6SN7 Tube OTL Headphone Amplifier

Here is a new design from Illuwatar - this is my first design where I have calculated all by myself, not just copying someone elses schematics. The start with this "little" amplifier was the Christmas weekends and all spare time I got together with the lack of parts to my power amplifier. To not be bored, I started with something that I could build out of parts that layed around (as a though in the beginning).
The basic design of the headphone amplifier was found when I read an article at www.tubecad.com about an idea of reducing distortion by creating it (sounds strange but it is true). Also, I wanted to try a OTL (Output TransformerLess) design and a headphone amplifier is a good start due to the higher impedance of headphones (32 - 600 ohm) compared to loudspeakers (1 - 16 ohm), requiring a less powerful output stage. And the amplifier should be possible to build using standard components - nothing exotic or special is needed.

The project started with a schematic of a headphone amplifier found at www.tubecad.com that showed a OTL design for headphones that used a tube called 5687. The design was fine and it wasn't too complicated - there was just one problem: I couldn't find that tube! So, I had to find a replacement and in case of difference, recalculate the component values. No direct replacement was found (the 5687 is a rather robust design that allows 30mA anode current and has a internal resistance of 2000 ohm) but I found a few candidates that could be used instead: E88CC (Ia = 25mA, Ra = 3000 ohm with noval-base) and 6SN7 (Ia = 20mA, Ra = 7700 ohm with octal base). Looking at the tubes, the E88CC was the best candidate but I choosed the 6SN7 instead - maybe not the best one but there was a few reasons:

* I had these tubes at home - I had E88CC's too, so this is not an excuse...
* The tube is bigger - the noval-based tubes are smaller than octal ones and they should look tiny at the top of the amplifier. In some cases, the look make a big choice - size does matter...
* Octal based tubes are easier to use - their sockets are bigger and easier to mount components at.
* The E88CC is a 130V tube, the 6SN7 can take 450V at the anode - some safety margin when the tubes aren't cascaded (as in the driver stage) - in case of component failure.

As output tubes I had some other candidates too, but during the design they was finally refused. That includes the 6AS7 power dual triode and the EL84 (that should be triode connected). The EL84 was refused due to my wishes to make a pure triode design. The 6AS7 was a hot candidate but was finally refused due to the high heater current it needed - the regulator couldn't take it (it is limited to 7A and with 2 x 6AS7 + 2 x 6SN7, 10A is needed to play safe). Otherwise, the 6AS7 is perfect for OTL-design due to it's low internal resistance. And it looks cool too.
During my experiments, I found the the high idle current needed for a class-A run of the 6AS7 created new problems that finally made the tubes go back in their boxes: the 100mA idle current (single channel) created a large voltage drop over the tube rectifier (both GZ34 and 5U4G was tested), making the regulator running at under-voltage (250V went down to 177V) and the entire amplifier was running bad. To fix the voltage drop, I had to go back to silicon rectifiers instead - and with new problems: Delay circuits for the anode voltage and inrush current limiter. Also, the regulator had to be cooled much better when taking almost 250mA of current (for a stereo amplifier) - or a choke had to be used instead (they doesn't heat up as regulators do).
These tubes are now waiting for future projects - the EL84's could make a excellent 10W pull-push power classical amplifier (with output transformers) and the 6AS7's may be used for high-voltage regulation experiments... or an other try to build an OTL.

The Amplifier Design
To understand exactly how this amplifier works and the idea behind "Inverse Complementary Distortion Cancellation", read the article about it at www.tubecad.com.

To view the schematics of the amplifier, click just here (opens in a new window).

Shortly, the theory behind the amplifier design is to reduce distortion by adding it - inverted. The inverted distortion will interact with the "normal" distortion and cancel it out. It is done practically with a two-stage design of the amplifier - a grounded cathode stage followed by a cathode-follower, where the load resistance is identical in both stages (and therefore, the anode currents are identical).
In my amplifier, the second stage (the cathode-follower) have been replaced by a split-load phase splitter where the load resistance have been splitted in half and put both at the anode and the cathode. Signals are taken out from both anode and cathode to drive a class-A output stage.
The output stage is simple - two triodes in a pull-push connection using single voltage and no transformer at the output. The idle current through the output triodes is set to 13mA (remember that the 6SN7 handles max 20mA). Output signal to the headphones are taken between the triodes via a large capacitor. The reason for the capacitor output is the single anode voltage - an alternative would be a +/- supply, but this creates new problems - the power supply becomes more complicated and the output stage needs a balancing circuit to zero out the output DC offset. For a speaker OTL this is feasible, but for a headphone amplifier, this is too complicated.

