Yet another headphone amplifier ty.ytka

tytka 13 Nov 2024 22:25 14 3867 Cool? (+22)

📢 Listen (AI):

TL;DR

Built a portable headphone amplifier for phones and laptops, powered from built-in batteries and using only a USB digital input.
The amplifier stage follows the LME49600 datasheet and adds a PCM2705 USB DAC, plus off-the-shelf battery charging and cell-balancing modules.
A four-cell 18650 battery pack provides the supply; the circuits can work from two cells in series, but 18650s were chosen for low price and long runtime.
The charging-module ground must not be tied to the amplifier ground, so an insulating insert was added to the rear charging connector; tests still showed the PCM2705 block degrades measured parameters.

Generated by the language model.

.

I hope I don't bore you, with my next headphone amplifier. But somehow it just so happened that I've had the verve for such projects lately.

This headphone amplifier was created to be used with a phone or laptop. Hence my decision to power it from the built-in batteries (for easier and more versatile use), and to only equip it with a USB digital input.

The schematic of my amplifier itself is overwhelmingly taken from the datasheet of the LME49600 chip. This chip is a very interesting buffer, ideally suited to the construction of such a device. It has been enhanced here with a USB DAC realised on the PCM2705 chip, which is easy to apply and works with various operating systems without problems. In addition, the amplifier has a battery charging circuit, as well as a cell balancing circuit. The latter two, are in the form of off-the-shelf modules. I made this decision mainly because these modules were already lying in a drawer, waiting to be used.

.

The circuits used, in terms of supply voltage, can work when powered by two cells connected in series. I, however, chose to power the device from a battery made up of four cells. And why did I choose these cells? Initially, I was thinking of using smaller sized cells, but as 18650 are probably the most popular, their price is more affordable than the smaller ones. Besides, on these, the amplifier can work for a really long time.

Before I go any further IMPORTANT NOTE for possible role models! .
The ground of the charging module, has a different potential than the ground of the amplifier circuit. They cannot be combined!
For this reason, an insulator has been glued into the charging connector hole located on the rear panel, in order to prevent accidental short-circuiting of the grounds when connecting the charger plug.

.

Back to my design. I have designed the following board:

.

Which I commissioned from a company (I think) in a known part of the world.

As for the casing. I think you will already recognise it in part. The main part of it is taken from one of my previous projects, which underwent a metamorphosis some time ago and received a new enclosure. So, these I have used here to make useful.
In completing the project on the mechanical side, a few more small components were needed. These include:

.

In order not to bore you with descriptions, hereafter just a few photos taken during the project.

.

Below I still include the results of the tests carried out. Looks to the fact that, as in the design of my amplifier in the mains-powered version, the converter block from the PCM2705 affects the measured parameters in a negative way. i will still review this carefully.

.

Greetings and feel free to comment.

About Author

tytka wrote 752 posts with rating 1831 , helped 8 times. Live in city Pabianice. Been with us since 2006 year.

Comments

Add a comment

-rafal- 14 Nov 2024 20:30

I have read the whole description and looked at the schematic and I still don't know what the circuits on the output are :) . Why USB type A? I don't think this is their typical use (being an "input")... [Read more]

tytka 14 Nov 2024 21:09

Sorry, the symbol for the output circuits will not appear on the schematic, but they are listed in the description, it is LME49600. And why USB-A? Due to lack of space, I couldn't manage to use a dedicated... [Read more]

remiorn 14 Nov 2024 23:12

I wonder why such poor performance. Over 20 years ago I made a similar circuit, only on a PCM2900 chip. And the parameters were much better, inter-channel transfer -90dB, noise -96dB, which is practically... [Read more]

djfarad02 16 Nov 2024 11:30

It is not known how it was measured. I presume some other card of unknown grade. [Read more]

tytka 16 Nov 2024 14:43

. Just out of curiosity I measured the DAC with the PCM2902. This is what the Behringer UCA222 I own is made on. I did this, the USB input from the UCA222 plugged into the computer, and its analogue... [Read more]

djfarad02 16 Nov 2024 14:47

I used to apply Wolfson's WM8776 cubes. Theoretically also with excellent performance. Equally, I encountered the problem of quite high noise levels. The chips were from aliexpress - maybe rejects. [Read more]

tytka 16 Nov 2024 15:08

Yours I cannot rule out. Maybe, in fact, this is the cause here. [Read more]

andrzejlisek 16 Nov 2024 15:21

Very nice, well made and certainly satisfying, so the purpose of the construction was fully achieved. However, at first glance it seems that, since this is an amplifier for headphones, its dimensions... [Read more]

acctr 16 Nov 2024 16:09

. What is the point of using an amplifier in such surroundings? The noise from outside nullifies the quality obtained from it. And those who want to cut out the noise use headphones with active noise... [Read more]

andrzejlisek 16 Nov 2024 16:38

. The only, and primary, point is to "shout out" the noise of the vehicle or ship during the journey. Even a 'muted' train inside is in fact loud. The headphones, were the cheapest ones, pressed into... [Read more]

remiorn 17 Nov 2024 12:08

. My experience with these circuits was that good PCB design and manufacture was critical. A circuit with improperly routed ground was able to exhibit noise levels some 20-30 dB higher. And of course... [Read more]

acctr 17 Nov 2024 12:13

. This is a very bad method because only hearing suffers. When I used to travel on Polish rolling stock I used earplugs and over that earphones. [Read more]

