input voltage stays above the critical value the output is the same. The current is constant even for those deep bass thumps and large crescendos. In fact a large capacitor is not necessarily a good thing.  Take a look at the ripple waveform in Figure 21. During the first part of the cycle the capacitor charges to the output voltage of the rectifier, while during the second part of the cycle the capacitor discharges until the rectifier voltage again rises to exceed the discharged voltage. As the value of the capacitor increases, the capacitor can store more energy so it looses less voltage during the discharge part of the cycle. As a result the charge time becomes shorter and the discharge time longer. With a very large capacitor the discharge time can be miniscule. However since the capacitor is large, it provides low impedance during this short charge time resulting in a large amplitude current pulse. Thus, the current drawn from the transformer is in the form of very short, high-amplitude current pulses. These pulses cause the transformer to ring like a bell. A smaller capacitor that draws a wider, lower amplitude current pulse has the advantage here. Another benefit of using a small capacitor is that it can be a high-quality film rather than an electrolytic resulting in even better sonic quality. The result of all this is a unique supply topology that regulates the current and lets the voltage seek whatever level is needed to provide the required current. Because the design uses as small a capacitor as needed and does not incorporate an inductor, I call it a Minimal Reactance Power Supply.

Let me share the complete design of the Line Stage Supply as shown in Figure 22. The first thing you will notice is that I have added a power-on-delay timer. Since the supply needs the load to participate in drawing current, it is important that the load be capable of drawing that current. The delay gives the tubes in the Line Stage a chance to warm up before applying the voltage to them. The output resistor is split to provide a bias to the 12V filament supply. Diodes D1 through D4 together with their associated components form the 12V supply used to power the timer as well as the Line Stage filaments. Common mode choke L1 along with capacitor C1 constitutes a power line filter and the high quality Lundahl LL1683 is used as a power transformer. A picture of the completed supply is shown in Photo1.

This power supply is an updated version of that featured in the May 2004 issue of audioXpress magazine. It now has a cascode constant current source and a full wave rectifier instead of the hybrid Graetz bridge shown in that article. The latest version is shown here and information about modifying the power supply for your specific requirements is here.