My Blameless MkII

[bimo]

I use 4mA CCS for LTP.
I think you should find the simulation for highest slew rate as posible, then lowest THD at 20kHz and highest PSRR. I have success design amp with Lin topology or Blameless. Only one modification after first implementation, it comparable with VSSA (although my wife prefered the sound of this amp).

 

[Ranchu32]

Bimo, can you post your schematic?

 

[AKSA]

No, Andrew, I think this is a bit overstated on both sides.

You are very helpful, and I appreciate it, and you do too, since Ranchu thanked you very graciously. But life on the forums is social as well as engineering, the fact is we are all human.

Cheers,

Hugh

 

[bimo]

Bimo, can you post your schematic?

I send you PM.

 

[Bigun]

the fact is we are all human.

you can please all of the people some of the time, some of the people all of the time, but never all of the people all of the time

 

[Ranchu32]

Calling Bigun...

Hi Gareth...

I took on your feedback and had another play around with the layout. Ignore the bottom half - its a mess - I'm trying out different ideas. What do you think about the top half?

 

[Bigun]

I think you are taking the right approach, layout the big parts that need big currents and work out how to fit in neat traces to carry those currents around - especially the ground.

You could eliminate the pcb mounting holes near the top edge, the power devices will anchor that edge of the board and as there are 4 of them I don't see the board placing any undue stress on the leads. This will give some space to optimize further.

Now that I've posted the TGM8 files you could also explore the option of starting with that layout and modifying it to suit yours.

 

[AndrewT]

Ranchu,
see the space where R46 is written.
That's where the HF decoupling should be fitted.

The HF decoupling at the big electrolytics is just about useless.

 

[Ranchu32]

Hello everyone,

Its been a while since I updated this thread and in that time I've been very busy researching and refining this project.

Layout

The board has undergone a redesign to reduce the length of high current traces by placing the bulk filter caps as close to the output transistor pins as close as possible. The pairs of output transistors are now horizontally opposed, whereas previously they were positioned along the top edge of the board. Subjectively, I believe that amplifiers with a large capacitance store close to the outputs sound better, and having the power supply on-board certainly simplifies wiring and chassis layout.

The LTP and current mirror transistors are oriented such that they can be thermally coupled (thanks Bigun).

Circuit

LTP CCS has been cascoded to further reduce hum and I've settled on ~6mA for the LTP pair after analysing a number of similar designs that use this same family of input transistors.

The LTP cascode has been removed and higher voltage parts substituted for the LTP. I will experiment with cascodes in a future re-design; however, I believe this circuit is enough of a departure from the Mk I that I want to evaluate these changes in isolation.

I spent some time reading Groner on Self (http://www.sg-acoustics.ch/analogue_audio/power_amplifiers/pdf/audio_power_amp_design_comments.pdf) and a very interesting thread on this forum (http://www.diyaudio.com/forums/solid-state/133018-current-mirror-discussion.html). I've adopted an improved current mirror here. The emitter degen resistors have been redimensioned to drop approx. 660mV, which I feel is a reasonable compromise.

 

[AndrewT]

The sch shows the VAS feed coming from the wrong side of the mirror.