Назад: 11: Power Supply Design
Дальше: 13: Liquid Cooling the Eimac 4CX250 Valves
12: Protection Circuitry
If you are contemplating building an amplifier you may wish to include some or all
of the protection circuits detailed below. If you are already the owner of a commercial amplifier you may also wish to make some changes to your amplifier to improve the reliability and life of the valves. Some commercial amplifiers currently on the market only offer the bare bones and often, due to cost constraints, omit adequate protection circuitry.
You may wonder why I should even mention filament (heater) protection, but we have all witnessed an electric lamp failing. Usually this occurs when switching on from cold: the filament breaks due to the high starting current. Filaments in valves suffer from the same problem when they are getting old. The severe thermal shock when switched on from cold can cause them to break. Let us look at this in more detail. A tungsten lamp filament runs at a very high temperature: it is white hot. Over time this very high temperature causes the filament to lose some of the metal in evaporation. Consequently it becomes thinner and eventually it cannot withstand the punishment and breaks at switch on. Valves can do exactly the same. High power transmitting valves for broadcast transmitters running high power have the same problem and the way that broadcasters handle the problem is a good method of getting maximum life from the valve.
Black Heat Operation
Valves used in high power broadcasting transmitters are very expensive and anything that can be done to prolong their life is advantageous. When a broadcasting transmitter is not required to be in service, but will be required to be brought into service soon, they use a technique known as black heat operation. With this technique the high voltage is switched off but the filament is not. It is operated at a much lower temperature by switching the filament voltage down to about 33% of its normal value. The filament remains hot but does not glow brightly, hence the term black heat operation. This helps to maintain a high vacuum in the valve as any molecules of air leaking in get converted to matter on the cathode, a permanent getter situation.
When the transmitter is needed to return to service the filament voltage is gradually increased to the operating level and the high voltage is turned on. This reduces the thermal stress on the filament. When a large valve is initially turned on from cold the filament voltage is slowly increased from zero in small steps to gradually bring it up to working temperature. It is allowed to bake for a short time before the high voltage is turned on.
The reason an electric lamp fails at switch on is because the filament resistance when cold is a small percentage of that when it is at operating temperature. Hence, it experiences a very high current before it starts to heat up. Typically the starting current – called the inrush current – can be as high as 20 times the normal running current. If the winding on the transformer can supply this very high current the inrush current will be fairly close to the theoretical maximum. Most low voltage filament windings can supply a current of 10 times the rated current for a short time, with only a small drop in voltage. Therefore it is prudent to connect an additional current limiting resistor in series with the filament at switch on and to short it out when the filament is near full temperature. This technique is called soft starting and is commonly used for large filament currents. It also takes the stress off the filament winding.
Prolonging Filament Life
All the valve manufacturers have issued engineering notes on how to prolong filament life. The most important factor is the current and hence the filament voltage. Mullard, RCA and Eimac – to name just a few manufacturers – all state that the filament voltage must be held within 2.5% of the nominal value to ensure adequate filament life. If the voltage is above the 2.5% limit, rapid deterioration occurs and as little as an extra 1% above the upper limit causes about 25% shortening of the filament life. For 2% extra it drops to ~50% valve life. Lower filament voltage is also damaging as it causes erosion of the cathode material when high cathode currents are flowing.
Eimac gives a comprehensive method to establish the optimum voltage for large valves. In this method the filament transformer is supplied with a variable AC input that can be varied in small increments, for example using a Variac transformer. The transmitter is run at full carrier power with the nominal filament voltage applied. The filament voltage is slowly reduced until the power output drops by 2 to 3%. This is the lower limit of the filament voltage acceptable. They then recommend that the filament voltage be increased by 0.25V as being the optimum value for that particular valve.
The filament voltages of some commercial amateur amplifiers are not well controlled. Often it is just a secondary winding on the main transformer supplying the high voltage rectifiers. A far better method is to use a separate filament transformer so that the input taps can be adjusted to give the correct filament voltage. This also allows other protection circuitry to be utilised to protect other parts of the amplifier. We will discuss these later.
The filament voltage should be measured directly at the valve socket when powered up to confirm the voltage is within the 2.5% specification. When high filament currents are used substantial voltage drops can occur in the wiring from the transformer to the valve sockets. If the filament winding voltage is measured with the valves unplugged expect to see at least a 10% increase of the voltage!
