9V Tube Headphone Amp

[CynicalMan]

     Recently, I finally got around to making a tube breadboarding setup. Since then, I've been playing around with diodes, triodes, charge pumps, and everything else that goes into tube stompboxes. This is one of the things that came out of my tinkering. It's a class A headphone amp using a 12AX7 running off of 9V. One of the main advantages of this design is that it uses the ubiquitous 1k:8 ohm audio matching transformer as an OT rather than any expensive and hard-to-find tube OT. Really, the only obscure part here is the 10 ohm 2W resistor, but that can be faked. I used a 16 ohm 10W resistor in parallel with three 68 ohm 1/4W resistors.

    This circuit looks simple enough, but there are a few tricks to increase power and headroom. Firstly, the tubes are running in parallel. This is a simple way to give around 20% more power than a standard class A stage. Secondly, the power stage is running in class A2, which means that the grid voltage is biased positive, and that the grid draws current. On mine, the grid voltage is 0.57V. This again makes the amp more powerful, while reducing clipping. 12AX7s often can't run at 9V because they gate and clip too much, but the combination of the low anode load and the positive grid bias here makes for a pretty clean amp. For more info about class A2, check this out. The JFET stage on the input provides a boost, as well as presenting a low output impedance to the power stage.

    This thing sounds pretty good to me. It has a slight tubey compression at maximum volume, although it sounds more like a thicker tone than normal clipping. At lower volumes, it has a chimey, punchy sound. This thing sounds nice when overdriven slightly, although it can get blatty if driven too hard. It has pretty high headroom, so that's not a big issue. Still, a dirty version of this circuit might be in the works. I don't know what the power output here is, but I think it's in the region of a couple milliwatts. With decent headphones it's loud enough, but far from painfully loud even at maximum volume. Anyway, suggestions are always welcome, and enjoy!

Edit: And, I just hit 1 post per day.  

Edit 2: I just tried out a 12AT7 and a 12AU7, and they both sound more or less the same as the 12AX7. So, you can probably use whatever dual triode you have lying around.

[therecordingart]

Sweet! Definitely something fun to play with. Dabbling in tube circuits is usually cost prohibitive, but this is a cool way to do it on the cheap.

[Brossman]

Is it possible to run this design as a pre- for say... use as a booster (a la Valvecaster)?

1/4" in/out; PS; footswitch... or even...not...?

[dune2k]

Does the Amp care about the headphones impedance? I don't care whether the max. volume changes with different headphones, but I wouldn't want to blow the "OT" when using the Amp with wrong headphones...

[CynicalMan]

Is it possible to run this design as a pre- for say... use as a booster (a la Valvecaster)?

1/4" in/out; PS; footswitch... or even...not...?

As is, not really. The voltage of the signal is actually attenuated in this circuit. The circuit is made to amplify power, but not necessarily voltage. However, you could probably do it if you use a 1k:10k or 1k:50k transformer instead of the 1k:8. Those usually have turn ratios of 1:5 to 1:10, which would supply enough voltage to use as a booster.

Does the Amp care about the headphones impedance? I don't care whether the max. volume changes with different headphones, but I wouldn't want to blow the "OT" when using the Amp with wrong headphones...

I've left this on without any load and nothing has happened. I wouldn't worry about it.

[therecordingart]

Is it possible to run this design as a pre- for say... use as a booster (a la Valvecaster)?

1/4" in/out; PS; footswitch... or even...not...?

I'm just starting to learn this stuff, but if you remove the OT, and cascade the triodes instead of running them in parallel...might work?

[CynicalMan]

The problem is that this is running off of 9V. If you use normal common cathode stages, then the 12AX7s will gate a lot, will sound splatty, and will sound quieter than the input because the gains will be too high, with almost no room for voltage swing. The trick with this amplifier is that the voltage is attenuated, and there is very little voltage swing in the power stage. The amplifier makes up for this by having a greater current swing than in a normal common cathode stage, and thus more power. 12AU7s run okay on 9V, but then you might as well build a valvecaster.

[PRR]

> the only obscure part here is the 10 ohm 2W resistor

Ten 100-ohm 1/4W parallel. Many builders have a clump of small 100r parts.

An option is to use a 12V wart (re-wire heaters for 12V). Get a pinch more headroom.

> in parallel. This is a simple way to give around 20% more power than a standard class A stage.

Optimized it should be twice the power. Unoptimized as here, but with grid resistor tweaked, it should nearly quadruple your audio power output (twice current swing, twice voltage swing, is 4X power).

> I wouldn't want to blow the "OT" when using the Amp with wrong headphones...

IMHO, that's super unlikely at this voltage, even with these low-rated parts. They don't have a voltage rating, but you can't make varnish so thin it won't take a kick-up from 9V.

And tube OT damage happens when amp is pushed LOUD (past clipping). That's normal for stage amps, not for headphone amps.

Also... what's the cost of disaster? $2? $7? It's not like seeing smoke from out a hundreds-buck part.

> remove the OT, and cascade the triodes...might work?

If it didn't "work", we'd never have radio or guitar amps.

