Huginn, High performance, Class A, Lateral FET headphone amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Started with this : http://www.diyaudio.com/forums/head...n-lateral-fet-output-headphone-amplifier.html

But I only did some simulations and played around with a layout before I stopped working on it.

However, I went back and had a look at it again and implemented some big changes and I ended up with this : http://www.diyaudio.com/forums/head...n-high-quality-class-lateral-fet-headamp.html

This amplifier module I actually built and spent a few weeks listening to it but I did not get around to make a finished amp and there was still some things I wanted to change.

Well, changes have been done and this is where I am today. See attached pictures for final schematic and layout.

The schematic does not show it, but as one can see on the layout there are 3 connectors on the board, for the inclusion of a DC servo on a separate PCB.

This is due to the fact that it makes it easy to choose between servo or non servo operation and it also makes it easy to experiment with different types of DC servo circuits without having to redo the whole HAB.

Also not shown on the schematic, the input JFETs are high frequency, low noise BF862, input cascodes are J111, current mirror is BCM62B, all from NXP.

Simulation results are incoming.
 

Attachments

  • Huginn-Schematic.png
    Huginn-Schematic.png
    47.3 KB · Views: 697
  • Huginn-Layout.png
    Huginn-Layout.png
    128 KB · Views: 693
Last edited:
Some simulation results :

5Vrms into 32 Ohm, THD20 : 0.001552%

5Vrms into 150 Ohm, THD20 : 0.000481%

5Vrms into 250 Ohm, THD20 : 0.000475%

5Vrms into 600 Ohm, THD20 : 0.000471%

It is stable driving up to 20nF capacitive loads, more than enough for any headphone/headphone cable available.

Attachments :

1 : Frequency response, 250 Ohm load.

2 : SquareWave, 20kHz, 2VPP Input, 10nS Rise/Fall time.

3 : SquareWave, 20kHz, 10VPP Input, 10nS Rise/Fall time.

4 : SquareWave, 100kHz, 2VPP Input, 10nS Rise/Fall time.

5 : SquareWave, 100kHz, 10VPP Input, 10nS Rise/Fall time.
 

Attachments

  • Huginn-Frequency Response, 250 Ohm.png
    Huginn-Frequency Response, 250 Ohm.png
    13.9 KB · Views: 612
  • Huginn-SquareWave-20kHz-2VPP-Input.png
    Huginn-SquareWave-20kHz-2VPP-Input.png
    14.3 KB · Views: 588
  • Huginn-SquareWave-20kHz-10VPP-Input.png
    Huginn-SquareWave-20kHz-10VPP-Input.png
    14.1 KB · Views: 575
  • Huginn-SquareWave-100kHz-2VPP-Input.png
    Huginn-SquareWave-100kHz-2VPP-Input.png
    22.3 KB · Views: 106
  • Huginn-SquareWave-100kHz-10VPP-Input.png
    Huginn-SquareWave-100kHz-10VPP-Input.png
    19.1 KB · Views: 106
Last edited:
Hi Neutrality,

The project looks great, congratulations. Maybe it is just me but I have problems interpreting your pcb color scheme. Where do you find the NXP products? Here in France, I could not find them @ my usual suppliers and Mouser does not seem to carry those either.

Jacques

You can get them from from rsonline.dk(rs-components.com), however they should be available from digikey.com as well.

Attached is bottom layer and top layer.
 

Attachments

  • Huginn-GBL.png
    Huginn-GBL.png
    23.7 KB · Views: 141
  • Huginn-GTL.png
    Huginn-GTL.png
    17.7 KB · Views: 124
Last edited:
Cooling, I will be dissipating 3.2 Watt in the heatsinks.

With the holes in the planes I can now put holes in the bottom under the PCB and in the top over the PCB, this will result in improved cooling compared to no holes in the planes.

Simple really. :)

To expand on this, PCB mounted heatsinks usually rely on natural convection, which basically means that if you get an increase in the airflow around the heatsink and within the heatsink fins you also increase the amount of power you can dissipate in the heatsink.

