A month ago I started thinking about rebuilding my good old headphone amplifier with TPA6120A. I didn’t realize it was that old until I decided to translate the original article in English. It’s been 7 years and I though it was 3 or 4. Time passes quickly, it seems. I spent some time thinking how to improve it. Now the PCB and components are ordered and once I get them, assembly will take place in the same chassis. What’s new and improved? A number of things in fact.
The list of updates is as follows:
- Single PCB to include all blocks except input and output connectors
- Encapsulated toroidal transformer for PCB mounting
- Double the capacitors in the regulator (2x2200uF for each polarity)
- All electrolytic capacitors are now Panasonic FR series (newer and better than FC)
- All 100nF decoupling capacitors are now polyester WIMA MKS02 (smaller, allowing them to be closer to TPA6120A). Not as great as the polypropylene Panasonic, but more than sufficient
- ALPS Blue Velvet potentiometer
- New heatsinks for the voltage regulators
- All possible components switched to SMD
- Rectifier diodes changed to 2A model
- Delay On circuit has now its own voltage regulator
- Delay On circuit error fixed – output was switched off when the large capacitors discharged, now this happens when the switch is turned off (well, around 50ms later)
- Delay setting trim potentiometer is replaced with Bourns precise model for easier setting of desired delay
- Power LED will now turn on and off together with the headphone relay
- The new PCB has option for 115V in addition to 230V (supported by the new transformer)
- New volume knob
Let’s take a closer look at the design decisions made, updates and why they were needed.
Let’s start with the schematic. It is almost the same as the original project. I won’t describe every single component in details. You can find BOM file attached at the end of the article with more information.
The original design included toroidal transformer, mounted with a screw to the chassis. I wanted to mount it on the new PCB. I was surprised when I opened the chassis. The fire resistant insulation on the transformer’s input and output wires had started melting at room temperature (never sensed any heat coming from the amplifier). I don’t know what was used for impregnation of those braided insulation tubes, but it is literally dripping on the PCB and is sticking the cables to the chassis. Otherwise the transformer is just fine. I know I can replace those and clean up the transformer, but I decided to go for an encapsulated toroidal transformer. It is easier to mount on a PCB, has better shielding and looks better. The transformer I picked has double primary winding so it allows for usage with both 115V and 230V. It is Chinese YHDC brand. 2x12V, 35VA. Basically the same parameters as the old transformer.
The old design had an analogue supply and a singe zener diode to provide the delay-on circuit with the required voltage. This time I decided to use completely separate supply block. It uses the same transformer though. I will start with the smaller supply block – the delay circuit one. It is single +5V supply so I used only 2 diodes to rectify the AC voltage. Those are 2A Schottky diodes. A 100μF capacitor follows. Large enough to ensure clean output voltage for the relay. The value was found by simulation in worst case condition. The circuit will work without that capacitor, but then fluctuation will be present in the regulator output as well as in the relay coil, which is close to the contacts and the sound path. The voltage regulator is 78L05, providing up to 100mA (more than enough for the relay and LED). The two smaller capacitors are as recommended in the datasheet.
The analogue supply has been changed as well. It uses the same Schottky diodes as the smaller supply, 2A/40V. The electrolytic capacitors are doubled now, 2×2200μF for each pole. This time I will use the newer FR series from Panasonic. The have similar parameter, but longer life. In fact all electrolytic capacitors in the schematic are the same type. Another change in the analogue supply are the heatsinks for the voltage regulator ICs. They have a solder pin for securing the position of the ICs. Regulator are again LM317 and LM337, using fixed resistor for setting the output voltage for lower noise. The 10μF capacitors, connected to the adjustment pins of the regulators are also ensuring low output noise in case of peak power drain at the output.
The delay circuit is almost the same as before in terms of component values. All components except the trim potentiometer and the time constant capacitor are SMD. The trim potentiometer is Bourns precision model with a small screw for setting the value. It is 20kOhms compared to 22kOhms in the old design. I just didn’t find the same value. This will reduce a little the maximum delay possible, but it will still be more than 2 seconds, which is more than enough.
Functionally, the circuit it very simple. The 555 timer IC generates a step at its output when the capacitor C12 is charged. This turns on the relay with a delay. After that the circuit is not generating anything, so no noise can be generated from it.
The previous version had a bug in that circuit. It was switching on the output with the desired delay, but it was also having a delay when switching it off. This might produce undesirable audible effects. The reason for that was that this circuit was drawing its power from the same point as the regulators for the analogue block, after the large electrolytic capacitors. This was fixed by adding a separate rectifier.
The last change in this block is the power indicator LED. It is now connected in parallel with the relay so it will show when the output is enabled.
