Building a Low Power Headphone Amp with the Tubelab SE board:
The Tubelab SE board can also be used to build a low power headphone amp. Most audiophiles will build the board in the normal fashion, using the popular output tubes and a resistive attenuator to reduce the output power for headphone use. This way the amplifier can still be used with speakers. The type of resistive attenuator depends on the impedance of the headphones being used.
There is another option. The board can be built up specifically for headphone (or other low power use) with output tubes capable of power far less than one watt. I have used tube types 01A, 30, and 31 for this application. Output power for the 01A and the 30 is about 100 milliwatts (1/10 watt), depending on how hard you want to push the tubes. The type 31 is capable of about 1/4 watt. 100 milliwatts is extremely loud through most headphones. You should have some electronics experience, and be prepared to do some experimenting if you do this. Since headphones come in impedance values from 8 ohms to 600 ohms some experimenting is needed to find the tube and transformer combination that works well with your particular headphones and your musical tastes.
There are two obstacles that must be overcome to use this approach, but excellent sound awaits those who succeed. I am not a headphone person so I can only relate the experiences of a customer who, after much experimentation, reached his own personal audio nirvana with an 01A amp. Both obstacles relate to the choice of tubes and transformers. This customer tried multiple power and output transformer sets and all 3 types of tubes mentioned above to arrive at his personal choice. I personally could not hear the differences between some of the different combinations, but they sounded different to him. I performed all of the experiments while he listened.
The tubes mentioned above operate at a much lower voltage than the typical 45 or 300B. The 01A works best at about 125 volts, the 30 likes about 150 volts, and the 31 can run as high as 175 volts. This means that the power transformer used for a 31 won't work with an 01A and vice versa. Therefore you should choose the tube first then get a transformer that delivers the right voltage for that tube. There are no catalog transformers with all three windings that support such a low output voltage. The three windings are a 5 volt 2 amp output to power the rectifier filament, a 6.3 volt center tapped winding to power the filament supply and a center tapped high voltage winding. A 250 Volt center tapped HV secondary would work for 01A's and 30's and a 300 Volt center tapped secondary is needed for 31's. I could not find a catalog transformer in stock anywhere that fit this criteria.
There are multiple ways to solve this problem and we tried most of them. First I set the amp up with a bench power supply (an old Knight kit that I got at a hamfest). The bench power supply provided the filament voltages and provided regulated adjustable high voltage. This is the same power supply that I used to design the original amplifier, so I know that it is capable of producing excellent sound. The customer spent several hours trying different tubes at different voltages, with three different types of headphones and several types of output transformers (covered below). From this experiment we learned that he really liked the sound of 01A's at 125 volts. The external power supply allowed us to zero in on the type of output tube and its optimum operating voltage just by turning the knob. Since we were using an external regulated power supply there was no audible hum. Hum can be a big problem with headphones since they are tightly coupled to your head.
Understanding the hum issue we both agreed to use a filter choke in the power supply (mount it off the board, wire it into the board in place of R4). This would allow more flexibility to adjust the high voltage output. The filter choke that we used was one that I had in my spare parts collection. It came from an old Hewlett Packard audio oscillator that I parted out. This was later changed to a Hammond 193B because it looked better.
Next we had to figure out how to get our 125 volts, and supply the filament voltages. The 01A's have 5 Volt filaments, no problem, the board has a 5 volt setting on the filament regulator, which does NOT use the center tap on the 6.3 volt winding. This means that a center tap on the 6.3 volt winding is not required when using 5 volt tubes. The 30 and 31 tubes require 2.0 volts for their filaments. This voltage can be produced by changing the value of R35. A center tapped transformer is normally required to operate 2.5 volt tubes (45 and 2A3) but the 30's and 31's draw such a low filament current that I could use a non center tapped transformer by adding a larger heat sink to the filament regulator.
The obvious solution to the transformer problem is to use solid state rectification. There are stock, cheap transformers that will work here. I had a Hammond 261G6 transformer that I got on Ebay for a phono preamp project. I mounted a pair of Hex Fred diodes in the board where the 5AR4 goes. The amplifier worked, we played with the value of C4 to get the voltage to 125 volts. Both of us agreed that it did not sound right. The FFT analyzer revealed 120Hz intermodulation products even there was no audible hum and very low residual 60Hz or 120Hz output. Substituting the external 125 volts, while leaving the filaments powered by the Hammond transformer solved the problem. We agreed to explore tube rectifiers. The customer had a NOS Mullard 5AR4 that he wanted to use, but this would require an additional 5 volt transformer to power its filament.
