Part 2 Simple regulators with line and load regulation.

In this section, we will discuss the addition of a voltage reference tube to our one tube voltage regulators. This allows the regulator to provide "line" as well as "load" regulation. This is perhaps the simplest combination that offers performance expected of a voltage regulator: namely, the voltage output is constant.

There are some new concepts introduced in part 2:

The same basic "unregulated" power supply is used throughout this section. It is the same one used in part 1:

Voltage Reference Tubes

The idea to be conveyed here is a sort of self contained mini voltage regulator. These devices maintain a constant and fixed voltage over a reasonable range of currents. In the solid state world, these are ZENER DIODES. In the tube world, these are rare gas filled diodes with no filament. There are perhaps 3 important classes of these devices.

  1. Voltage regulator tubes. Examples are 0A2, 0A3, 0G3, VR-150. These devices provide a relatively constant voltage over a moderate range of currents, typically 5 to 30 or 40 mA. Voltage regulator tubes will typically change 2 to 4 volts over their operating range of currents, and the absolute voltage will change a few volts from device to device.
  2. Voltage reference tubes. Example 5651. These devices ore intended to be operated over a narrow range of currents and are designed to produce a constant and repeatable voltage that remains stable over time. These are usually used in feedback multiple control tube style regulators (in other words, complex regulators). They produce good references, but are not intended to be "regulators", as the operating current range is very narrow.
  3. Neon or Argon "light bulbs". Example NE2-H. These devices make relatively good low current regulators, except that they are not really characterized for this service. Voltage and regulation characteristics vary from device to device, and the "turn on" characteristic varies with ambient light intensity. Still, they do make cheap regulators, particularly if you're going to "adjust" the output voltage anyway. Be cautious: some neon lightbulbs have a built in resistor in series with the bulb, so that they can be directly connected to line. These don't regulate.

Unlike solid state zener diodes, these devices also require a higher voltage to "start" them than their operating voltage. For instance, an 0D3 regulates at 150 volts, but you may have to apply a current limited 185 volts to get it to start. Because of this characteristic, it is impractical to place a large capacitor across these devices, since in the best case, they will form a "relaxation oscillator", and in the worst case, explode if you do so. On the other hand, they are usually much quieter than the equivalent solid state zener, and have a better "temperature coefficient". [This latter means the voltage stays fairly constant as the device heats up. High voltage zeners will change quite a bit as they heat up].

Note that voltage regulator tubes are, in fact, shunt regulation devices. However, with a 5 to 40 mA current rating, these are appropriate for shunt regulators only at relatively low powers. In part 5, we will discuss using a shunt tube to form a higher power shunt regulator.

The way you use voltage regulator tubes is actually pretty easy: at nominal conditions, choose an operating current of something like 15-20mA. Then, choose a dropping resistor from the supply to the plate of the VR tube to supply this current. For instance, suppose the supply voltage was 450 volts when it produced 15mA. Suppose we want to regulate to 300 volts, using 2 0A2 VR tubes. Then, we would need to drop 15mA at 150 volts, for a resistor of 10k. If the input dropped to 400 volts, this current would decrease to 10mA. If the input increased to 500 volts, the current would increase to 20mA. All of these conditions are within rating for the 0A2. If they were not, we would need to juggle voltages and resistance values until they were. Just remember the supply must be high enough to light the tubes. In this example, at least 370 volts.

Operating "line" limits

In part 1, I mentioned how the output voltage changed with a change in line voltage. Those limits were +/- 20%. Now, that's a pretty big change. For a nominal line of 117V, 20% change is equivalent to a change from 94 volts to 140 volts (or from 188 volts to 280 volts). Probably a more "normal" change is +/- 10%, which is equivalent to a range from 105V to 129V. (or multiply by 2 for 240V mains)..

I will show a 10% variation in line in the circuits in this section, just to give you an indication of "what happens". When we get to a "real" example, we will probably design for +/-10% changes as well.

A simple triode regulator with line and load regulation.

This looks a lot like the "improved" triode regulator we discussed in part 1. But there is one important difference. In this circuit, I have added one resistor and 2 series connected voltage regulator tubes, to provide a 300 volt reference. This reference will stay about 300 volts as long as I can provide enough voltage to "light" the regulator tubes. The voltage divider setting the control grid voltage is thus made constant with line variation.

The regulation characteristics of this circuit are:

There are several things to note on this regulation picture. First, note that in addition to plotting both 200V and 250V "set" points on the same picture, I have also shown the line voltage variation. Also notice that the vertical scale has been expanded. This allows you to get an instant picture of all the conditions you will likely encounter, and the improved regulation allows an expanded scale to see more detail.

This second point is important. For the previous "improved triode" regulator, at 50mA draw, the voltage varied from 226V to 250V to 276V as the line changed from -10% to nominal to +10%. With this circuit, the change at 50mA is 239V to 254 volts! Because the voltage on the plate does vary, even maintaining a constant grid voltage allows the output voltage to vary somewhat.

The ripple from this supply was about 100mV. This improvement is the result of the additional filtering action of the regulator tubes. There is only a couple mV of ripple on the grid!

