Tubes 102 - Intro to Power Supplies for Tube Circuits

vigilante398

Authorized Vendor
I had planned to add this post on to the previous post about tube preamp design, but as I was writing it I found it getting way longer than I had planned, so I figured I would make it a separate topic.

So how do I power these tubes?

Well let’s start with the heater, without that you won’t be getting any signal regardless of what you put on the plates. The heater is also known as the filament, but I’m used to calling it heater, so that’s what I’m going to continue to do.

Our 12AX7 dual triode has three connections for the heater, the two ends and a center tap. It may help to think of the heaters as a pair of resistors forming a resistor divider.

1.JPG

2.JPG

Since the only goal is to get it hot, and resistors are non-polarized, you can either use AC or DC to heat it up, the heaters don’t care. The only important thing is that there needs to be approximately 6.3V between pin 4 and pin 9, and 6.3V between pin 5 and pin 9. There are two common ways to do this. In amplifiers it’s common to have a 6.3V transformer winding on your power transformer, so you can tie pins 4 and 5 together, and connect your 6.3V AC lines to pin 9 and your combined pins 4/5.

3.JPG

In pedals we’re a lot more likely to have DC running around in the pedal, and it’s not uncommon to run a pedal at 12V. So in this case we leave the center tap unconnected and connect +12VDC to one heater pin and GROUND to the other heater pin.

4.JPG

I’ve also seen (and built) plenty of pedals that take a 9V input and step it down to 6V using a regulator like L7806 and do the same setup as found on amplifiers.

5.JPG

One trick I’ve seen to get exactly 6.3V is to add a diode between the regulator’s ground pin and ground to adjust the voltage. A schottkey diode with a 300mV forward voltage would give you exactly what you’re looking for. I’ve seen this in one DIY project (TH Customs if I remember right) and it works, but I don’t generally recommend it as it removes the heatsink of the regulator from ground, so it’s harder for it to dissipate the heat it’s going to generate. But it’s a neat academic exercise.

6.JPG

So how much current is the heater going to pull?

It’s Ohm’s Law time! On a 12AX7 (or similar dual triode in the same family) the nominal resistance of the filaments is 42ohms on each side. So if you power it with 12.6V like the datasheet recommends, it puts both resistances in series to give us 84 ohms, and we have

I = V/R

I = 12.6/84

I = 0.15 = 150mA

If we’re running the heaters off 6.3V we have the two resistances in parallel, which gives us an equivalent resistance of 21 ohms, so we have

I = V/R

I = 6.3/21

I = 0.3 = 300mA

So in pedals where current draw is often a big consideration, as we have to power them somehow and many pedal supplies are very limited in output, it makes sense to use the method that will pull the least amount of current. This is why it’s common to use a 12V supply for tube pedals. But…

How critical is the heater voltage?

Not very! The datasheet calls for 12.6V or 6.3V but we commonly use a straight 12V or 6V. How low can we actually go? We can absolutely get away with powering a 12AX7 with 9V on the heater instead of 12V! This is the way I chose to hook up heaters in my pedals, so every tube pedal I make runs the heaters in series straight off the 9V supply. This gives us an additional advantage in current draw, since with a 9V supply:

I = V/R

I = 9/84

I = 0.107 - 107mA

When we use a lower voltage for the heater supply, as long as it's still high enough to heat the tube, we can save power since the heaters are fixed resistances. To be completely honest I have no idea what the "minimum" voltage is before you start noticing a difference in sound, but 9V gives no audible difference compared to 12V and only pulls 107mA on the heater as opposed to 150mA on a 12.6V supply.

7.JPG

Does it matter which pin is positive and which pin is negative?

Nope! Remember the heater is like a resistor, there’s no polarity, there just needs to be the right voltage between the two points.
 
I have my heater connected, what do I do now?

This is where it gets dangerous. If you’re not comfortable with high voltage, this is a good time for you to say “well this has been a fun learning exercise, I should probably stop now so I don’t get hurt.” I will reiterate this point multiple times, as it frankly can’t be said enough. The voltages present in tube amps and even in tube pedals CAN KILL YOU and will hurt the whole time you’re dying. I don’t want this to feel like gatekeeping, but I can’t emphasize enough that if you aren’t comfortable taking the necessary safety precautions, you really shouldn’t be working on high-voltage circuits. It doesn’t matter how much experience you have, you should always keep in mind how possible a painful death is when dealing with high voltage.

