HeadLine

Headphone Tube Amplifier (OTL)

Page 2: December 08, 2006

Amp Design Alternatives

There are several design issues that need to be resolved and determine in great deal what amp to build:

• Can we do without an output transformer or not. Looking at the impedance of the headphone (300 Ohms) my first reaction would be that this would not be necessary.
• Will the amp be used as a pre-amp as well? This will determine whether we need an additional power reserve and connections.
• What type of output tube(s) to use. There are designs with a single EL-84 put in triode mode, but there are also designs based on two parallel triodes. I like 12SN7 tubes a lot, and I have some left ... (see picture)
• Will we use feedback or not?
• Will we use a simple design with traditional passive components or use modern FET alternatives for example for the current supply of the Cathode Follower stage.
• Single Ended Triodes or Push-Pull?

As said above, designing a tube based headphone amp is not that simple. First of all it does not seem logical to use an output tranny, which means that an OTL design is immenent. But, since good headphones will have a 200-400 Ohms impedance the amp needs to be able to provide sufficient output power.

I therefore expect the MF to be using a 2-stage design with a Cathode Follower in the second stage. After all, it uses two double triode tubes so the options are quite limited.

General Remarks

For OTL output stage, and headphones with 300 Ohms load we'll need an output capacitor with a value of 100-220 uF. We'll probably bypass the cap with a smaller value of 1/100 of the main value. Too bad we'll have to use such a high value but my feeling is that it's still a better better choice than most output transformers.

What Tubes to use

The tubes to be used in a headphone amplifier must posess certain qualities.

• A low output impedance, which is of-course dependent on both the circuit and the tubes used. In order to have a low output impedance in a cathode follower design, a tube must have a relatively low internal impedance and a relatively high mu.
• Feedback will help in getting a better output impedance. Feedback will reduce the gain of the amp however, but when choosing a tube with a good amplification factor (u), enough gain would remain at the output.
• It must be able to drive enough current. For my headphones about 20mA per channel is needed.

Circuit Design Alternatives

I decided to work on some alternative designs. below You'll find a few of the possible design options. Please keep in mind that I simulated these designs and since I did NOT build them all yet I strongly advice you to have a good look at them before even think of building one of these yourself. The design I've chosen for my amp is described further on the next page, others may be for later reference.

1. Simple

Below you'll find the simple version of a headphone amp. Two tubes are used: The first stage is for amplification and the second stage is a cathode follower to make sure the output impedance will probably acceptable for some headphones (of 400 Ohms). But in order to cope with the relatively difficult headphone such as the Sennheiser HD600 series it may be advisable to put two tubes in parallel in the output stage.

2. Simple with Feedback

Feedback is a way to ensure that the amp will deal with difficult loads easier. A lot of things are to be said about feedback, and most of it is negative feedback anyway. But in this case it would not hurt at all using feedback for this amp.

A second circuit description in JPG

3. Using an EL84 for the output stage

Here is the circuit for the EL84 based headphone amp. It's simulated with SPICE.

Btw, this amp can be made with EL34 as well.

4. Using two parallel triodes for output stage

I noticed that I had a few 12SN7 types around. I therefore decided to investigate their use for output stage. Simulations in SPICE with an output load of 300 Ohms show that these tubes, which are from a previous generation may be just be what I'm looking for in a headphone amp.

Well, maybe this is for me a good starting point as I have most of the material in house needed to build this amp. I have to take into consideration though that the best achievable value for the output impedance is according to the formula: Ro = (Ri+Ra)/(u+1) and since Ri is 7700 for the 12SN7, Ra is 0 or higher and u is 20.5 or so, the best value will be in the order of 175 Ohms. This is lower than the 300 Ohms of the HD650 or the 600 Ohms of the HD424, but it is relatively hight value. Hmm...

The 12SN7 tubes offers a plate current of 8mA per channel, so I will choose the value of cathode resitors such that we will more or less get to this value.

The choice for the 12AT7 tube is not very strict. I think that the ECC 82 tube does not quite have the punch I like and it does not offer much headroom in amplification also. The ECC83 works fine except that it offers far too much gain.

---

Design for HeadLine

I've chosen to build the following design. The best strategy when building something like this is working backwards: Start with the desired result and what you need and work back to make construction/design decisions.

Well the Sennheiser heaphones I own are 300Ohms impedance, and at high volume levels the voltage level is around 3 Volts. This means that 10mA is used per channel (and most likely double that in peak load situations). This means 20mA per channel. Simulations in Spice tell me that indeed 18mA would be reached for some tube configurations.

The 12SN7 tube would deliver about 10mA per channel so I would need two halves of this tube, or in other words I'll have to use both halves of this double triode in order to provide the desired amount of current. Of course tetrodes or penthodes would be able to drive this load easier, so would push-pull designs. But I would prefer OTL and SET setups.