Inrush Current Protection
To protect the filament from excessive inrush current, a low value resistor connected in series is the simplest solution. For the 811 and 572B valve the filament current is 4A at a voltage of 6.3V. If a 0.1W resistor is used an additional voltage drop of 0.4V will occur. The resistor can be the chassis wiring, by deliberately using higher resistance wire from the transformer to the valve socket to adjust the voltage.
When at full temperature the 811 and 572B filament exhibits a resistance of 1.575Ω, when cold it is about 0.1Ω. Hence, it tries to draw ~63A at switch on. The extra series resistor limits the peak inrush current to about half the value. Both of these valves, as well as the 3-500Z, are classed as quick heat filaments and as such experience a rapid change in filament temperature. For multiple valve amplifiers the best option would be to have a separate filament winding for each valve, with each winding having only enough capability to supply the required running current. This means a flux-limited transformer should be used similar to a welding or battery charging transformer design. The inrush current demand should hold down the voltage until the filament temperature rises, so giving a gentle soft start to the filament.
For valves such as the 6146 and the 4CX250 series a better option is to wire the filaments in series. This means the secondary winding needs to be double the valve filament voltage, but it is inherently a better method as the two act as the additional dropper resistor to limit the inrush current. A further small value series resistor helps to slow down the filament heat-up period.
Unfortunately, with the directly heated filaments used in the 811, 572B and the 3-500Z this isn’t possible as the filament is also the cathode and they have to be connected in parallel due to the grounded grid topology. But for other types with indirectly heated cathodes it is the best method.
Stabilised Filament Supply
A circuit I developed for the QQV06-40 and the 4CX250 series provides a very stable filament voltage and incorporates inrush current limiting. The time to reach full temperature without inrush current limiting is about 2 seconds, but with the stabiliser it is about 5 seconds. The stabiliser requires a ~3A supply of 12V for the 2 x 4CX250 and 12.6V for the QQV06-40 when the filaments are wired in series. The raw DC supply is provided by two 6.3V AC windings of 3A connected in series. The basic circuit is shown in
Fig 12.1. It will hold the filament voltage within 0.1% over temperature.
Fig 12.1: Stabilised DC filament supply.
The stabiliser uses a common audio amplifier IC as the controlling element, the TDA-2003. This is a 741 op-amp with a 3.5A output capability. To relieve the power dissipation in the IC, an additional series pass transistor is used, a
TIP-41C which has a 6A collector current rating and so will handle the current easily. The reference voltage uses a red LED as this is a stable 2V Zener and has the required temperature compensated characteristic. In series with the output is a 0.22Ω resistor to limit the current output of the supply.
The raw DC supply is shown in Fig 12.2. It uses 1N5822 3A Schottky rectifiers to reduce the diode forward voltage drop.
Fig 12.2: Unregulated DC supply for the filament stabiliser.
High voltage supply protection
In earlier days, when equipment only used valves, this topic didn’t apply. But today the use of silicon rectifiers causes major headaches! In older equipment the high voltage rectifiers were valves and when the equipment was first switched on the filaments were cold. The rectifiers didn’t begin to supply any high voltage until they heated up, in unison with the other valves. So the high voltage gradually rose from zero in a slow and controlled manner. Today, with high voltage silicon rectifiers, the high voltage comes on with a ‘bang!’ And that can also cause some spectacular bangs in the valves and the power supply, with the emission of much so-called ‘magic smoke’!
All valve manufacturers state that it is good practice not to switch on the high voltage until the valve filaments have had a chance to reach operating temperature. Switching on with a cold filament can cause ionisation of any remnants of gas within the valve. If the filament is allowed to idle for about 30 seconds to one minute at full temperature this is less likely to be a problem. Consequently, some of the older commercial amateur amplifiers had time delay circuits to prevent the high voltage transformer being activated until the delay time had elapsed. This of course means the filament transformer needs to be separate from the high voltage transformer so it can be switched on first. Some amplifiers used a thermal switch in series with the high voltage secondary winding, so a single transformer could be used. These ultimately gave trouble when switching several kV of AC into a rectifier stack.
In an effort to cut costs these days this method has gone out of the window and there is often only one transformer with all the windings needed.