So how can I keep myself safe while working on high-voltage circuits?

There are a couple precautions I recommend when working on high-voltage circuits:
  • The one-hand rule. When working on a live circuit like probing voltages during operation for debugging, etc., I keep one hand behind my back. It’s become a reflex now, when my right hand gets close to the circuit, my left hand goes behind my back. I’ve known some techs that stick one hand in their pocket. Anything you can do to avoid having both hands in the circuit at the same time. If you discharge high voltage into your hand it will absolutely hurt and you may get burns, but it won’t kill you. If you have both hands in and the high voltage gets a chance to jump across your heart, you will absolutely die.
  • Treat every circuit as though it is energized until proven otherwise. It doesn’t matter how long it’s been unplugged, you avoid touching ANYTHING inside the circuit until you’ve measured anything and everything that could store high voltage. I recommend using one clip lead and one probe lead on a multimeter to measure things. You can use the clip for negative and clip it to the chassis so you can probe voltages with one hand behind your back. ALWAYS check EVERYWHERE before you touch ANYWHERE.
  • Respect the circuit. Working on pedals and amps is fun, but when high voltage is involved it’s time to get serious. You should always be at least a little bit afraid, because being too confident can lead to carelessness.
  • If you don’t want to follow these steps, please don’t work on high-voltage circuits. Again I don’t want to be the gatekeeper of tube circuits, I’m not the high-voltage police, but even if I’m not legally responsible I don’t want to be morally responsible for someone getting hurt. If you’re not comfortable enough to be safe, please don’t do it. No great tone is worth dying for.
Okay, I understand the risks and I am going to be completely safe. Where do I start?

We already talked about the heaters for the tube, so your tube is heated up and ready to pass some signal. For the real meat, you’re going to need some high voltage for your tube’s plates. As we discussed earlier with tubes, the plate resistor goes between the high voltage supply and sets the gain of the circuit.

Where does the high voltage come from?

There are generally two options for high voltage, transformer or switched-mode power supply (SMPS). We’ll start with transformers as it’s simpler (though probably more likely to kill you if you’re not careful). In amplifiers you need a lot of current, so a transformer is the best option for voltage conversion. You’ll need a rectifier with your transformer, as unlike the heater, the plate needs DC. If the high-voltage secondaries of your power transformer don’t have a center tap, you’ll need a bridge rectifier, like this:

8.JPG

If your power transformer has a center tap on the high voltage secondaries then you can get away with a full-wave rectifier like this:

9.JPG

You will absolutely need to filter this high voltage DC line with some nice big capacitors, and full amplifiers (and more complicated preamps) will have RC filters for each stage of power to isolate them, but the details of how to go about with that is a bit outside the scope of this, so I’ll post some links at the bottom for further reading on power supply filtering.

In pedals where we already have a DC input, it makes sense from a size, cost, and complexity standpoint to use a switched-mode power supply (SMPS) for high voltage. I won’t get into the mechanics of how a SMPS works, but I’ll post a link at the bottom for those that like to understand things. For all points and purposes you can think of SMPS as a voltage regulator that operates more efficiently than a linear regulator (like L78** etc) which means it loses less power in the form of heat. SMPS come in two varieties, buck (step-down) and boost (step-up). We need to take a 9-12V input and kick out some serious high-voltage, so it will not surprise you that we will be using a boost SMPS.

The supply I (and countless others) use for high voltage in pedals is known famously as the “nixie supply” as it was originally used to provide the high voltage for hobbyist nixie tube projects. This is the configuration I use:

10.JPG

C1 needs to be big but the exact value isn’t critical, I’ve since dropped to 100uF for size. I’ve been told the inductor value isn’t critical either, but 100uH inductors aren’t hard to find, so I’ve always stuck with them. Most drawings recommend a 2A or larger inductor, I use 1.8A as they are easy for me to find in a good size package and reasonably priced.