_J23 20 Nov 2024 23:14

The amplifier looks very good, and the DAC measurements suspiciously poor. Isn't packing batteries into such a simple design an excess of form over substance...? If it were the headphone amplifier itself... [Read more]

tytka 21 Nov 2024 14:50

The PCM2702 chip, has not been in the TI range for some time now. So it is also more difficult to obtain it. Yes my amplifier, especially with those rather large batteries, may seem like an overdesigned... [Read more]

FAQ

TL;DR: With 4×18650 cells, this DIY USB headphone amp targets phone and laptop use, but the builder found the USB DAC section dragged results down; as one commenter put it, "good PCB design and manufacture was critical." This FAQ helps DIY audio builders choose the output stage, battery setup, USB connector, and measurement method while avoiding ground and layout mistakes. [#21306482]

Why it matters: This thread shows that a headphone amp can work well in practice while still measuring badly because of DAC implementation, grounding, and test setup.

Option	What the thread says	Main trade-off
USB-A input	Chosen because the case had no room for USB-B and to avoid confusing it with the charging socket	Unusual as a device-side input
USB-B input	Suggested as the more typical "printer" style input	Better convention, but too bulky here
USB-C input	Suggested as a single modern connector	Risk of confusion with the charging port
4×18650 supply	Chosen for long runtime and low cost because 18650 cells are common	Large enclosure

Key insight: The analog amplifier section measured better than the full USB DAC plus amp path. In this build, the PCM2705 stage, its power supply, grounding, and the measurement chain were the main suspects, not the LME49600 output buffer alone.

Quick Facts

The amplifier was built for phone or laptop use, powered from an internal battery pack, and fitted with only a USB digital input for easier portable use. [#21302297]
The design uses 4×18650 cells even though the circuits can run from 2 cells in series; the stated reasons were longer runtime and lower cost due to 18650 popularity. [#21302297]
The builder gave a hard safety warning: the charging module ground and amplifier ground are at different potentials and must not be connected; an insulating insert was added at the rear charging connector. [#21302297]
Reported reference figures from a similar older USB audio build were about -90 dB inter-channel transfer and -96 dB noise, near what the commenter considered feasible for a 16-bit converter. [#21303723]
One commenter reported that poor ground routing alone can raise noise by 20–30 dB, and moving a 100 nF capacitor a few millimetres worsened signal-to-noise ratio by a few dB. [#21306482]

What is the LME49600 and why is it used as the output stage in a headphone amplifier?

The LME49600 is the headphone amplifier’s output buffer, chosen because the builder considered it well suited to this kind of device. The schematic was taken largely from the LME49600 datasheet, and the chip forms the final drive stage after the USB DAC. In this thread, it matters because the analog stage alone measured better than the full DAC-plus-amp path, so the output stage was not the only suspect. [#21302297]

How does the PCM2705 USB DAC affect the measured performance of a headphone amplifier compared with the analog stage alone?

It worsened the measured results compared with the analog stage alone. The builder explicitly wrote that the PCM2705 converter block negatively affected the parameters, then later showed improved results after tweaking the PCM2705 supply and separate tests of the amplifier without the PCM stage. That points to the DAC section, its supply, or its integration as the main measurement bottleneck. [#21305432]

Why would a headphone amplifier powered from 18650 cells use four batteries instead of just two in series?

It uses four 18650 cells to gain much longer runtime and keep cell cost low. The builder stated that the circuits can work from two series cells, but chose a four-cell pack because 18650 cells are very common, cheaper than smaller formats, and let the amplifier work for a really long time. The trade-off is a much larger enclosure. [#21302297]

How should I isolate the charging connector ground from the amplifier ground in a battery-powered headphone amp?

Keep the charging module ground physically separate from the amplifier ground and prevent any accidental contact at the connector. The builder warned that the two grounds sit at different potentials and cannot be combined. To avoid a short through the plug body, an insulator was glued into the rear charging-connector hole. That is the thread’s clearest failure-case warning. [#21302297]

What is a cell balancing circuit and why is it important when charging a 4-cell 18650 battery pack?

