Recovery and restoration service:frontea online,corp.
‹ 2026/01/15 ›
Good evening. Recently, for late‑night listening in the bedroom, I’ve been using the PopoDAC almost exclusively — it makes it easy to fall asleep. Sweat
As for speaker listening, the current PA setup sounds great… but sometimes I feel like listening with headphones too.
However, I don’t have a proper headphone amplifier at home… Sweat
So, I’ve decided to build a headphone amplifier that can fully bring out the capability of the PopoDAC.
The target headphones are my familiar AKG K701, and I’ll tune the drive feel to match them.
Let’s quickly define the design goals.
Since the source is the PopoDAC, the main requirements are:
a) Extend the audio bandwidth up to just under ~100 kHz so the energy feel can be enjoyed
b) Reproduce the PopoDAC’s MC Profile sound quality faithfully
c) Avoid excessive “amp coloration”
d) Designed to drive ~60 Ω headphones
e) Allow ±9 V to ±16 V supply operation for convenience and tonal variation
f) (Excluded here) No attempt to reconstruct headphone soundstage
So we end up with five requirements (one excluded for now).
“Considering the K701 at the center, an output of around 2–6 Vpp * 120 mA seems appropriate.”
“Since the main source is the small PopoDAC, I want the enclosure to be as compact as possible.”
Possible headphone amp configurations:
1.Fully discrete
2.Op‑amp buffer IC
3.Op‑amp discrete output stage
Option 2 would be the safe choice, but…
“Hm? Hmm? Unfortunately, the LME49600 can’t be bought easily! And I can’t find BUF chips either!” ><
So I’ll go with option 3. ^^;
The overall structure is:
Front stage = op‑amp voltage gain
Rear stage = bipolar transistor push‑pull current gain
Wrapped together with global negative feedback as a composite amplifier.
(Of course, at this level it’s Class‑AB.)
Here is the schematic — this is the final version.
(Some parts differ from the actual selected components, but the overall structure is unchanged.)
The distinctive part this time is the Vbe multiplier section (Q1, Q2).
The relevant area is the group of transistors and resistors between the OPA output and the push‑pull output pair.
I want to choose between ±9 V, ±12 V, and ±16 V supplies depending on mood and usage, and also allow wiper adjustment around ±12 V to tune the “texture” (idle current).
This gives me both convenience and tonal flexibility — satisfying preferences (b) through (e).
Explaining using the upper channel in the schematic:
I separated the portion that injects (feedback) current through R15 (R16), and the bias‑adjustment portion consisting of R11, R1, RV1, and Q1.
Simulating with the resistor values shown in the schematic, the output‑stage idle current falls within the following ranges:
1.±9 V supply → ~10 mA (RV1 wiper at .3)
2.±12 V supply → ~10 mA (RV1 wiper at .5)
3.±16 V supply → ~10 mA (RV1 wiper at .7)
The RV seems usable with just the right amount of adjustment. ^^
For many people who build amps like this, this area is probably the “fun but troublesome” part.
Once you lock in the bias, it’s fixed — essentially “set in stone” — so when you get bored, you end up thinking, “Alright, time to build another one…”
Since I want to enjoy and experiment until it breaks, I added this variable feel.
By the way, the LT1010 datasheet shows a standard example using LM334 for bias adjustment.
This time, I tried to make a reasonably adjustable discrete version — but honestly, if I had more budget and procurement power, I’d prefer to rely on an IC. Sweat
Anyway, I tested a breadboard prototype with R11 and R12 removed (listening at low volume), and it sounded good, so I’ll move on to PCB design. ^^;
(The op‑amp is NE5532…)
If it were IC‑to‑IC, I would have soldered everything onto a universal board, but with 3 transistors × 2 channels, I decided to use a PCB.
(As you can see in the breadboard sample photo, the transistors point in all sorts of directions — I didn’t feel confident arranging them neatly on a universal board… ><)
The reasons are:
For thermal compensation, I want clean thermal coupling using a heat panel
As a substitute for overload protection… Sweat
(I’ll probably blow it up eventually from having too much fun, so I’ll order multiple copies — that’s my “protection policy,” haha)
The PCB layout has relaxed spacing and looks beautiful for DIY. ^^
I finished the Gerber files and ordered from JLCPCB.
One more PCB feature:
I used dual‑side pour ground.
Top side = analog lines AGND
Bottom side = power lines PGND
AGND and PGND join at a single point using the 0 Ω resistor shown in the schematic, effectively connecting at the output stage.
(In actual assembly, it will be shorted.)
(The model photo shows parts used only for size matching, so some components look odd.)
That’s all for now — next time will be assembly once the PCB arrives.
Stay tuned♪