But at 9V, normal techniques work "poorly". Without coil loading, tube voltage is even less than 9V, about half lost in the resistor. And in normal resistor loading the plate current is super small. Or it can be large, if (as in Alex's plan) the grid current is forced large. Perhaps larger than plate current. Which leads to a LOSS of current, with small gain/loss of voltage, and little or no useful stage gain.

I suspect it is just as loud with the tube bypassed. But that's not the point.

> 12AT7 and a 12AU7, and they both sound more or less the same as the 12AX7

Probably. At least from the tech side, you are not using the features which make AU different from AX. AU makes more current without crossing the no-grid-current line, but you are way past that line. AX makes more gain into high impedance load, but 1K is low-low-low for any of these tubes. It comes down to grid-cathode impedance. This determines both input impedance and voltage gain. And at similar plate current, AU AT and AX all tend to have similar grid-cathode impedance. There's also grid-plate current gain, which falls-off FAST (unlike transistors) and is not documented on the AX. I think your large resistor to B+ leads to fairly similar plate currents. AX may have a little less current gain, but also a better cathode impedance, so it may be a wash. To make a real change you'd need larger drive current (more than a small JFET) and a tube with a larger cathode (more heater power). And that gets into lily-gilding.

> Dabbling in tube circuits

Note that any insights you glean from this circuit do NOT carry-over to "normal" tube audio. Alex has completely crossed a line that audio designers fear to get near. He has abandoned high gain and low distortion, the main virtues of tubes. He's even violated book ratings on 12AX7 (no positive grid operation), but the tube police won't arrest him because the abuse is (by traditional tube use) too small to care about.

> use as a booster

Take a 0.1uFd cap from the plate to a jack, 50K to ground. That skips the 10:1 voltage loss in the OT, should give a hot guitar level IF you can beat the JFET gate hard enough. You might want an LPB in front for a bit more level. If it gets too strong, the 50K can be an output level pot.

[PRR]

It looks like triode section of 9U8 (6U8 with 9V heater) working at 9V plate will pass 2mA at zero grid, swing 0-4mA with +/-1V drive. This suggests gain into 1K load of almost 2.

9U8 data does not include grid current. My hunch is 9U8 triode is half a 6J6 twin-triode, Data for 6J6 says grid impedance will drop to 200 ohms (but infinity over the sub-zero half of the grid line). So there's really no power gain.

6J6 is a sweet little beast. And there is a 5J6 which is actually rated 4.7V heater. Two 5J6 can be heated with 9.4V (or 9V, same diff) and you get four low-power-gain triodes to jigsaw together. That suggests going push-pull-parallel with the plates. A 1KCT grid transformer will be smoothly loaded over the whole audio cycle. And a 8:1K transformer may be driven with LM386 for gain and grid drive. And 5J6 is a gut on the market (popular TV tuner tube), sells for $4 each.

[therecordingart]

PRR - As always...I really appreciate the time you take to respond. Thank you!

[CynicalMan]

Thanks for the tips! As I'm sure is obvious, my comprehension of tubes is in little bits and pieces gleaned from various books and websites, and is in no way a complete understanding.

An option is to use a 12V wart (re-wire heaters for 12V). Get a pinch more headroom.

I'll mention that when I put this project up on my site. I can't really confirm how it would sound because I don't have a 12V adapter, but I think I'd prefer the 9V version. I like the slight distortion I can get at maximum volume here.

> in parallel. This is a simple way to give around 20% more power than a standard class A stage.

Optimized it should be twice the power. Unoptimized as here, but with grid resistor tweaked, it should nearly quadruple your audio power output (twice current swing, twice voltage swing, is 4X power).

Wow, I didn't realize it would normally be that much. I just got the 20% (0.8dB) from LTSpice, and it sounded right to me when I put in the other triode. Still, it wouldn't be 4x, because using the triodes in parallel means that they're both drawing grid current. This means a lower input impedance, lower input voltage, and less power. It's worth pointing out that I did try using a JFET stage with a lower output impedance, which would mean that less signal would be lost due to grid current. But that means lowering the gain of the JFET stage, so I didn't notice any change in overall loudness.

> remove the OT, and cascade the triodes...might work?

If it didn't "work", we'd never have radio or guitar amps.

But at 9V, normal techniques work "poorly". Without coil loading, tube voltage is even less than 9V, about half lost in the resistor. And in normal resistor loading the plate current is super small. Or it can be large, if (as in Alex's plan) the grid current is forced large. Perhaps larger than plate current. Which leads to a LOSS of current, with small gain/loss of voltage, and little or no useful stage gain.

I suspect it is just as loud with the tube bypassed. But that's not the point.

> use as a booster

Take a 0.1uFd cap from the plate to a jack, 50K to ground. That skips the 10:1 voltage loss in the OT, should give a hot guitar level IF you can beat the JFET gate hard enough. You might want an LPB in front for a bit more level. If it gets too strong, the 50K can be an output level pot.

The tube stage attenuates voltage. There is a bit of current gain but not much. The plate current is still higher than the grid current. Anyway, I think the whole circuit still has voltage loss. Which means that it wouldn't work as a booster. The simplest way to use it as a booster is to put in a step-up transformer instead of the step-down transformer, as I suggested earlier.

