Final Touches And Laying it All Down

Aside from going to production, there are only some minor changes left to do with the circuit at this point. So sadly, this is the last post about this short project. In this post I will review some minor changes done to the circuit schematic and give some pointers regarding the layout.

The first issue to address is an added low pass filter. Eventually, to save time and effort I decided not to make it adjustable, however the component is optional. The reason this was added to the layout was that there were too much trebles while I listened to the proto-build. Another thing a low pass adds to a feedback amplifier is stability. This concept requires some mathematical coverage and will not be reviewed in this post (maybe in the distant future, I have several others in the queue). In this topology, the simply can be calculated mathematically by “comparing” it to the resistor (when the imaginary part of the capacitor impedance equals the resistors real one),

C_{LP} = \frac{1}{2\pi R_f f_P}

Here C_LP is the low pass capacitor, R_f is the feedback resistor value and f_P is the pole frequency, namely the frequency in which the filters starts attenuating. For a pole frequency, we will place some frequency between 10 and 20 kHz and a resistor of kHz is used in the schematic. Choosing 15 kHz yields that a capacitor of ~2.4 nF is needed. At this point it is important to remember that only specific capacitor values are actually made. A short sweep on EBay shows that it is simple enough to find 1, 2.2, 3.3 and 4.7 pF capacitors. Placing them in QUCS and sweeping obtains

It can be seen that the choice is between a 2.2 nH and a 1 nH. I am quite inexperienced in these issues, so the correct choice might be something different altogether. Perhaps the proper solution should have been a series feedback inductor, but that I will only learn with time, I guess. The usage of capacitors is more common, however, for two reasons:

  1. They are more accurately manufactured.
  2. They tend to be less lossy.

Losses are the unpleasant truth about real components. They have parasitic properties, such as resistance, that cause them to dissipate energy.

The second change to the schematic was replacing the decoupling capacitor with 1000 uF. This change is probably useless. Why? because although this allows frequencies as low as 50 Hz to be amplified unobstructed, 100% of the headphones I can afford really cant make these frequencies significant. So this might change eventually, on-the-fly.

All set to laying out the circuit! After annotating all of the components (someone has to explain to me why this isn’t called numbering or indexing) a netlist file is exported out of the schematic and imported into the layout editor. You start off with something like this:

Don’t panic (I did). This is just all your components lumped together. Bursting them aside is simple. Let’s consider this:

  1. The LM4808 amplifier is the “heart” of the board.
  2. The charging component should be close to the USB connector.
  3. All connectors\switches should be on the edge of the circuit.
  4. LEDs are cool. Put more than one if you want to.

I’ll add a personal touch here. In a lot of headphone amplifiers you see the input and output sitting on the same side. I’m a fan of the In\Out layout, namely one side is all of the inputs and another (not necessarily opposite. See the raspberry Pi 1 vs. the 2!) has all of the outputs. So arranging according to these guidelines:

The white lines, called the ratsnet (because rats, apparently, weave nets…), indicate where a net starts and where it is supposed to connect to. For this specific project, I applied some simple design rules:

  1. Unless absolutely necessary, all audio signals go on the top layer.
  2. If an audio signal goes on the bottom layer, then it’s stereo counterpart also goes on the bottom layer.
  3. The feedback structure has to be as symmetric as possible.
  4. Audio lines are 0.1 mils wide (yes, this project made me think in retard units).
  5. Power\DC lines are 0.2 mils wide.
  6. As much ground as possible.

Eventually this came out (yes I’m showing off in 3D):

There were some additional problems in transferring the manufacturing files to the factory, that forced me to use round instead of oval holes, but hopefully I’ll resolve them by next time. Hope you enjoyed this project. The next and final post about it will be after assembly.

Cheers!

Leave a Reply

Your email address will not be published. Required fields are marked *