There is one additional item to note with this circuit. Notice that the "low line" 250v curve indicates less regulation at higher load currents. This is due to the tube "saturating". There is not enough plate voltage to maintain good regulation at the cathode. This is called regulator dropout. I'll discuss this in more detail later.

A Simple Pentode Regulator with Line and Load Regulation.

Similar to the pentode regulator we discussed in part 1, the voltage regulator tubes can be added to that circuit as:

Again, one resistor and the 2 VR tubes have been added. Note, however, that this time I'm also running the screen voltage from the regulated supply. Since there is some added screen current, I've lowered the dropping resistor from 8k to 5k, in order to supply this added current.

Here is the regulation characteristics that can be achieved:

This is actually quite good as a regulator. Notice that the stiff screen source and stiff control grid source allow essentially "perfect" regulation. [meaning that the output resistance is 1/gm. The curves indicate an output resistance of about 70 ohms!]. Also notice that the line regulation is also quite good,;it's about 0.5% for +/-10% line voltage change.

The output ripple was about 20mV, due to the plate isolating the output ripple from the LC filter, and the VR tubes providing low ripple to the screen and control grids.

There is one additional benefit of the pentode in this application. The tube can maintain regulation when the plate voltage is lower than the screen voltage. Notice the low line 250V case is much better (but not perfect... in fact, the imperfection here is not due to the plate "saturating", but rather the increase in screen current starting to pull the VR tubes out of regulation range.)  It is the lower available voltage drop of the pentode, along with its constant current characteristics that provide the dramatic improvement in performance in this simple circuit.

Again note that since there is no feedback conditions (other than cathode follower degeneration), there is no major "frequency response" issues with this regulator. It will "sound" pretty good. I'll introduce the concept of plotting the output impedance vs frequency when we get into regulators with defined feedback loops later in this series.

Note that this circuit could be turned into a variable output power supply. Simply replace R1 and R2 with a 100k to 500k pot. Do place about 10k in series with the point connecting to the VR tubes though, so that when the wiper is turned so that the capacitor on the grid connects towards the VR tubes, there is still 10k resistance isolating the VR tubes from the capacitor. Available output voltage effective control range is from about 50 volts to about 275 volts. We will discuss the available current below.

Supply Limits and Tube dissipation

The realistic limits on the regulated power supply are controlled not only by the power supply (transformer, rectifier, filter inductor) but also by the tube. The tube provides a limit in current and power. Lets initially assume the power supply components do not limit the circuit (they are big enough). Then, what limits the output voltage and current?

In the above circuit, the screen voltage is held at 300 volts. Thus for cathode (output) voltages approaching this value, there would not be enough conduction. A practical limit in this instance is probably 275 volts. Also a practical limit is also probably about 30 volts (anode to cathode) drop. For the rectifier and filter combination, this means the filter output would have to be 275+30=305 volts. This occurs at about 35 mA. At this point the tube is dissipating 30v*35mA= about 1 watt. At 275 volts, the maximum output current would be about 35 mA. If we content ourselves with 250V maximum, the screen to cathode voltage is 50 volts, so the minimum anode to cathode voltage might be about 25 volts. This occurs at about 120 mA, so 120 mA is the maximum available current at 250 volts. (The tube is dissipating 25v*120 mA which is less than a watt).

If we consider -10% line conditions, 275 volt output would be limited to about 20 mA, due to the filter output voltage (305V at 20 mA), and 250V output current would be limited to about 50 mA. Notice that the curves above confirm this analysis: at 250 volts, above 50 mA, the output voltage starts to drop. I should add, though that I'm ignoring screen current, which is not *quite* fair. In fact, the screen current from the "higher voltage" supply is attempting to provide about a third of the output current at 50 mA (i.e., 16 mA) under low line high voltage conditions, and this is starting to pull the VR tubes out of regulation.

At the low voltage end of our hypothetical variable supply, power dissipation and./or maximum tube current will limit the available output. Let's consider the case of 50 volts output. (Potentiometer turned towards the ground end). In this case, we have 250 volts screen to cathode... we can get lots of conduction. The plate voltage will be relatively high. At high line and 200mA, the filter output is about 290 volts. Thus, 240 volts and 200mA flows in the tube. (48 watts). This exceeds the 35 watt tube rating. In fact, working thru the math implies we can draw about 140 mA from the supply at 50 volts. At 100 volts, we could draw about 180 mA. At 150 volts we could draw about 240 mA. At 200 volts, we could draw almost 500mA. (At high line, the filter output produces about 270 volts at a 450 to 500 mA load. The output will have dropped to about 195 volts for 75 volts plate to cathode. For 35 watt dissipation, this is 466 mA) However, at 200 volt output and LOW line, the filter output will be about 220 volts at 400 mA, so about 400 mA is the limit at 200 volts.

Notice that in some of the conditions above, the limit was high line, while in other conditions, it was low line.

In the next part, we will talk about error amplifier tubes and demonstrate their effect on power supply regulator performance.

go to part 3