Silicon Rectifiers – the Bad Points
Switching on a high voltage transformer when silicon rectifiers are used means that the high voltage rises almost instantly and the inrush current in the transformer, diodes and smoothing capacitors can reach very high currents for a brief time. If you happen to catch the AC mains just at the peak of the waveform the current spike can be severe. Diodes and smoothing capacitors, as well as the main on / off switch, can react adversely to this punishment by failing, usually in a short circuit mode. If the mains input fuse doesn’t blow quickly extensive damage can be caused.
A much better way is to use a delay timer to hold off the switch-on of the high voltage and to apply some surge limiting at the AC input. This is a type of soft-start and using it will greatly prolong the life of the components. If this isn’t possible a ‘glitch resistor’ should be connected in series with both the high voltage secondary winding between the rectifiers and the smoothing capacitors to limit the current surge to sane values. Another resistor in series with the high voltage output acts as a sacrificial fuse should a valve decide to go short circuit. High voltage fusing of DC is a difficult topic as although the fuse ruptures an arc can develop across the fuse due to the spattered metal. DC arcs are very difficult to extinguish! AC fusing is simpler as the voltage drops to zero every half cycle so it is more easily extinguished should an arc develop.
Using a full soft start method delays the high voltage switch-on and limits the inrush current. To incorporate all this isn’t a great deal of effort compared to what it can save you in blown components and stress on the valves.
Simple High Voltage Inrush Current Limit
For an amplifier which uses the same transformer to supply the filaments as well as the high voltage then the circuit below at least protects the high voltage bits, if not providing the filament delay. It can be incorporated into most commercial amplifiers without too much effort.
A typical amplifier input circuit is shown in Fig 12.3. The live (line) input from the mains lead passes via a suitable fuse and then via the main on/off switch. It then passes via a resistor R1 to limit the inrush current at switch on. Across R1 is a relay contact (RY1) rated to carry the envisaged input current. This supplies the transformer primary winding. In series with the secondary winding is another resistor, R2, that limits the peak surge current into the rectifiers.
Fig 12.3: Input circuit of power supply.
At switch-on the series resistor holds the surge current down to sane proportions; with a 10Ω resistor and a 240V supply it is 24A RMS worst case. When the high voltage across the smoothing capacitors rises to beyond a certain value relay RY1 is energised, so shorting out the resistor. This is performed by the circuit shown in Fig 12.4, which shows the additional bits that make the soft-start work.
The bottom capacitor in the string of electrolytic capacitors will rise to a certain voltage when fully charged up. In this example it will be about 375V across a 450V capacitor. The potential divider of R3 and R4 reduce this to about 34V. Zener diode D1 will not conduct until the voltage exceeds 24V. This voltage is about two-thirds of the final voltage. Hence, when the voltage rises above this value it turns on TR1, which pulls in RY1, so shorting out the input current limiting resistor. This occurs in about half a second from switch on. As long as D1 is conducting the current limit is not operational and full power is available to drive the rectifiers. TR1 is any small NPN transistor with a reasonable current gain, such as a 2N2222 or equivalent. The 12V relay supply is obtained from a low voltage winding on the main transformer used to drive the antenna relays etc.
Additional glitch or current limiting resistors are also provided to limit the surge current into the smoothing capacitor bank and to feed the high voltage to the valves.
The values shown in Fig 12.4 show the principle and the exact values needed will vary from type to type. It is simple to change the trip point by changing resistor values or the Zener voltage to suit your application. The trip point must not be set too high as the high voltage can drop on peak anode current loads: two-thirds value seems to work for most cases. If a short circuit capacitor, or some other failure, stops the circuit tripping, the input resistor across RY1 will burn up and go open circuit, a sort of sacrificial fuse.
Fig 12.4: Voltage sensing circuitry.
Regulated high voltage supply
Earlier a reference was made to a regulated high voltage supply. For semiconductor applications it is the norm to use a series regulated supply to provide the collector voltage. The norm is for a 13.8V DC supply with the ability to supply many amps of current. The simplest method is to use a conventional transformer with a step down secondary, a bridge rectifier and reservoir capacitor to supply an unregulated supply to the series pass transistors. Generally, this method performs the regulation at the positive side of the supply (high-side regulation) as the voltages involved are not too high. However, sometimes it is more convenient to use a different method, called low-side regulation. If a high negative voltage regulated supply is required this becomes tricky with a conventional high-side regulator.