A "cell balancing circuit" is a battery-management circuit that equalizes the charge state of cells in a series pack, preventing one cell from drifting away from the others during charging. In this build, it was added alongside the charger as an off-the-shelf module because the battery pack uses 4 cells, not a single cell. The thread presents it as a necessary part of the charging arrangement. [#21302297]

Why did this headphone amplifier use a USB Type-A input instead of USB-B or USB-C?

It used USB Type-A because the enclosure had no room for a proper USB-B device connector and the builder wanted to avoid confusing USB-C with the charging socket. The batteries occupy almost the full case width, so connector size drove the choice. This was a packaging decision, not a claim that Type-A is the ideal device-side USB audio connector. [#21303535]

USB-A vs USB-B vs USB-C for a DIY USB DAC/headphone amp input — which connector makes the most sense and why?

USB-B makes the most conventional sense, but this thread shows that mechanical constraints can override convention. One commenter said USB-B would fit better as an input, while USB-C could unify the interface. The builder rejected both: USB-B would not fit, and USB-C could be mistaken for the charging port. USB-A worked here because the case was dominated by batteries. [#21303535]

How can I improve noise, channel separation, and overall measurements in a PCM2705 or PCM2902 based DAC?

Improve the DAC supply, shorten critical grounding paths, and verify the board layout before changing chips. In this thread, the builder got some improvement after improving the PCM2705 supply. Another commenter said bad ground routing can add 20–30 dB of noise, and even moving a 100 nF capacitor a few millimetres hurt signal-to-noise ratio. Those are direct, practical fixes. [#21306482]

What measurement setup should I use to reliably test DAC noise and crosstalk without the sound card skewing the results?

Use a known reference path first, then compare the DUT under the same wiring. 1. Measure your sound card alone as a baseline. 2. Connect the USB DAC output to the computer’s line input. 3. Compare both traces before judging the DAC. The builder did exactly that with a Behringer UCA222 and also measured the reference card itself, because commenters questioned whether the measurement card was limiting the result. [#21305432]

Why can PCB layout and ground routing worsen DAC noise by 20–30 dB in audio circuits?

PCB layout can worsen noise by 20–30 dB because small ground-path and decoupling changes directly affect sensitive DAC nodes. One commenter described seeing that scale of degradation from improperly routed ground. The same person said moving a 100 nF capacitor only a few millimetres away from the IC worsened signal-to-noise ratio by a few dB. In audio DAC work, millimetres matter. [#21306482]

How close should 100 nF decoupling capacitors be placed to PCM2705 or similar audio ICs for best signal-to-noise ratio?

Place the 100 nF decoupling capacitor as close to the IC as physically possible. The thread gives a concrete warning: moving a 100 nF capacitor just a few millimetres away worsened signal-to-noise ratio by a few dB in a similar design. That is not a theoretical point here; it was reported from hands-on testing during a thesis project. [#21306482]

What is galvanic isolation in USB audio and how can it help with ground loops in DAC measurements?

"Galvanic isolation" is an isolation method that passes data while breaking direct electrical ground continuity, reducing loop currents between connected devices. One commenter suggested inserting galvanic isolation in the USB path because the DAC is still powered from USB, and a ground loop could be depressing the measured DAC performance. In this thread, isolation is proposed specifically as a measurement and troubleshooting aid. [#21311948]

PCM2705 vs PCM2702 vs PCM2900 vs PCM2902 — which chip is better for a simple DIY USB headphone amplifier?

This thread does not identify a universal winner, but it does rank them by practical experience and availability. A commenter recalled better figures from PCM2900, about -90 dB crosstalk and -96 dB noise. Another suggested PCM2702 might perform better, but the builder replied that PCM2702 has been out of TI’s range for some time and is harder to obtain. PCM2902 was tested here through a Behringer UCA222 and still measured poorly in this setup. [#21312709]

How do I design a portable headphone amplifier for use with a phone or laptop without making the enclosure too large?

Keep the power system simple and define “portable” before you design the case. This build was never meant for walking use; it was meant to be fairly easily portable at home or away, while remaining power-independent. The large 4×18650 pack delivered that goal, but commenters noted the enclosure was too large for travel use and suggested smaller battery choices or USB-powered operation. [#21312709]

What are the most likely causes of suspiciously poor DAC measurements in a DIY headphone amp powered by batteries but fed from USB?

The most likely causes are the USB DAC stage itself, power-supply quality, grounding, PCB layout, and the measurement chain. The builder suspected the PCM2705 block early, improved results by working on its supply, and showed the analog section alone behaved better. Commenters added two more causes: sound-card-limited testing and USB ground loops. One concise diagnosis was, “good PCB design and manufacture was critical.” [#21305432]

Generated by the language model.