Cooler air is sucked into the cabinet through the holes in the bottom cover, through the PCB planes holes, through the heatsink fins and then leaves the cabinet through the top cover holes above the heatsink. This gives you a chimney effect.
 
Well at least you have a rationale behind it. And using convection like that is a wonderful thing: a system that tends to self-regulate. i.e. The hotter the heatsinks get, the greater the convection effect gets, causing more airflow, cooling the heatsinks more. It's a beautiful thing.

The reason I asked is that since each hole makes its plane into a loop, with a significant enclosed loop area (see Faraday's Law), every time-varying magnetic field will induce a corresponding time-varying current in the loop. It might not be a significant concern, since it's not an "obviously bad" loop like it would be if it was a loop that included, say, the amplifier input resistor. Nevertheless, the induced currents will induce voltages across the impedance of the copper itself, which will present themselves between whatever pairs of connections can be found connected to the loop.

The two possible problem scenarios, if there would even be any problem, would probably come from your transformer and its primary and secondary leads, since the fields associated with those could be relatively large in amplitude because of their close proximity, and, from RF (radio frequency) fields from external sources, since they are high frequencies, to which the copper will be a much more-significant impedance (although the RF's fields are usually mostly electric fields, rather than magnetic, after they get more than one or two wavelengths from the transmitting antenna).

I don't have any idea about whether or not any possible effects from those might be significant. Just wanted to mention the theoretical possibility.

If there were to be a problem, I guess you could always simply solder some copper mesh (screen) across the holes, to eliminate the enclosed loop area, eliminating the induced currents, while still allowing the airflow.
 
Last edited:
Well at least you have a rationale behind it. And using convection like that is a wonderful thing: a system that tends to self-regulate. i.e. The hotter the heatsinks get, the greater the convection effect gets, causing more airflow, cooling the heatsinks more. It's a beautiful thing.

The reason I asked is that since each hole makes its plane into a loop, with a significant enclosed loop area (see Faraday's Law), every time-varying magnetic field will induce a corresponding time-varying current in the loop. It might not be a significant concern, since it's not an "obviously bad" loop like it would be if it was a loop that included, say, the amplifier input resistor. Nevertheless, the induced currents will induce voltages across the impedance of the copper itself, which will present themselves between whatever pairs of connections can be found connected to the loop.

The two possible problem scenarios, if there would even be any problem, would probably come from your transformer and its primary and secondary leads, since the fields associated with those could be relatively large in amplitude because of their close proximity, and, from RF (radio frequency) fields from external sources, since they are high frequencies, to which the copper will be a much more-significant impedance (although the RF's fields are usually mostly electric fields, rather than magnetic, after they get more than one or two wavelengths from the transmitting antenna).

I don't have any idea about whether or not any possible effects from those might be significant. Just wanted to mention the theoretical possibility.

If there were to be a problem, I guess you could always simply solder some copper mesh (screen) across the holes, to eliminate the enclosed loop area, eliminating the induced currents, while still allowing the airflow.

I wont say you are wrong, because you are absolutely right about it. :)

However, I am not really worried about it that much.

First of all, the transformer is outside the case, since I am using a external wallwart to supply it. I do not like working with mains voltage if I can avoid it.

Secondly, RF induces currents, that is a bigger issue and to be honest I have no idea if it would be a problem or not.

Might have to do some thinking. :)
 
Well, thought about it some more and this is my solution.

I could solder some wire mesh over the PCB holes but while it will work it just does not "feel" right.

My option is a ton of 2mm diameter holes under the heatsink, with the power and power ground planes between the holes. Sort of a mesh like structure with round holes.

A simple solution and since I am not paying per hole, it cost nothing to add a bunch of holes to the PCB.

EDIT : Wont let me upload the picture. A standard 151 kb .png file. Weird.

Anyway, should be easy to visualize.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.