No R, No C, No Batteries, No Hum
In the previous article, I described a small SE amplifier that produced a couple watts and used no resistors or capacitors (and no batteries, solid state devices etc). Article can be gotten here. This also included the power supply. In this article, we will describe a regulated power supply that can be used with that amplifier that produces a well filtered low impedance set of voltages. This supply also uses no resistors and capacitors. That is, the described power supply consists of ONLY transformers, inductors and tubes.
250VDC at a load of up to 100mA.
Approximately 130 ohm output resistance (resistive).
Line and load regulated. (Output remains within the range 237 to 255 volts for any combination of load 0 to 100mA, and line from -5% of nominal mains to +8% of nominal mains voltage.)
Approximately 40 mV output ripple.
Also produces 75VDC at up to about 15mA, with about 2 mV ripple.
Also provides 6.3VAC for heaters at 1.7 amps.
Actually, this is a rather simple circuit. There are two sections of filter, consisting of an inductor followed by a shunt regulator. The shunt regulator uses VR tubes driving the screen (so the current through the VR tubes is only screen current: 5 to 20mA at any line/load combination). The first section produces produces about 265 to 275 volts. Approximately 75 mA current at nominal line flows into the first 6KG6. The worst case dissipation in this tube occurs with no load and high line, and is about 33 watts, within rating for that tube. There is about 5 volts of ripple at this stage's output. That actually comes in handy to insure the VR tubes ignite. This will be discussed below. The first stage also provides the primary "line" regulation. The current in that stage varies from about 120mA at high line to about 70mA at low line. It is important NOT to use too low resistance choke and DO use tube rectification: See below.
The second stage filter consists of a second inductor and another 6KG6 shunt regulator, set to operate at a lower voltage. The second stage produces about 250V and reduces the ripple to about 40mV. About 40mA flows through the second shunt regulator tube. However, this current does vary quite a bit depending on the load current. At nominal line and no load, about 65mA flows in the tube whereas at 100mA load, only 16mA flows in the tube.
For this circuit, I used a 5R4. The reason was this device has sufficiently high effective resistance so the regulator tubes don't get too hot. If you consistently find the voltage to be too low, you can replace it with a 5U4. In the 100mA range, the effective resistance of a 5R4 is about 226 ohms. A 5U4 offers about 175 ohms, and a 5AR4/GZ34 offers about 55 ohms! A 5Y3 offers about 310 ohms, by the way.
Auxilliary +75V Output
In the amplifier this supply was developed for, the first stage ran from 75 volts at about 10 mA. I've used a 6CB6 as a constant current source driving the 0C2 used to supply the 75 volts. This eliminates the need for any resistors here as well. It also provides a good ripple reduction, so the hum level at the 75 volt point is less than 2 mV!
So how do the VR tubes light?
When you first turn on the supply, the voltage on the first regulator rises so that the first set of VR tubes ignite. Since there is yet no load from the second stage, the hum level is about 20 volts. As the current in the second stage builds up (remember, in an inductive circuit, current cannot change instantaneously), the AC ripple allows the second stage regulator tubes to ignite, causing regulation. Then the voltage at the first stage drops a little, but since THOSE tubes are already lit, they remain in conduction, so long as there is enough resistance in the second choke. The ripple at the first stage drops to about 5 volts (from the load of the second stage and the current flowing in the first shunt regulator stage).
An L-R filter (which is how you describe this kind of filter) produces lowest ripple at the lowest load resistance. For that reason, shunt regulators work well, but series pass regulators won't work at all in this application. (For light or no load conditions, even dynamically, the ripple increases very much, making the supply useless). This kind of filter would be quite impractical for high voltages and high currents.
Since a reasonably high resistance INTO the first regulator is needed, solid state rectifiers, or even 5AR4 or "damper diodes" would not work well: the current at high line would be too high.
With the circuit shown, all devices are run well within their ratings, at any combination of line and load.
It may be desireable to have several VR tubes available, in case the first ones you select are at the "limits" of their ratings. For example, 0A2 nominally operates at 150 volts, but according to the RCA book, individual tubes can operate anywhere from 140 to 168 volts. Most of the samples I have run between 148 anf 154 volts, but I do have 1 0A2 that wants to regulate at 163 volts. If the voltages are far from nominal, either the first or second regulator will "hog" current.
Speaking of current, how does one monitor the current flowing? Measure the chokes resistance (unpowered), and measure the voltage drop across each choke.
So how does the amplifier sound now?
Combined with this circuit, the original R/C less amplifier is quiet! (Less than 1mV of residual at the speaker). Sounds quite good. Has to be experienced rather than described.
What's next? Hmmm, how about a No R No C push pull amp?