> Dabbling in tube circuits

Note that any insights you glean from this circuit do NOT carry-over to "normal" tube audio. Alex has completely crossed a line that audio designers fear to get near. He has abandoned high gain and low distortion, the main virtues of tubes. He's even violated book ratings on 12AX7 (no positive grid operation), but the tube police won't arrest him because the abuse is (by traditional tube use) too small to care about.

Well, I think guitarists abandoned "low distortion" ages ago.
If anyone wants a completely clean tube headphone amp that plays by the rules, go here:
diyaudioprojects.com/Solid/12AU7-IRF510-LM317-Headamp/

It looks like triode section of 9U8 (6U8 with 9V heater) working at 9V plate will pass 2mA at zero grid, swing 0-4mA with +/-1V drive. This suggests gain into 1K load of almost 2.

9U8 data does not include grid current. My hunch is 9U8 triode is half a 6J6 twin-triode, Data for 6J6 says grid impedance will drop to 200 ohms (but infinity over the sub-zero half of the grid line). So there's really no power gain.

6J6 is a sweet little beast. And there is a 5J6 which is actually rated 4.7V heater. Two 5J6 can be heated with 9.4V (or 9V, same diff) and you get four low-power-gain triodes to jigsaw together. That suggests going push-pull-parallel with the plates. A 1KCT grid transformer will be smoothly loaded over the whole audio cycle. And a 8:1K transformer may be driven with LM386 for gain and grid drive. And 5J6 is a gut on the market (popular TV tuner tube), sells for $4 each.

That's a neat idea. I think I have a 6J6 lying around somewhere, I should try that out. I think the 12AX7 grid impedance is higher, around 1k-2k in this circuit. I tried out class AB1 and AB2 power stages when I was designing it, but they were only marginally louder than the parallel class A2 setup. Also, they required a phase splitter, and I preferred the simplicity of the class A2 stage.

[CynicalMan]

Unverified layout:


Please let me know if you verify it.

[JAROPDA]

Hi I just got all the components and I will start making this headphone tube amp.

I have impatient to know how it sounds and provides quality
comment and defects if any ...


regards

and many thanks by the circuit

[CynicalMan]

To me it sounds clean, with a bit of compression. The only problem I can think of is that it isn't that loud. It's good enough in my opinion with decent headphones or iems, but the volume is pretty moderate.

[JAROPDA]

Hello, I've been riding the circuit on a breadboard only following the pattern ... but it sounds like nothing, but nothing I have reviewed does not ring.
I have two questions the pin-valve, how are connected by pin numbers?
Can you tell me the resistance 2w 2Ohm going to heater that number is in the valve to connect?
and the transformer is a tm013 1k / 8 ohm but how do I connect? has three pins on the primary and secondary three
what are they?
   primary pin 1 pin 2 ---->>>> -----> 9 volt outlet valve?
I need help
thank you very much

[CynicalMan]

You want a 2W 10 ohm resistor. If you don't have this, put 10 100Ω resistors in parallel, and use that instead of the single resistor.
You connect one side of the resistor to +9V. You connect the other side to the tube's pin 9. Then, you connect the tube's pins 4 and 5 to ground.

Primary pin 1 connects to +9V. Primary pin 2 is not connected to anything. Primary pin 3 connects to the tube's pins 1 and 6.
Secondary pin 1 connects to both the left and right headphone hot wires. Secondary pin 2 is not connected to anything. Secondary pin 3 connects to the headphones' ground.

[JAROPDA]

oh thanks ok
I test today

regards

[JAROPDA]

Hello I have been riding the circuit in a pre-punched board and soldering tin that I had the feeling that the breadboard connections are not very safe.
The attempt has been nil, I'm not sound demoralizing, I tested the connections that you gave me to the pins of the valuvula and transformer.

if you do me the favor to tell me:

  each pin of the valve that is part of the circuit
the transformer that will pin valve pins
and the pin is connected MPF102 as 1-2-3 maybe there comes the problem
  What is the pin 1 to 2 - 3?

also may have damaged the transistor by connecting the same with evil and perhaps the valve is broken.

if you do me the favor to introduce me to be forever grateful.
the scheme is very easy but it is not going to happen.

thank you very much
a greeting

[iccaros]

I think you are asking for these


and the tube

[JAROPDA]

Hello I have been testing the circuit and does not sound like anything.
I connected the pins of the transformer as you said.
I have one doubt the valve pins 8 and 3, do not connect to anywhere?
 I've tried it and does not sound too well.
I fear that I have messed up the valve by connecting wrong, they can ruin for having put another pin 9 volt?
I have tried in my tube amp and if it works as a preamplifier.
The transformer impedance gives me the data sheet:
I had to give primary 1k
                             8 ohm secondary

just gives me another pin in 75 ohm and 1 ohm resistance
I get the impedance in the primary between pin 1 and 2 is 38 ohms and 1 and pin 3 75 ohm
in the secondary pin 1 and pin 2 1 ohm and pin1 and pin 3 1.3 ohms
If you have any idea what might happen I'd appreciate it much

thank you very much

a greeting