Valve (Tube) High Voltage Time Delay

There is often a need for the application of a valve amp's high voltage (commonly known as B+) supply to be delayed until the cathodes are up to temperature. In some cases, this may prolong the life of the valves, although hard evidence for this is hard to come by. I recently had to provide just such a delay circuit for a number of 4-way 200W/ channel valve power amps that were having regular valve failures for no apparent reason.

The delay circuits I made up were not as elegant as the circuit shown here, but they work well. At the time of writing it's been several months without a valve failure, so I remain hopeful that adding the delay has indeed improved matters.

There are a few circuits shown on the Net for delay circuits. I've not looked at any in great detail, but I'd expect them to work well. One I saw used a high voltage MOSFET to perform switching. This is a good scheme, but it requires a dedicated 6.3V winding on the transformer because the circuit and its supply will 'float' at the full DC voltage during the delay period. Another uses photo-voltaic optocouplers and a MOSFET. This solves the problem with the 6.3V winding, but the optos may be difficult for some people to obtain.

While using a MOSFET is certainly a good solution, it does place a semiconductor in the high voltage circuit. You won't hear it (regardless of what some might claim), but a surge or spike on the mains may cause failure unless precautions are taken. I've chosen to use a relay for switching, and despite initial concerns there's no reason that the contacts won't outlast the amplifier.

The circuit shown uses a 4000 series CMOS counter, and you cannot use a 74 or 74HC series device because they will not tolerate 12V. Naturally you could reduce the supply to 5V, but then you'd need to use a switching MOSFET that's guaranteed to turn on fully with 5V on the gate. You'll also need to use a relay with a 5V coil which might be a harder to find (and it will draw a higher current). I leave the necessary changes to the constructor if you want to use a 5V circuit.

The delay circuit is shown below. It uses a 4020 CMOS counter, which is reset at power on by C3 and R3. After the reset period (about 50ms), the circuit counts AC cycles obtained via R2 until the appropriate output goes high. When that happens, the counter is stopped and the relay is energised. D5 is used to stop the counter, which is done by forcing the clock input to the counter high. When the relay contacts close the AC is connected to the rectifier.

The time delay is set by selecting the appropriate output from U1. I have not shown intermediate timings which can be selected by using an 'AND' gate if you really need to set the time to something specific (see below for more info). As shown, the timings are as follows ...

Delay	50Hz	60Hz
Q9	10.25 seconds	8.5 seconds
Q10 - 1 (Pin 14)	20.5 s	17 s
Q11 - 2 (Pin 15)	41 s	34 s
Q12 - 3 (Pin 1)	82 s (1m 22s)	68 s (1m 8s)
Q13 - 4 (Pin 2)	164 s (2m 44s)	136 s (2m 16s)

Table 1 - CMOS Output Timings (Default In Grey)

The delay you use depends on many variables, but it's not very likely that you'll need the long delay unless your valves are particularly fussy or you have good reason to be paranoid based on past experience. As you can see from the table, in the simple form shown each output simply doubles the time delay created by the one before. It's very rare that you really need a specific time, and for most output valves a delay of 82 seconds (50Hz) or 68 seconds (60Hz) will be perfect (U1, Pin 1).

If you want to, you can include an AND gate so that intermediate values can be obtained, but I wouldn't bother. One situation where you might want to include the AND gate is if you need more time than the circuit can provide as shown. For example, if you gate the three outputs I included ( '2' AND '3' AND '4' ) you can increase the delay to 164 + 82 + 41 seconds - a total delay of 287 seconds (4 minutes and 47 seconds at 50Hz).

If you do decide that you need an intermediate or extended delay, you can use a 4082 dual 4-input AND gate to sum the outputs as desired. Unused inputs to the AND gate(s) must be tied to the positive supply (cathode of D3 zener). You can work out the effective delay for any output from the delays described above. If you use an output that's one less than one of those listed in the table, the delay is half. For example, the Q8 output has a delay of about 5 seconds at 50Hz (4s at 60Hz). Very short delays are probably not useful.

R4 in series with the relay is to limit the voltage to the coil. You can normally expect a DC voltage of about 15V (the rectifier is a voltage doubler), so once you know the relay coil current it's easy to determine the resistance needed for R4 using Ohm's law. If the 12V relay coil measures 250 ohms and you have 15V at point 'A', then we get ...