Amp Circuit

Below you find the circuit of the amp as built. (Click on the picture to enlarge it)

I stuck to this design for the greatest part, although I was forced to use 12ax7/ecc83 tubes in the first stage due to my depleted supply of 12at7. As you can see there is nothing exotic about this headphone design. Instead of the 12SN7 I could have used the ECC82/12AU7 tube, but I like these 12SN7 tubes a lot.

Also, due to my limited supply of high-voltage capacitors I was not able to use 400V/470uF in the power rails as I calculated. Probably due to that there is a very light humm in my headphones wich is present on all volume leverls. I've ordered a few 400V/470uF capacitors that should make my power supply very.very quiet.

Power Supply

I've decided to put the power transformer outside the amp, as does MF. An external 12-0-12 or 15-0-15V Wall Wart transformer should do but I did not have that one at home and I was not prepared to spend lot's of bucks for the first headphone that I would build. Therefore I used an old 10-0-10 Volt/4A powersupply that I bought in a local dumpstore for just 3 Euros and I used the 20V as a starting point. This 20V AC is then fed into a power regulator providing 12V stable for the heaters, and also to a step-up transformer that converts it back to about 250Vac.

After rectification with 1N4007 diodes, caps and choke the powersupply provides a near perfect 220V DC (between 200 and 250V is OK)..

I take it that the power supply is always up for improvement anyway, but again I would like to build something simple heren.

(Look for the power supply circuit in the figure above.)

Switches, Connectors etc.

Since a headphone is a delicate piece of equipment, sudden pops and clicks might very well damage the headphone. Therefore we might want to build a slow-start circuit for the amp although I want to check the need for such a solution first.

Update: Since I decided to use a SRPP I do not expect many sudden pops and clicks, and I've decided not to implement such a circuit at his moment.

---

Modification 1

During the prototyping phase (see next page) I had to swap some parts in the power supply. At that moment I decided to implement a "SRPP" for the first/gain stage of the amp. In the original idea I used two half tubes for the first stage, opposite halves so I could swap the input tubes at some point in time and extend the life of the tubes. But I decided that I rather have super sound than super long life from my tubes.

Since the power supply was giving me troubles I decided to do something extra besides eliminating faulty components: Make the first stage a SRPP and improve my Power Supply Rejection Ratio a little.

Anyway, the resulting circuit looks like this:

Since I did not have a step-up transformer 230V-12V I used the one with 18V secondaries. The result is that when used as a step-up transformer we get a lower B+ voltage. I changed the resistors in the power supply and added CRC filter with a resistor value of 1K and as a result all more or less works out just fine.

I will publish measured voltage levels as soon as I've decided whether I will change the step-up or leave it as is.

---

Modification 2 (Jan 21, 2007)

Although the first modification has been succesful, there are a few and mostly minor details that I want to fix in the second modification.

• Use of a better volume pot: The old pot is a cheap 22K Log pot which performs fine but cannot be compared to a 50K log of ALPS
• Change a few resistors in the PSU and the cathode resistor of the output tubes to a 5W wirewound version. I used 1W Beyschlag before (as shown on page 3) and although there was no sign of overheating the power rating should be 2W at least according to my calculations.
• Change the LED's to lo power versions. The current blue LED use 30mA each and that's close to too much for my heater supply. The diodes have a rating of 1A. And as heaters consume 300+300+150+150 mA when heated up, another 60mA is a lot.
• There sometimes is a slight transformer humm at 2/3 of the volume setting, which I found out is probably related to input wire routing. I use screened microphone cable for the wire going from RCA connectors to volume pot and from pot to grid of the input tubes. Stil I found that it makes a difference how to route these cables.

I changed the pot and the resistors, and made progress with the wiring. Waiting for low power LEDS to arrive.

The output impedance of the amplifier is not as low as I would want it to be. But after all, if that would hev been the primary concern I would have chosen for a solid state solution. The output impedance based on the formula would be (Ri + Ra)/(u+1) which translates for the 12SN7 tube into (7700+0)/(20+1)=380 Ohms. Since I use two tubes in parallel the effective output impedance would be 190 Ohms. Simulations show an output voltage even lower, in the order of 120 Ohms, but this is with a standard load resistor of 22k parallel to the load of the headphones.

I'm still thinking about adding an (optional) feedback loop for the amp, as this will have a good effect on the output impedance and the distortion figures. The gain I have from the amp is high enought to try it.

 

<< Page 1; Intro

^^ Back to DIY Page ^^

Page 3; Construction >>

 

© Maarten&Annemarie, November 2006

last modified 13 March, 2007