Amplifier progress update

It’s working! It’s working!

After some delay, I’ve gotten the PCB and components, put it all together — tested it, briefly wondered why it was letting through 5A instead of 2A, added a missing wire link — and it works!

Messy desk

First impressions, sound-wise

Take one: Initially, the potentiometers were set randomly, and although there was some sound coming out of the speaker, it was very quiet. I started a “blind search”, turning POT1 and POT2 left and right, listening for changes in the sound. It was late at night, and the PSU was causing difficulties. Not only was its ventilation fan droning like a vacuum cleaner, but to overcome the 15V limit I added a lead acid battery in series, so the tone changed with every minute as the battery discharged.

During the short breaks when the PSU drone periodically switched itself off, the sound was smooth in the treble, but somewhat garbled at lower frequencies.

Take two: I borrowed a different PSU, and focused on a ‘hump’ in the sensitivity, around 7.5V DC at the output, where the gain seemed to be the highest. I then attempted to fine-tune what I heard by adjusting POT3. At this setting, vocals were a little strained, making Awolnation’s ‘Sail’ extra visceral.

At 10V DC out, the gain seemed lower, but the sound was also cleaner. Super smooth highs — supershigi’s “Chrono Trigger — Fall” cover sounded fantastic. ‘S’s were articulate, not noisy hash. The Dayton 3.5″ “point source” speaker was smooth and detailed, and managed to be non-harsh and non-sibilant in spite of the ‘hot’ tonal balance.

What’s Next?

I’ll be able to do a more complete analysis and tune when I get access to an oscilloscope, and I’ll have more confidence about what’s happening with the audio signal. At the moment though, I can only guess that there was some mid-band saturation happening, perhaps due to the current source being a bit low at 1.5A.

PSU design — this may be an interesting project in its own right, but to be honest, I want to get it over and done with, quickly. Something like a 2 or 4 × 40W SMPS with 24V out, 4V of head room for linear regulators (one per channel). Done!

At a full 12V, the amplifier’s cooling fan is also too loud for me. On the other hand, at 5V it was barely spinning. It’s nice that it doesn’t strictly need PWM control, but some speed control could be useful as the device ages. A small project with a speed controller.

I’d like to add a soft-start feature. While it’s probably not essential for bass, tweeters may require protection against turn-on/off thumps.

General bug-fixing:

  • Fitting some of the components onto the board was a bit awkward due to wrong sized holes or the wrong pitch. There may be one or two eye-rollers in there, like missing + and – symbols for the power supply polarity. Or R15: a hot resistor, nestled uncomfortably close to some electrolytics.
  • I’m not entirely sold on the centre-mounted CPU cooler. Although the spring-loaded clamp is self-balancing, the TO-220 and TO-247 cases have to be perfectly positioned for it to be nice and straight. Nearby solder joints pressing against the plastic side weren’t helping, either.
  • The number of potentiometers could be reduced to a maximum of 2.
  • While probing the PCB (with great difficulty I might add), I had an amazing idea: why not add some special “test points” for easier access to the most important voltage levels?
  • Some buzzing could be heard when the fan was running off a common power supply, reminding me to add some filtering to isolate the long-tailed pair from the main MOSFET load.
Top layer. The power MOSFETs will be mounted underneath, together with the heatsink.
Fat tracks should help to wick some heat away from Q4 and Q5.

Last week…

Both the PCBs and electronic parts should be coming in this week (fingers crossed)! I could probably improve a hundred different things, in both the schematic and PCB, but TBH I was getting tired of tweaking and perfecting the theoretical design. It’s time for real hardware!

Maybe the trimmers will produce crackling noises during adjustment… Maybe the MOSFETs won’t be well matched… Maybe the soldering of some components will be tricky due to the large copper areas… maybe some of the components won’t physically fit well on the board…