The high voltage power supply of the amplifier is rather straightforward: a classic tube rectifier design using a GZ34 (or a 5U4G), followed by a 100uF high-voltage electrolytic. The power transformers are 2 x 115V/50VA toroids (the amplifier uses two due to the plan of using more powerful output tubes - with 6SN7's only, one transformer is enough). The special with this power supply is the usage of electronic regulation. By using a regulator instead of a choke gave me some advantages: lower weight, less space needed, lower cost and easier to obtain. There are some disadvantages too: have to be built (contains several components), fragile and generates heat (needs cooling). In this compact design, the advantages wins.

For the tube heaters, there are two power supplies: one that gives 5V regulated to the rectifier tube (it can't share heater voltage with the other tubes) and one for the 6.3V heaters of the four 6SN7.
These two supplies uses their own transformers (one each) of 7V/50VA and two regulators are used (three-pin, 7A adjustable low-drop regulators) - one for 5V and one for 6.3V. The 5V supply is powerful enough to handle both the GZ34 and the 5U4G rectifier tubes. For the 6.3V, the four 6SN7 is enough - the four tubes takes almost 4A together (all tube heaters in parallel).

What I got
The final amplifier (with 6SN7 only) gave me following result:

* Class-A drive of headphones down to 32 ohm with superb sound quality (depends of tube brand).
* Classic design - tubes and transformers are placed at the top of the chassie.
* Rather compact - a design initially made for the 6AS7 can't be smaller. Event the octal 6SN7 take their places. By using the E88CC and silicon rectifiers, the design could be more compact. Maybe a design of that kind will see the light someday (as I can use it at my work)...
* Rather low price due to usage of standard transformers and electronic regulation.
* Simple, straight-forward design without global feedback - easy to repair and gives good sound reproduction.
* All components (except the tubes) can be found in almost all electronic shops. The tubes are common today and a decent guitar amp repair shop should have them (if you live in Sweden, there is LH-Musik that have a very good selection of tubes - and they repair guitar amps too...).

There are some disadvantages with a tube design too (not specially mine - these are common for almost all tube amplifiers):

* Hot and Heavy - this can't be avoided, transformes are heavy and tubes are heated to operate. Also the heater regulators are hot due to the high currents needed. The E88CC tube with 300mA heaters would make a design both ligher and cooler.
* More expensive than a semiconductor version - using modern OP-amps and MOSFET's, a decent headphone amp could be built for a fraction of the cost, in a smaller box, running from a 9V battery...
* High, lethal voltages inside - extra care have to be taken when working with a tube amp due to this. Even when switched off, high voltages can be found in capacitors giving a nice jolt that guarantees wakes you up. If you can't take the high voltages - stick with silicon instead. If you get killed by building an amplifier, there is no use building one - I have warned you, these projects are done at your own risk! Know what you are doing and be safe!

Future plans - improvements:
This was my first try at the tube department to design something from scratch (even if some help was taken from the TubeCAD article) and it worked fine too. Still, there are things that I would improve:

* More output current by using a different tube in the output stage (the 6AS7 for example).
* Better regulators for the heater - this is required if more powerful output tubes are used.
* A low voltage version using the E88CC. This one should be possible to shrink too.

Amplifier Pictures
The prototype setup to test the amplifier before making the final one.

The prototype board is simple - made out of a Veroboard (160x100 mm), two PCB-mounted octal-sockets and a lot of soldering pins placed around the board. The two tube sockets are connected to 16 of the pins, eight around each socket. This makes it easy to build a prototype that are stable enough to test with.
Close-up of the prototype board.
The 250V regulator. This is a light-weight (and cheap) alternative to a choke.
6SN7GT dual triode - the active part of the amplifier. Two of there tubes are used for each channel. This picture shows a chinese brand that I got when the amp was made - I will try to test other brands too to see what tubes are best. The small black thing at the right is a transistor in a TO-220 and is put there for size comparison.
Internal view - all is hardwired and free-mounted except for the regulators that are mouned at prototype boards.
Low voltage regulators for tube heaters. There are two of them at the same board - one for 5V to the rectifier and one 6.3V for the other tubes.
The finished amplifier.
If you build one yourself, send me a mail and tell me about your sucess - and a picture of the design would be nice. Also, improvements and alternations are welcome if yoy like to share them.

