Building an ultimate low power class A amplifier – my way

The two solder points (input/output) are not enough support for this module with the heatsink making it top heavy.
Actually, there is 6 connection points aligned with Prasi’s CRC PCB. I’ve used 1 A diode cutoff wires for connection. On the picture, all connection points are not visible and this is 1.4 version PCB, where connection points were not well aligned with CRC PCB.

No bracket.jpg
 
It is fun to drive A/B stage with opamp.
Nice project


So .001% at 1k

.01% ? at 20k

Something been wanting to do myself for 1 to 3 watt speakers
Since low power numbers with opamp usually very good.
Might hold up better at 4 ohms with another pair output devices.
 
  • Like
Reactions: lineup
Measured, rounded up distortion figure for 10W/8Ω is 0.0007. I don’t have low distortion signal generator for 19 or 20 kHz, so with NFB being 25 dB down at 20 kHz, we can estimate distortion as 0.013.
 
Last edited:
  • Like
Reactions: lineup
Here is result of 32 tone signal test. From DAC with nominal 2Vrms output, 710 mV multitone was feed to amplifier input. That was maximum before REW signal generator would proclaim DAC output clipping. At amplifier output was chaotic signal with 16 V peak to peak. Through attenuator, output signal was forwarded to Cosmos ADC. Result speaks for itself. It looks like multitone measured direct at some DAC output, rather than from amplifier output.

Multitone.jpg
 
@tombo56, in post #19 you mentioned you would add corresponding voltages for various bias currents as measured at R16 or R17. I can't see a post where you have listed that information, can you provide some details thanks. I have your pcb's half built, so would be looking for some biasing info in the next couple of weeks, thanks. would also be interested in your final bias selection for both 4ohm and 8ohm loads.

Regards,
Gary.
 
Didn’t expect anyone is near building phase.
Bias current can be measured through voltage at R16/R17 resistors.
  • 0.8 A 160 - 170 mV
  • 1 A 250 - 270 mV
  • 1.2 A 340 – 360 mV
Those values don’t change with actual current. Lateral MOSFETs have negative temperature coefficient, and bias current will drop from cold to hot state for some 70 – 80 mA.
Actual current depends on heatsink temperature. Lower voltage is for moderately warm heatsinks and higher is for heatsinks at 40 °C.
 
As this is an amplifier reaching above 1 MHz and using the whole outer PCB surfaces as grounded shields, it would be a good practice in principle to place resistors slightly above PCB surface, to minimize parasitic capacitances. It is good in principle but not required. During development, this amplifier worked and was harassed as a 1.8 – 2 MHz device. All was well.

Elevated resistors.jpg



It is enough to lift NFB resistors R3, R4, R5, R6. Rest, if you like.

My opinion is that good amplifier must have a great safety margin. For offering this to DIY fellows, frequency response was reduced to 1.3 MHz as this increases phase margin to some 80°.
Mine amplifier is exactly the same as published. I can’t have different pudding than offered to anyone else.

So, I used 0.5 mm thick Teflon strip to elevate resistors before soldering. I’m sure you can find something suitable as well.

Strip.jpg
 
Thanks for the bias info, now have that printed out. I am probably 40% through the build, taking it slowly. I thought I had some Exicon laterals stashed away, but now can't find them so will have to order from Profusion UK, so that will be 3 weeks delivery or so. Good tip on the resistors, I can do that., I normally sit 0.5W resistors 2mm above the pcb and 3W resistors 5mm above the pcb.

I thought Vunce might have started his build as well.

The boards I got from JLCPCB look great, you have done a great job with the layout. I see you have gone for the fashionable purple, I went for blue this time.
 
Last edited:
Ah, no. I wouldn’t dare at 16 Vpp. It was 8 Ω resistor.

It is worth to mention that hash, rising with frequency from -150 to -135 dBV levels, was not produced by amplifier or used DAC. It is a Cosmos ADC artifact. Here is what designer measured for Cosmos ADC.

ES9023-Multitone-0dB.png