Fig 12.5: High-side and low-side positive voltage regulators.
If Fig 12.5 is examined we can see the difference between the two types. In practice, both use the same principle: an unregulated DC supply and a variable resistance element to vary the output voltage. The regulator doesn’t know which of the two terminals is connected to system ground, it can be either the negative or the positive terminal. All the regulator does is to maintain a constant voltage between the +ve and the –ve terminal. If the top diagram is considered, if the negative terminal is grounded it becomes a positive voltage regulated supply with the control element in the high-side line. However, if the positive terminal is connected to system ground it becomes a low-side regulator with a regulated negative voltage output.
Additional glitch resistor fitted to a Heathkit
SB-230 amplifier (photo: Rad, ZS6RAD).
The control element VR can be several different methods, either a bipolar trans-istor or a MOSFET with the base / gate driven by the control circuitry. The device chosen needs to withstand the extra off-load rectified voltage and a rating of 500V would suit most applications. The current demand is quite low, a 2A or greater device would be adequate. The device would need to be well heatsinked to dissipate the power. A 500V switch-mode MOSFET would require very little drive power to make the circuit work. Using a circuit such as Fig 12.5 the output voltage can be held to better than 0.1% regulation.
Unfortunately such a circuit cannot always be easily incorporated into the Delon doubler, as the no load to full load voltage variation is often quite high and suitable low cost semiconductors to withstand the extra voltage are rare. In this case the best method would be to control the AC mains input to the transformer primary winding using a Triac regulator with a zero crossing detector circuit. As an example of what can be achieved, in Fig 12.6 an ~800V stabilised supply using a low cost high voltage transistor as the regulating device is shown. For equipment where the valve rectifiers have been replaced with semiconductor diodes this addition allows the extra voltage developed to be held below the capacitor maximum rated voltage.
Fig 12.6: High voltage regulated supply.
The circuit shown is a bi-phase rectifier, but a bridge rectifier can be used with the same method. The secondary voltage of the transformer would be selected to supply a little extra voltage than normal; about 2.5% would be adequate. The sample voltage is taken from the mid-point using the capacitor sharing resistors to perform a times two divider. This voltage is at about 425V above ground. The final potential divider is performed by R3 / R4 connected in series to obtain sufficient dissipation using 1W resistors. It could also be a single 220kΩ of 2W rating. The series pass control transistor TR1 is a BUT-11A in a TO-220 package. This device can withstand up to ~500V collector-emitter at 4A of current. The collector tab is at ground potential, bolted directly to ground, and this supplies sufficient heat sinking.
Resistor RS is a surge-limiting resistor, which could be placed in series with a suitable fuse if desired. The other normal inrush current limiting and diode sharing resistors are not shown for simplicity. Resistor R11 is a 5W wire-wound type to set the minimum voltage should the regulator malfunction.
The emitter of TR1 is lifted up by D4, which supplies a negative bias voltage. (D4 needs to be a high voltage diode.) The sampled HT voltage is used by TR4 and TR3 connected as an emitter coupled pair to supply base drive to TR2. The BUT-11A has a current gain of about 40 for small currents. For low current loads TR1 is partially off, supplying just sufficient current to the capacitor reservoir to maintain a constant voltage. When the load current increases the sampled voltage falls slightly so turning on TR1 harder to supply more current.
A 9V DC supply is obtained by a half-wave rectifier using a 6.3V heater winding. This supplies the low power regulator circuitry. A 5.6V Zener supplies the reference voltage to TR3. Alternatives for the BUT-11A are the MJE-13007 and the BU-505 which are all TO-200 packages with adequate voltage rating.
The method to set up the supply is to short R11 by a jumper wire temporarily and to measure the off-load voltage across the supply. When the supply is loaded to full output the voltage will fall by some amount. Suppose it is 900V off load and 800V at full load. To prevent voltage variations the adjustment should be so that it supplies a little less than 800V with no load. Using this simple circuit allows regulation within a volt or two between no load and full load. It allows transformers with poor regulation to be pressed into service for intermittent high peaks of output current. It also means that the value of the reservoir capacitor can be reduced from the normal types of supply. Unlike high-side regulation there is ample scope to extend this method to supplies of several kilovolts output with suitable control transistors.
Fig 12.7 shows the bridge rectifier connection details.