Listening impressions

This is an interesting amplifier to listen to - first of all, a decent pair of headphones is needed (forget these "streestyle" phones - they are only suitable for usage on the run with a MP3-player). The OTL-design makes it possible to really listen to the tubes (if the music itself is boring...) as there are no transformers at the output. The ideal solution would be to get rid of the output capacitor too, but this makes the amplifier too complicated. To reduce the effect of the capacitor at the output, a big one of good quality have to be used. I have choosen a type with very low ESR (internal resistance) and with a value of 560uF/400V. It is also bypassed by a high-quality polypropylene capacitor of 4.7uF/400V. These low-ESR electrolytics are common in switched power supplies and therefore, they aren't too expensive and rather easy to find.
The sound of the amplifier really depends of the tubes - I have tested two types for the output stage: a chinese brand (shown at the pictures) and a russinan brand that I don't know the name of (marked with a pentagon). During the tests, chinese tubes was used for the input/driver stage - only the output stage was run with different tubes. What was my impressions?

Chinese tubes: Clear sound with good response and bass reproduction. A disadvantage with this tube is a slight hum could be heard. When testing, I used a pair of Sennheiser HD475 headphones (32 ohm).
The chinese setup gives the amp a similar behavour as my pre-amp I built earlier - fast response and capacity to handle transients.

The "pentagon"-tubes: The hum was gone (still chinese in driver stage, same headphones) but the sound was different too - not as clear as with chinese tubes all over. The bass was more dominat too. I will try more with these tubes as these tests was done in the workshop and was not really seriously perfomed.

Finally, I want to find more brands of the 6SN7 to test with - both Sovtek and Svetlana would be nice to test (in case they have the type). Any NOS 6SN7 that I can find will get a run in the amp too.
Also, a 300 ohm headphone would be nice to try with - I think this amplifier is more suited to work with a little higher impedance than 32 ohm.

Update: I have got new headphones (Sennheiser HD575) with a impedance of 120 ohm to play with - these phones have a much better sound compared to my old Sennheiser (that wasn't bad either).
Also, I have corrected a few things in the amplifier to make it better - specially the heaters have now a fixed potential to the anode voltage, wich made the hum finally disappear, even with the chinese tubes that was more sensitive to hum - they are now complete silence!
When listening with this new setup, I can actually say that the russian tubes aren't bad at all - I start to like them more. The chinese tubes are still good - now when the hum problem is solved, I will give them a chance again.
The new headphones really makes the amplifier to show off - I have tried different types of music and I'm really satisfied. When listening, I connect my CD-player (a Marantz CD-67SE) directly to the amp to eliminate any other sources to noise or other interference. Also, no tone controls are in the signal path.

Design tips (if you plan to make one):
* Play safe when choosing components. Always round up capacitor voltages to the higher value - high-voltage capacitors are often better performes and they will survive changes in transformer output. Resistors should be metal-film or similar - 0.6W types works fine in this design.
* Can't find correct capacitors (specially for output)? No problem - parallelling smaller ones works fine too - just remeber the voltage requirements.
* Try different brand of tubes, specially in the output stage - you will notice the difference.
* Don't make the chassis too small - it is both safer and easier to build if there is space inside. Cooling is also better with a roomier box. Pay some attention to the physical design and layout - you want to put the amp on display for everyone, aren't you?
* Instead of the electronic high voltage regulator, try a classic choke instead. Choose one that gives an final output of 250V at 50mA (each channel takes about 25mA) or load it to give the desired voltage. A minimum fo 10H/50mA should be selected.
* Always use components of high quality - it pays both in reliability and sound quality.
* Shunt bigger capacitors with smaller ones for optimum performance (normally a ratio of 100:1).
* Tubes are hot when running - leave enough clearance around them. Don't put your final amp in a tight space.
* Finally - enjoy the music! Don't let the amplifier drag the attention from the important part!

Schematics to download
Amplifier Shows one channel
Power Supply To be added...