• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Tube Build for High-Impedance Headphones - the Ubiquitous 6AS7G OTL

L0rdGwyn

Active Member
Forum Donor
Joined
Jan 15, 2018
Messages
233
Likes
534
Hey ASR,

Here is another amplifier I designed. The 6AS7G family OTL headphone amp is everywhere these days. I have a large collection of these tubes, thought I'd do my own version of the circuit, it's now something I build on a select basis for Head-Fi friends, they seem to like it. This type of amplifier is typically paired with high-impedance headphones given the relatively high output impedance of the circuit (e.g., Sennheiser, Beyerdynamic, ZMF).

Nothing innovative, but tried to make what I felt was the best version of this common topology, here is an overview:

-Dual mono power supply, dual secondary toroidal mains, HEXFRED diode SS rectified, properly snubbed
-Time delay circuitry at startup, 30 second delay before application of B+, allowing the tube heaters to fully warm before power is applied and avoids excessive grid-to-cathode voltages on the power tubes (which are direct-coupled)
-Separate Maida style HV regulators for left and right channels with built-in soft start functionality
-50VDC elevation of the heater supply, avoids noise associated with heater-to-cathode leakage and obeys the maximum heater-to-cathode ratings of the Tung-Sol 5998
-Gyrator loaded 6J5 inputs, unbypassed cathode resistors, direct-coupled to output tube grids
-6AS7G / 5998 outputs, sections wired as parallel cathode followers with active loading
-265uF film output caps
-Output bypass circuitry at shutdown, giving the output caps a low DCR path to ground to discharge, as opposed to discharging through the headphones (these DC transients manifest as unpleasant pops at shutdown and can damage the drivers)
-Properly rated fuses and safety ground included ;) (Sorry, couldn't help but poke fun given the current Carver Crimson 275 debacle).

That's the design at a high level, here are some photos of the most recent one I built. This particular build used dual mono V47 Goldpoint stepped attenuators as the owner has a hearing imbalance due to working in the trades.

PXL_20220105_194039336-2.jpg


PXL_20220105_194106925-2.jpg


PXL_20220105_194143948-2.jpg


With the Russian 6J5 and 6AS7G I have been including with the amp.

PXL_20220105_194304213-2.jpg


Here it is in operation with some fancy tubes, GEC L63 and 6AS7G. The amp is dead silent, no hum or noise to be heard even with relatively sensitive headphones.

PXL_20220105_202034824.NIGHT-2.jpg


Here are some measurements with two different output tubes.

Russian 6AS7G
Output Z - 50ohm
1mW into 300ohm - 0.11% THD
1mW into 120ohm - 0.22% THD
1mW into 80ohm - 0.31% THD
1mW into 32ohm - 0.67% THD

Tung-Sol 5998
Output Z - 30ohm
1mW into 300ohm - 0.032% THD
1mW into 120ohm - 0.094% THD
1mW into 80ohm - 0.15% THD
1mW into 32ohm - 0.38% THD

Frequency response at the two extremes.

300ohm

Airmid FR 300ohm 5998.png


32ohm

Airmid FR 32ohm 5998.png



1mW into 300ohm square waves.

1kHz

SDS00002.png


10kHz

SDS00001.png


100Hz

SDS00003.png


1kHz sine wave into 300ohm dummy load just before the onset of hard clipping. So just about 1W out.

SDS00004.png


Think that's it. It's a good sounding headphone amp, I actually won't have one for myself (I have too much tube gear already), but others have been very happy with the sound. Not running a business, just building for friends :)
 

paulbottlehead

Active Member
Manufacturer
Joined
Dec 22, 2021
Messages
209
Likes
834
You may also want to explore putting the time delay on the gyrator. If you add a high value resistor from plate to ground on the driver stage and slow start conduction of the gyrator, you can keep the output tube grid voltage at ground potential for a bit of time while things warm up without having to use relays and deal with potentially audible clicks and pops as things turn on.
 