Fig 12.7: Regulated high voltage supply using bridge rectifier.
Medium voltage regulated supply
For a simple lower voltage regulated supply a form of the bucket regulator is often adequate. In this topology the regulation is done at the high-side as the voltage is not too high. The regulation is not as good as the previous versions, but it can hold the output voltage within ~2%, which is better than an unregulated type of supply can achieve. Fig 12.8 shows the basic circuit.
Fig 12.8: Medium voltage ‘bucket regulator’.
The transformer secondary uses either a bi-phase or a bridge rectifier to produce an unsmoothed DC output. The regulator is a silicon-controlled rectifier (SCR) which is used as a series pass element. The reservoir capacitor is placed after the SCR and it receives current pulses to charge up the capacitor. The gate pin of the SCR is also supplied with rectified pulses and when the peak of the pulse exceeds the cathode potential the SCR conducts for each pulse of input signal. The voltage at which it conducts is determined by the regulator formed by TR1. This samples the output voltage and when it exceeds the required voltage it pulls down the gate of the SCR via R4, so terminating the switching. TR1 is a 200V rated device in a TO92 package.
The SCR shown (BT-151/800) has a maximum rated voltage of 800V and can pass about 7.5A, so it is not stressed by a supply of 250mA at up to ~500V. The
BT-151 series are available in 500V, 650V and 800V versions. (The BT-152 series can handle more current and are available in 450V, 650V and 800V versions). The transformer secondary voltage only needs to be a little more than the required maximum output voltage, a 350-0-350 winding would cater for a ~500V DC supply. The SCR will require insulating from the heat sink with an elastomer or mica washer with an insulating bush. The output voltage sampling resistor (R5) needs to be calculated for the required output voltage range. With the values shown the minimum voltage is ~300V. The various resistors need to be adequately rated for the voltage and power dissipation. In most cases 1W or 2W resistors would be required. Diodes D1 to D4 are 1N4007. D3, D4 prevents the SCR gate being driven too positive. Zener diode D5 can be a 500mW rating. Reservoir capacitors C1 and C2 can be 350V rating. R7 and C4 are RFI suppression components.
Inevitably valves wear out or fail for some other reason. When this occurs replacement is a necessary chore. However, unlike transistors, it is a simple operation because they are plugged into valve bases and simply pull out. But there are a few important points to be aware of when fitting new valves.
The replacement valve may be from NOS (New Old Stock) and have been in storage for some considerable time. Even valves recently manufactured can be a problem. The reason is that the high vacuum within them tends to suck in tiny amounts of air. It is virtually impossible to make a perfect hermetic seal between the pins and the glass envelope. If the valve has been in storage for a long time it is almost guaranteed to have ingested a few molecules of air. The danger is that if this valve is fitted and the high voltage supply is simply switched on it can cause ionisation to occur in the valve. To prevent this potentially catastrophic failure precautions need to be taken.
The way to condition a new valve is to remove all the high voltage supplies and then power it up with just the filaments and any cooling fan running. The valve needs to bake for a length of time to allow the hot cathode to find any air molecules and convert them into material on the cathode. Depending on the cathode construction this may be as short a time as 30 minutes up to several days. The longer the cathode has to eliminate the ingested air the better. Some high power broadcast transmitting valves are continuously pumped to keep the vacuum hard.
Some of the later valves have an active getter, which is a coating on the anode that continues to absorb any air molecules and prevent ionisation. The Svetlana 572B valve has a titanium coating on the graphite anode to perform this function.
- 1. Notes on Terminology
- 2. Safety When Working with Valves
- 3. 1: Introduction
- 4. 2: Basic Valve Theory
- 5. 3: Working with Valve Characteristic Curves
- 6. 4: The Class A Amplifier
- 7. 5: Operating Modes
- 8. 6: Causes of Distortion
- 9. 7: Measuring Amplifier Linearity
- 10. 8: Designing an Anode Tank Circuit
- 11. 9: Grounded Cathode Linear Amplifiers
- 12. 10: The Grounded Grid Amplifier
- 13. 11: Power Supply Design
- 14. 12: Protection Circuitry
- 15. 13: Liquid Cooling the Eimac 4CX250 Valves
- 16. 14: Purchasing an Amplifier
- 17. 15: Notes on RF Linear Amplifier Operation
- 18. 16: VHF RF Power Amplifiers