Last but not least is the amplifier block. There are a few updates here as well. The first design change I made were the 100nF decoupling capacitors. The ones I have used in the old design were great, but they were just an overkill, since they are not in the sound path. They were also quite large, which made it difficult to place them next to the power supply pins of TPA6120A. This is an important requirement, described in its datasheet. For that reason I decided to replace those capacitors with smaller ones, which will be as close to the pins as SMD ones. I chose WIMA MKS02 series. These are low voltage (63VDC), which allows them to be so small (2.5mm). They are polyester, while the older ones were polypropylene. They seemed to fit the best into the requirements.
I didn’t like my volume potentiometer at low levels, so this time I decided to go for a better one. I chose ALPS Blue Velvet series. It uses the same technology (conductive plastic), but is much larger (probably 10 times). This should solve my issue with the unbalanced levels in left and right at very low levels. At least in theory.
Another addition are a couple of 1kOhm resistors, connected to the normally closed contacts of the relay. This will provide a small load for the amplifier even when headphones are not connected, preventing saturation of the output when no input signal is present. This will be useful when connecting headphones after the amplifier was turned on.
The same signal connectors will be used as before. Those are panel mounted RCA and 6.3mm TRS. The schematic shows connector on the PCB, which I will not mouth and solder the wires directly instead. I want to keep the number of connections at minimum. This will make the chassis assembly a little more tricky, but I don’t do that many times anyway.
I have mentioned as point 1 in the list at the top that there is brand new PCB design. A single PCB will have all the components except the signal connectors, the mains power connector and switch and the fuse. Those will not move from where they were mounter previously. Let’s have a look at the brand new PCB:
The first thing that catches the eye is the separate ground planes. There is one for the analogue supply, one for the amplifier and another one for the delay-on circuit and its own supply. A closer look will reveal a few details. The analogue supply plane extends along the amplifier plane to separate it from the transformer and power lines on the bottom side of the PCB. The required connection between those two ground planes is made on the current path. This is between small electrolytic capacitors and the decoupling ones next to the TPA6120A. This should reduce the noise in the amplifier ground plane to minimum. The third ground plane for the delay-on circuit is at a larger distance from the other two just to make sure it does not interfere in any way. Another fact that has to be mentioned here is that this plane is connected directly to the transformer’s middle point (pin 12). This is as good separation from the analogue part as there can be. You can also see that all ground planes have many vias, connecting the top and bottom ground planes of each block. This minimizes the parasitic inductance in the ground planes as well as the parasitic capacitance between the two sides of the board.
In the new design I decided to use arcs instead of 45° corners for the tracks in the amplifier block. This is a technique used in RF design and high frequency digital design, but it is also popular to some extent in the audio world. I personally doubt that it has any audible effect in such a small circuit, but at least it looks nice. 🙂 I also tried to make circuit as symmetrical as possible for both channels. Unfortunately the chassis I have isn’t large enough to allow me to make a perfectly symmetrical design.
I shall look again closely at the amplifier area. The design is very similar to the old one. The datasheet of TPA6120A describes in detail the most appropriate PCB layout approaches. I tried to follow those as much as possible. Those include the following: short traces between resistors and pins, same for the decoupling capacitors, no ground plane under the pins to avoid parasitic capacitance, thermal dissipation area on both sides (especially for the DWP package and high output power designs). The calculation and selection of resistor values is also described in detail in the datasheet, so I won’t repeat it here. All resistors are Vishay MELF 0204 SMD except the input 51Ohm resistors. Those are 0102 size, because the larger ones were not available when I placed the order. Also those are not loaded with much current, so there is no need to use larger ones. Tolerance is 1% for most and 0.1% for the feedback 1kOhm resistors. Temperature coefficient is 50ppm for the 1% resistors and 25ppm for the 0.1% ones.
One last thing to mention here is the mains input connector. It is meant for both 115V and 230V. The PCB has those connections marked as well as the neutral line.
At the moment I am waiting for PCB manufacturing and components delivery from China. This will take some time. When I have all the parts, I will assemble and post another article about the finished product with more photos.
As I promised in the beginning of the article, the BOM file. I can’t seem to be able to share zip files here, so if you are interested in making your own PCB, take a look at this same project on EasyEDA, where you can even order PCBs. You will see 3 PCB versions there. That is how the design evolved. Use the last version.
I am sure the licence note on the right side of this web page is well visible, but let me remind you the following:
This project has been developed and released as an open source by Stefan Penov.
Licensed under Creative Commonns Attribution-NonCommercial-ShareAlike 4.0 License: https://creativecommons.org/licenses/by-nc-sa/4.0/
Under this licence you as a user are free to share and adapt this work under the following terms:
>>> Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
>>> NonCommercial — You may not use the material for commercial purposes.
>>> ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.
>>> No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.