P1 on this drawing is a 5k trimpot, and this is what sets the high voltage. I recommend setting this right around the middle before supplying power, then when power is supplied measure (with one hand behind your back, never put both hands in the circuit at the same time, per the warnings earlier) the output and adjust the trimpot slowly until it gets to the desired value. You’ll have to figure out the best way to do this for you, personally I put a test point on my boards that is a large enough through-hole that I can stick the multimeter probe into it and it will stay there, which frees up my right hand (my left hand stays behind my back at all times) to use a small screwdriver and turn the trimpot.

What ratings do the SMPS components need to be?

You really aren’t seeing the high voltage until you get past the rectifier, so there’s no point in using massive components for the whole things. All resistors can be 1/4W (or even 1/8W), and C7 and C6 can be 50V capacitors. C1 needs to be sized based on your expected input voltage, I use 16V rated caps but higher is of course great. The rectifier diode needs to be an “ultra-fast rectifier” like UF4007, simple rectifiers like 1N4007 will not work.

C2 just needs to be sized for the high voltage that you are planning to run the circuit at.

How high should I run the voltage?

The SMPS schematic posted above is capable of providing around 500V maximum, and you really shouldn’t run a tube preamp that high, your tubes won’t like it. The 12AX7 datasheet recommends a plate voltage between 100 and 250V, but keep in mind that the plate resistor in your circuit will drop the voltage a bit so the plate won’t see the full voltage of the supply (unless it’s a cathode follower, we’ll cover that in another chapter), and even then tubes are generally forgiving devices, and it’s common to exceed the maximum voltages and get away with it.

I still run most of my preamps around 230-240V on the supply, and I use 250V capacitors. Fortunately for us the SMPS is very stable and the voltage doesn’t swing around, so we don’t need a huge margin above the supply voltage. SMPS have a regulated output, which means the output voltage is independent of the input voltage. If I set my regulator for 235V with a 9V then plug a 15V input into it I will still get 235V from the SMPS.

I think that’s pretty much it for power, let me know if there’s anything I missed covering and I can address it. Until next time, this has been “semi-coherent ramblings of a non-expert that has built a couple things with tubes.”

More reading on power supply filtering in amps: http://valvewizard.co.uk/smoothing.html

More reading on SMPS: https://en.wikipedia.org/wiki/Switched-mode_power_supply#Explanation

My previous post on tube preamp design: https://forum.pedalpcb.com/threads/tubes-101-intro-to-tube-preamp-design.10625/
 
Last edited:
I suddenly feel the need to only play acoustic instruments. I can't go near my fridge, and I'm having recurring nightmares of a maniac toaster chasing me with an audio probe.

Seriously, great stuff! Thanks for sharing it with us. I recently built my first SMPS.
 
That "Nixie Supply" circuit has room for improvement.
1. Regulation stinks because it uses Q1's Vbe as a reference. Vbe changes by 3mV/°C or about 0.5%/°C. That means that a 20°C temp change makes the HV change by 10%. Trimpots are unreliable and I do not trust them in this application.
2. The feedback loop is not compensated, so loop stability (freedom from oscillation) is unknown.
3. There is no soft-start or current limiting, so during startup C1, C2, M1 & L1 see large currents. L1 saturates which makes for even larger currents. The IRF740 is rated for 400V & 40A, so it can probably take it. Some wall-wart power supplies will not like the large surge currents and may motorboat during startup.
4. C1 & C2 should be bypassed with film or MLCC capacitors because at 45KHz (switching speed) and it's harmonics, aluminum electrolytics act more like resistors than capacitors.
5. The 555 was never designed to drive MOSFETs safely or efficiently. In switching applications, MOSFET have large gate currents.
6. There is no undervoltage lockout. If the power source sags, the MOSFET won't get enough gate voltage and that increases the stress on the MOSFET substantially.
7. You can't just buy inductors at random and expect good results. Inductor design for switching converters is not simple and books have been written on the subject.
8. The UF4007 is a good choice for rectifier because it switches quickly. Do not be tempted to sub 1N4007 because they are slow and this will increase stress and noise.

Having said all that, I have a tube pedal which uses that very circuit and so far, no problems.