With a fairly typical relay having a 12V coil and needing around 50mA coil current, you can expect the circuit to draw about 400-500mA from your 6.3V supply. Most amplifiers will have sufficient reserve on the 6.3V winding because they are typically rated at 3A or more, and some have two or more windings so you can use the one with the lowest current drain. If the worst happens and you can't spare 0.5A or so, then the circuit can be powered from an auxiliary transformer. You can also use a 5V winding if you have one spare, and the voltage will be around 12V. R4 will not be needed and R1 should be reduced to around 220 ohms or less. Do not omit the zeners!

The circuit shown is only one of many possibilities of course. The time delay circuit shown in Project 167 is an alternative timer that could also be used, based on a 555 timer. These are often not the best choice for long time delays though, because you need an electrolytic capacitor and a high value resistor for the timing. This combination is always somewhat variable, and much more so in a chassis where heat is present. Valve amps always run at elevated temperatures, so the time delay will not be entirely predictable.

Although the relay you use will not be rated for 400V AC or more, that doesn't actually matter. The relay will never have to break the circuit while voltage is applied, so it won't arc. A standard 250V relay with two sets of contacts in series is rated for 500V, but even a single pair of 250V contacts can be used. Once the relay energises, it stays energised until the power to the amplifier is switched off. At that time, the AC from the transformer isn't there any more so the relay breaks a 'dry' circuit - one where there is no current flowing. Only the high voltage connections from the power transformer are shown. Heater, auxiliary and tapped primary windings aren't included for clarity. The circuit will also work with valve rectifiers, but there will be a high surge current because the heaters/ filaments will be at full temperature when the relay closes.

I can't recommend valve rectifiers for anything, and it's always better if you use silicon diodes (usually with series resistors). In most cases, the effective plate resistance of a valve rectifier is around 75Ω up to 400Ω (see the datasheet for the diodes you use). If the series resistors are omitted, you may get a higher voltage than expected, and their inclusion mitigates this. Some people like to include capacitors in parallel with silicon diodes, but that's usually superfluous. If the rapid turn-on of silicon diodes presents a problem, a simple 'soft-start' circuit can be employed. It usually won't be necessary to go that far, as the HT windings generally have fairly high resistance that minimises current 'surges'.

A DC soft-start delay can use the same power supply as the main delay, and can use another identical delay circuit (with the take-off point the next highest delay from the 4020 of 'delay 1'). For example, if you use Q12 (Pin 1) for the main delay, use Q13 (Pin 2) on the second 4020 delay circuit, with the relay shorting out a suitable limiting resistor. This would typically be around 1k, 10W. The two delay circuits can share the same reset circuit, but must be clocked separately so the clock is inhibited separately for each delay circuit. I leave the details to the constructor.

If your HV winding is centre-tapped (full wave rectifier) the relay contacts will have a pulsating DC voltage across them when open. For example, if the windings are 2x300V AC, the peak voltage across the contacts will be about 600V. While this might seem like a recipe for disaster, the contacts won't arc because they are open. There will be a current 'surge' as the contacts close (which will happen with both circuits) and there'll also be a small arc due to contact bounce. However, as noted above, the relay will never have to open with any voltage across the contacts because they only open after the mains is turned off.

You need to ensure that all HV wiring uses cable with insulation intended for the voltages used. If you are used to working with valve amps then you'll already know this, and you will also be aware that the high voltages used are potentially lethal. The relay you use must be designed to operate with at least 250V AC across the contacts. As noted, the relay will never have to break the high voltage, so a 250V relay will be adequate for HV transformers with an output of 300V AC or more. You may elect to use a relay rated for the full voltage if you prefer, but an 'ordinary' double-pole relay with the contacts in series is cheaper and much easier to get.

As far as I'm aware, there's very little real evidence for claims that allowing the cathodes to reach full operating temperature before DC is applied makes any significant difference, but valves are certainly not what they used to be. The system mentioned in the intro had many valve failures, and now seems to be stable after the addition of a delay circuit. However, this isn't evidence per se but merely an anecdote at this stage. Real evidence can only be gained from lab tests and statistical analysis, something I'm not in a position to spend time on.

Adding the delay won't hurt anything though, and that's the main thing. In the heyday of valve equipment amps used valve rectifiers, and these provided some delay as a matter of course because it took time for the (heavy duty) cathode(s) to get to temperature. It has to be considered that these valves had the full AC voltage across them before the cathode(s) were hot, so they should have shown failures too.

There's probably not much point worrying whether a delay should (or should not) be used, it's simply an add-on circuit that might help prolong the life of your expensive output valves. As noted, it can do no harm, and since it's not expensive or difficult to add it may be worth including. It's easily removed if you decide that you don't want it, or you can simply reduce the delay to 10 seconds which will not be noticed. If (perish the thought) I were to build another valve amp, I would certainly include the delay because it can only help.

High Voltage DC Time Delay