- Joined
- Jan 15, 2018
- Messages
- 294
- Likes
- 674
The DarkVoice 336SE is one of the more popular entry-level tube headphone amplifiers, it was reviewed on this site to much criticism. I was introduced to a group of individuals on Head-Fi who were making component changes / modifications to their own circuits. I thought it would be a fun project to do my own "modifications", a nice Head-Fier sent me his spare stock DarkVoice to do with what I pleased.
Here is a photo of the stock circuit.
A quick description: it uses a passive CRCRC-CRC supply. The 6SN7 input tube is loaded with a plate resistor and cathode biased, the cathode resistor is unbypassed. The 6SN7 plate is then direct-coupled to the 6AS7/6080 grid. The 6AS7/6080 cathode follower is biased with a 1K 25W wirewound resistor. The output is capacitor coupled with 30uF per channel, giving a f-3dB point of 17Hz with 300ohm headphones and an output impedance of roughly 75ohms. As such, using anything less than 300ohm headphones leads to significant LF rolloff.
I thought the circuit would benefit from a regulated power supply, an active load on the 6SN7, as well as other various "fixes" based on what I discovered inside.
Here is a quick summary of the changes I made:
1) Removed CRCRC-CRC filter from the power supply - replaced with single 330uF resevoir cap and Maida style HV regulator. This allowed an increase in the B+ to 180VDC from ~140-150VDC in the stock design with much better PSRR.
2) Removed plate resistor load from 6SN7 input tube - replaced with DN2540 - J310 "gyrator" load. This fixes the plate voltage of the 6SN7 and allows the tube to find its plate current based on the bias. The reference voltage is generated via a LND150 cascode CCS across a resistor to ground. I designed the PCBs to mount on the existing standoffs from the final CRC of the passive supply.
3) Removed 6080 1K wirewound cathode resistors - replaced with Vishay 1.2K non-inductive wirewound resistors, adjusting the bias point to account for the higher B+ and maintain the bias current of the original design.
4) Removed 6SN7 1K cathode resistors - replaced with KOA Spear 300ohm carbon film resistors, again to adjust the bias.
5) Created a virtual center tap via 2x 100ohm resistors on the heater supply. Elevated the heater supply 50VDC above ground. Added twisted heater leads (previously straight, solid core 18AWG wire). This cured the well-known hum issues of the amplifier, as the original designer simply grounded one leg of the heaters.
6) Increased the output capacitance from 30uF per channel to 66uF per channel. This gives a f-3dB point of 20Hz with 120ohm headphones.
7) Power LED is run from the heater supply. As such, rerouted the power LED ground to the heater supply center tap to float it on the elevated supply, adjusted current limiting resistors for the halved voltage due to the newly balanced supply.
Here are the new tube bias points:
B+ 180VDC
6SN7: 80Va / 4.5mA / 1.4Vk
6080: 75Va / 85mA / 105Vk
There are a few other things to do (startup protection diodes on 6AS7/6080 grid, replace volume pot, replace the power switch per the request of the original owner), but overall the circuit is more or less complete.
The sound is drastically improved and 6SN7 tubes that were previously unusable due to almighty 60Hz hum are now completely silent.
Not really a DarkVoice anymore, but it sure sounds good might build an improved version of this circuit in my own chassis without the size limitations.
Here is a photo of the stock circuit.
A quick description: it uses a passive CRCRC-CRC supply. The 6SN7 input tube is loaded with a plate resistor and cathode biased, the cathode resistor is unbypassed. The 6SN7 plate is then direct-coupled to the 6AS7/6080 grid. The 6AS7/6080 cathode follower is biased with a 1K 25W wirewound resistor. The output is capacitor coupled with 30uF per channel, giving a f-3dB point of 17Hz with 300ohm headphones and an output impedance of roughly 75ohms. As such, using anything less than 300ohm headphones leads to significant LF rolloff.
I thought the circuit would benefit from a regulated power supply, an active load on the 6SN7, as well as other various "fixes" based on what I discovered inside.
Here is a quick summary of the changes I made:
1) Removed CRCRC-CRC filter from the power supply - replaced with single 330uF resevoir cap and Maida style HV regulator. This allowed an increase in the B+ to 180VDC from ~140-150VDC in the stock design with much better PSRR.
2) Removed plate resistor load from 6SN7 input tube - replaced with DN2540 - J310 "gyrator" load. This fixes the plate voltage of the 6SN7 and allows the tube to find its plate current based on the bias. The reference voltage is generated via a LND150 cascode CCS across a resistor to ground. I designed the PCBs to mount on the existing standoffs from the final CRC of the passive supply.
3) Removed 6080 1K wirewound cathode resistors - replaced with Vishay 1.2K non-inductive wirewound resistors, adjusting the bias point to account for the higher B+ and maintain the bias current of the original design.
4) Removed 6SN7 1K cathode resistors - replaced with KOA Spear 300ohm carbon film resistors, again to adjust the bias.
5) Created a virtual center tap via 2x 100ohm resistors on the heater supply. Elevated the heater supply 50VDC above ground. Added twisted heater leads (previously straight, solid core 18AWG wire). This cured the well-known hum issues of the amplifier, as the original designer simply grounded one leg of the heaters.
6) Increased the output capacitance from 30uF per channel to 66uF per channel. This gives a f-3dB point of 20Hz with 120ohm headphones.
7) Power LED is run from the heater supply. As such, rerouted the power LED ground to the heater supply center tap to float it on the elevated supply, adjusted current limiting resistors for the halved voltage due to the newly balanced supply.
Here are the new tube bias points:
B+ 180VDC
6SN7: 80Va / 4.5mA / 1.4Vk
6080: 75Va / 85mA / 105Vk
There are a few other things to do (startup protection diodes on 6AS7/6080 grid, replace volume pot, replace the power switch per the request of the original owner), but overall the circuit is more or less complete.
The sound is drastically improved and 6SN7 tubes that were previously unusable due to almighty 60Hz hum are now completely silent.
Not really a DarkVoice anymore, but it sure sounds good might build an improved version of this circuit in my own chassis without the size limitations.