In a former life I was a switching power supply designer and this circuit gives me a headache. I picked up some UC3843s from EG a while back and have plans to design a proper HV switching regulator.
 
The datasheet calls for 12.6V or 6.3V but we commonly use a straight 12V or 6V. How low can we actually go? We can absolutely get away with powering a 12AX7 with 9V on the heater instead of 12V! This is the way I chose to hook up heaters in my pedals, so every tube pedal I make runs the heaters in series straight off the 9V supply. This gives us an additional advantage in current draw, since with a 9V supply:

I = V/R

I = 9/84

I = 0.107 - 107mA
Have you measured the heater current? It's probably higher than you think because the filament resistance is temperature dependent. When we run the filaments on lower voltage, they run at a cooler temperature and consequently, the resistance is lower. Lower resistance = higher current. Different tubes respond differently to reduced filament voltage. When I'm testing tubes for HiFi use, I check the gain at the nominal heater voltage, then lower the heater voltage until the gain drops to the lower spec limit. The tubes that meet spec at the lowest heater voltage are the better ones in my book. YMMV.
 
That "Nixie Supply" circuit has room for improvement.
1. Regulation stinks because it uses Q1's Vbe as a reference. Vbe changes by 3mV/°C or about 0.5%/°C. That means that a 20°C temp change makes the HV change by 10%. Trimpots are unreliable and I do not trust them in this application.
2. The feedback loop is not compensated, so loop stability (freedom from oscillation) is unknown.
3. There is no soft-start or current limiting, so during startup C1, C2, M1 & L1 see large currents. L1 saturates which makes for even larger currents. The IRF740 is rated for 400V & 40A, so it can probably take it. Some wall-wart power supplies will not like the large surge currents and may motorboat during startup.
4. C1 & C2 should be bypassed with film or MLCC capacitors because at 45KHz (switching speed) and it's harmonics, aluminum electrolytics act more like resistors than capacitors.
5. The 555 was never designed to drive MOSFETs safely or efficiently. In switching applications, MOSFET have large gate currents.
6. There is no undervoltage lockout. If the power source sags, the MOSFET won't get enough gate voltage and that increases the stress on the MOSFET substantially.
7. You can't just buy inductors at random and expect good results. Inductor design for switching converters is not simple and books have been written on the subject.
8. The UF4007 is a good choice for rectifier because it switches quickly. Do not be tempted to sub 1N4007 because they are slow and this will increase stress and noise.

Having said all that, I have a tube pedal which uses that very circuit and so far, no problems.

In a former life I was a switching power supply designer and this circuit gives me a headache. I picked up some UC3843s from EG a while back and have plans to design a proper HV switching regulator.
I would be very interested to see what a boneyard version of this looks like.
 
One question, what is the configuration if the tube doesn't have a center tap (like pin 9) and you are running it via 6.3 Volt DC supply?
Is creating a virtual center tap the only way to go?
 
One question, what is the configuration if the tube doesn't have a center tap (like pin 9) and you are running it via 6.3 Volt DC supply?
Is creating a virtual center tap the only way to go?
No, if the tube doesn't have a center tap there's no reason to create one, it just means you run the heater off 6.3V and you're good to go. The center tap of heaters is just so you have different options of heating the tube, it isn't required.
 
No, if the tube doesn't have a center tap there's no reason to create one, it just means you run the heater off 6.3V and you're good to go. The center tap of heaters is just so you have different options of heating the tube, it isn't required.
Perhaps I'm missing a very easy point, but if you have two filaments with no center tap (and thus only two pins), how can I have 6.3V over each of the filaments?
I know its a very stupid question but wouldn't running one pin to the power supply (6.3V) and the other to ground just give 6.3V over both in series? Which I guess is not good
 
Perhaps I'm missing a very easy point, but if you have two filaments with no center tap (and thus only two pins), how can I have 6.3V over each of the filaments?
I know its a very stupid question but wouldn't running one pin to the power supply (6.3V) and the other to ground just give 6.3V over both in series? Which I guess is not good
If you have two pins for the filament then you only have one filament, not two. The only thing to do (with a DC connection) is give 6.3V to one pin and ground the other.
 
Back
Top