The next experiment was to try a 6 volt filament rectifier. This would not require an additional transformer since the 6 volt filament is isolated and can be powered by the main filament winding. I decided to try a 6X5 because I have hundreds of them. The 6X5 has a different pin configuration than the 5AR4 however it can be made to work in this board by adding a few jumper wires. I wired it up and we proceeded to test. The value of C4 had to be increased to raise the voltage to 125. We both agreed that the sound was much improved. I believe that the sound was equal to the bench supply. The customer however had this 5AR4 in his pocket and he was going to hear this amplifier with it, so more experiments were needed.
A 5 volt rectifier requires a separate 5 volt filament winding since the filament is connected to the cathode in a 5AR4. The cathode is the B+ output on these tubes. Since the 6.3 volt filament winding is grounded it can not be connected to the 5AR4. I have several 5 volt transformers (again from Ebay), so I connected one of these. Since the 5AR4 is a more efficient rectifier than a 6X5 we now had too much voltage, so the value of C4 was reduced to reach 125 volts. I could not hear the difference with the 5AR4 but the customer could (probably because he paid a lot for it), so that is what we will use in the amp. I tried some of my cheap 5AR4's and he could pick out his Mullard even with his back turned so that he could not see which tube I put in. I could not tell the difference.
We tried several alternative power transformer solutions to see if it influenced the sound. I tried using three separate transformers from Mouser Electronics (the cheapest solution) and a few other combinations that were influenced by looks. I could not hear the difference between any of them. He could only "hear" the difference when he could see which ones were connected.
Since this customer cared more about looks than price we wound up using all new Hammond power transformers. He wanted all of the transformers to be the enclosed type. No open frame transformers. The main power transformer was a Hammond 269AX, the choke was a Hammond 193B. These came from Antique electronic supply. We wanted a Hammond 167M5 for the 5 volt filament transformer, but no one had it in stock (it was special ordered). There are plenty of open frame transformers available, like the Hammond 166M5. We wound up using an old power transformer from a TV alignment generator, cutting off the unused windings. It was the same size as the desired Hammond which will replace it when it comes in.
Believe it or not the power transformers are the easy part. The output transformers are more difficult, since no one makes headphone specific output transformers. Headphones come in impedance ranging from 8 ohms to 600 ohms, with 32, 300 and 600 ohms being common. the difficulty arises from the fact that each require different drive voltages. A low impedance set of headphones (8 to 32 ohms) can usually be driven by a conventional output transformer made for driving a speaker. Often the sound can be improved in this case by putting a resistor in parallel with the headphones to optimize the load presented to the tube. I put a 500 ohm dual pot in parallel with the headphones and adjusted for optimum sound.
High impedance headphones are more difficult to drive. A conventional transformer will not deliver enough drive voltage to make much sound with high impedance headphones. There are a few possible solutions available here. There are some output transformers that have a high impedance winding. There are transformers made for driving a "70 volt line" ( the Hammond 1645 is one example). These are usually intended for public address applications so check the frequency response before buying these. There are a few output transformers that have a 600 ohm tap (the Hammond 1650G is one example). It is interesting to note that these transformers are intended for push pull operation and are not air gapped for single ended use. It has been my experience that the very low current used in this amplifier (5 to 10 mA) will not cause core saturation with an output transformer rated for 20 watts or more. I have successfully used larger (60 to 80 watt) push pull transformers with the 45 tube at 30mA.
The problem in this case was that the customer had 3 different sets of headphones in impedances from 40 to 500 ohms. He desired that the amplifier would work with all of them and any other headphones that he may acquire. This would require a transformer with multiple output windings, or multiple transformers. In my usual fashion I set out to test several sets of output transformers. I have a large collection of transformers and I tried about 20 sets of them. The usual single ended transformers work as expected. They give good results with the 40 ohm phones (with a 24 ohm resistor in parallel), but the volume level is very low with the 500 ohm phones even with no loading resistor. Transformers made for driving a high impedance load (vacuum tube interstage transformers) work reasonably well with the 600 ohm phones, but were virtually useless with the 40 ohm phones. They can not deliver the required driving current.
I was considering a two transformer solution. I had planned to use a conventional single ended output transformer for driving low impedance phones and a 600 ohm to 8 ohm transformer (reverse connected) to drive high impedance phones. The other alternative was a transformer like the Hammond 1650G. To test the single transformer theory, I hooked up a pair of UTC LS-57's. These are vintage push pull transformers that have a very good reputation and a matching price tag. I got my set at a flea market for $50 complete with a non working amplifier attached to them. They have two separate primary windings with taps and they also have several output taps from a few ohms to 500 ohms. I tried various primary combinations using clip leads, while he listened. The customer really liked the sound of the UTC's and I was not going to part with my set. I agreed to loan him mine until he found his own pair. He got a set on Ebay (for a lot of money) and they were installed. I installed 3 different headphone jacks, each connected up to different taps on the secondary of the transformer. This allowed a proper match for each of his different sets of headphones. Another user has found his personal audio nirvana.