All-Tube Preamp Power Supply

For the love of God and all that is Holy, tube power supplies employ POTENTIALLY LETHAL VOLTAGES. Don't mess with this stuff unless you know what you are doing.

Based on Steve Bench’s tube regulated power supply schematic my all-tube preamp power supply board uses a EZ81 rectifier tube feeding a CLC Pi filter followed by 2 RC filters, which is then series regulated using an EL34 with voltage reference via a series pair of 0A2 cold cathode regulators. Reference values for 200V and 250V out are marked on the board, set by the value of R12. Outside of the heater supply section this design has but a single silicon diode in the voltage reference supply, everything else is hollow state!

I chose the best compromise of simplicity and performance. An error amplifier feedback circuit would improve things but the extra complexity didn’t appeal to me – instead I added additional filtering ahead of the circuit to clean up the incoming DC. Cleaner power in means cleaner power out!  I also disliked Steve’s choice of an expensive and hard to find EL509 for regulation when a common-as-the-cold EL34 can do the job with only minor modifications.

Additional film shunt capacitors are recommended between the power supply and your circuit for local filtering. I also recommend bypassing all the electrolytic caps with small film caps (0.01-0.1uf). I use Nichicon LKX for high voltage electrolytics, UKA/UKT 105 degree audio caps for the heater supply. Output ripple and noise at 200V (20ma draw) is approximately 2.5mv, regulation of +-0.1V. This design is nominally rated for 50ma at 250V (approximately 4-6 small signal tubes, depending on tube type and bias level). For example my CCS line stage and phono stage each draw approximately 20ma, so this design can run both simultaneously.

Required AC is centre-tapped 550VAC (275-0-275) of at least 100ma. A 2A 6.3VAC heater supply is required for the power supply board alone. Heater voltage is elevated by B+. I recommend Edcor transformers, the XPWR106 (550 CT 125ma, 6.3V 4A) fitting the bill precisely.  At least 350VDC raw voltage needs to be present at the first cap to light the 0A2 regulators.

A ground lifted dual channel 6V heater supply is also included on the board. Required AC is 9 to 12VAC (if you have a spare 12.6V tap on your power transformer use it here!) Nominal output is 2A (1A per regulator) but this may be increased to 3A by selecting heavier duty low noise 78XX regulators. Heat sinks are recommended on the Q1/Q2 regulators. The ground of this circuit has no reference to the ground of the rest of the board, it is fully isolated so you may elevate the heater ground as required by your application. The two HT+ outputs are independent and can be bridged or run in parallel.

By default the regulation is via a pair of 7805s with 1V diode drop to ground to output 6V; this arrangement allows you to tweak the output as desired or increase the voltage to compensate for subsequent filter stages (for example: I run 7V output then feed additional RC filter stages to reduce noise and ripple, these filters then drop the voltage down to 6V). If you use 7806 regulators you can omit D6/D7 D10/D11 and put a jumper across these slots to HT- ground.

Part II for Advanced Builders – Modifications to Run Directly Heated Gas Rectifier Tubes

For some extra challenge and way-cool implementation of exotic gas rectifier tubes I’ve adapted my design to use a directly heated xenon or mercury full wave rectifier (2.5V heater). This isn’t a straight substitution for the default EZ81 – significant modifications need to be made to the high voltage side and a heater warmup period of at least 30 seconds needs to be included before applying high voltage. And gas rectifiers cannot be run into a capacitor input so a choke loaded scheme needs to be placed between the tube and the rest of the power supply circuit. From there on it’s the same as above.

This is advanced stuff and assumes a certain level of competency on your end to implement. A basic scheme is shown below to show how I did it. Separate transformers are required for the heaters and the DC heater supply circuit to accomplish this. You need to power on the heaters first, then after the delay the high voltage is applied.

I used an off-the-shelf adjustable delay relay - see Amazon or your local electronics hobby shop, they are cheap and easy to find. Get something that can run on 5VDC and then wire it to the output of the 6VDC heater supply circuit.

Wire your transformers so that the heater transformers are powered on directly by the power switch. The live wire to the high voltage transformer is routed through the delay relay with the timer set to at least 30 seconds. Be sure to include a separate fuse for this high voltage circuit. Your goal is to power up all the heaters and the DC heater supply when you flip the switch; the DC heater supply will then power the delay circuit, and after the delay it will close the relay and power up the high voltage transformer.

I designed this around the 3B22 xenon full wave rectifier which uses a 2.5V heater and requires a minimum of 6.5A filament current (10A is a safe bet). This will also support the 82 mercury vapor full wave rectifier, or any other 2.5V heater full wave UX4 directly heated rectifier you may encounter. Insert obligatory "Mercury is dangerous mkay and you shouldn't use it or even have it present in your home and mercury tubes will shank you in your sleep and kill your dog too".

Output B+ is taken off the centre tap of the 2.5V filament transformer (if you are using a non CT transformer, take B+ off the filament wire). This is then routed into a 15 Henry choke input, with the minimum load set to approximately 20ma by a 25 watt 20K resistor (this dissipates 8 watts continuously so go overboard on the wattage and mount it appropriately for heat dissipation, heat sunk to the chassis being ideal). From here connect it to C2 47uf capacitor on the power supply board.

Due to the inefficiency of a choke input much higher voltage is required than the standard capacitor input circuit, with 750VAC CT (375-0-375) resulting in 360VDC at the output of the choke. For rough estimates of B+ with choke loading multiply one side of the CT output by 0.9 (so in this case 375x0.9= 338VDC, close enough for government work). 

This presumes the choke is loaded properly and working as it should, if not you can get wild voltage fluctuations and you may exceed the ratings of your components. Take B+ voltage, divide by the expected current load in milliamps, divide by 1.2 = approximate choke size in Henries. Then select the closest off-the-shelf value. In this case 360V/20ma/1.2= 15H. Oh and this arrangement only works if you have a relatively stable current draw. 15H is perfect for the 20ma each of my CCS preamps draw, but if you wanted to run both preamps off one supply for 40ma total load then you'd need to recalculate (example 360V/40ma/1.2 = 7.5H).

Choke input is essential; gas rectifier tubes have extremely low output impedance and running them into a capacitor input will cause them to arc and fail.

If all this sounds complicated, expensive, and inefficient, that's because it is. Just stick to the default design unless you desperately want to have a purty glowing gas tube - performance is not any better, the parts cost goes up dramatically, it runs hot, and the output ripple is a little worse. Using a gas tube in this design is purely for cool factor and the satisfaction of doing things the hard way. 

OH and one more thing: gas rectifiers are hideously noisy and spew interference while running, making them the absolute worst choice for powering a phono stage and they should be kept far away from sensitive components. Capacitor C2 ahead of the choke snubs ringing harmonics but should not exceed 0.33uf. Adding this capacitor increases B+ voltage approximately 20V.