OP
L0rdGwyn

L0rdGwyn

Active Member
Forum Donor
Joined
Jan 15, 2018
Messages
233
Likes
534
You may also want to explore putting the time delay on the gyrator. If you add a high value resistor from plate to ground on the driver stage and slow start conduction of the gyrator, you can keep the output tube grid voltage at ground potential for a bit of time while things warm up without having to use relays and deal with potentially audible clicks and pops as things turn on.

Thanks for the idea Paul, I'll keep that in mind for future DC coupled circuits, may use that approach if I decide to DC couple my 45 amplifier, which is something I'm considering along with some other changes. This one is all the way done though, the relay-switched time delay works well, no ugly noises on startup. The full circuit was tested on my protoboard last year.

Here is the power supply board. Includes raw DC for both channels, mains transformer snubbers, bleeders, secondary fusing, timer circuitry and relays. When the relays close, the reservoirs connect to the Maida regulators, which have a built-in soft start. With the heaters already warmed and the soft-start B+, everything comes up nicely. At shutdown, the outputs are bypassed so the caps don't discharge through the headphones, all nice and quiet like.
 
Last edited:

paulbottlehead

Active Member
Manufacturer
Joined
Dec 22, 2021
Messages
209
Likes
834
Your scope shots sure to look a lot nicer than mine; maybe I shouldn't have bought an Owon!
 
OP
L0rdGwyn

L0rdGwyn

Active Member
Forum Donor
Joined
Jan 15, 2018
Messages
233
Likes
534
Here is another one, in bronze.

PXL_20220126_005021532-3.jpg


PXL_20220126_005039949-3.jpg


PXL_20220126_005054258-3.jpg


PXL_20220126_005136420-3.jpg


PXL_20220126_005151821-3.jpg


PXL_20220126_011611594.NIGHT.jpg
 
OP
L0rdGwyn

L0rdGwyn

Active Member
Forum Donor
Joined
Jan 15, 2018
Messages
233
Likes
534
Really nice work! Where did you buy the black plates to mount the octal sockets? Or did you have them made custom?

Jack

Thank you! All custom, all done by Dave at Landfall. I do my layout and design chassis in CAD software and shoot him the files. Back in 2020, I was doing powder coating at home, having him send me the raw aluminum chassis. That inspired him to get a powder coating rig, much nicer than mine, so now he does the coating.
 

TriodeLuvr

Member
Joined
Dec 23, 2021
Messages
44
Likes
87
Location
Central Texas
Thanks, that's good to know. I do most of my own CNC milling, but it's very time consuming. I probably won't build more chassis from scratch when the two I still have are used up. One thing I'm curious about - why did you have the custom plates made, rather than mounting the sockets directly onto the chassis? Was this so different socket types could be used more easily?
 
OP
L0rdGwyn

L0rdGwyn

Active Member
Forum Donor
Joined
Jan 15, 2018
Messages
233
Likes
534
Thanks, that's good to know. I do most of my own CNC milling, but it's very time consuming. I probably won't build more chassis from scratch when the two I still have are used up. One thing I'm curious about - why did you have the custom plates made, rather than mounting the sockets directly onto the chassis? Was this so different socket types could be used more easily?

The sub-plates were added only for aesthetic purposes, but they also worked out well from a layout standpoint as I could mount tags / components to the four screws that mount the plates. I've done other chassis with the sockets directly mounted, I think I was going through a Yamamoto copycat phase.
 
OP
L0rdGwyn

L0rdGwyn

Active Member
Forum Donor
Joined
Jan 15, 2018
Messages
233
Likes
534
Building a few more of these for folks on Head-Fi before it's retired. Here's a white one with a three position toggle on the back for two sets of RCA pass throughs.

PXL_20220501_201736657-2 (1).jpg


PXL_20220501_201802159-2 (1).jpg


PXL_20220501_205825911.NIGHT-2 (1).jpg


PXL_20220501_210325867.NIGHT-2 (1).jpg


PXL_20220502_231011161.NIGHT-2 (1).jpg
 
Top Bottom