Mjølner, another all discrete Lateral FET Headphone Amplifier
 

A simplified variation of my Thor headamp.

Still all discrete, with Lateral FET output and DC Servo.

Removed a few stages, only single BJT VAS instead of Darlington VAS and no driver BJT's. Using 2SK213-216/2SJ76-79 TO-220 Lateral FET's for output.

Optimized for 150-600 Ohm loads.

Simulated performance is :

1 kHz and 100mW/5VRMS into a 250 Ohm load : 0.000122% THD

20 kHz and 100mW/5VRMS into a 250 Ohm load : 0.001876% THD

Performance is slightly worse into a 150 Ohm load and slightly better into a 600 Ohm load, as expected.

Attached are schematic and my layout(Work in progress)

Separate AGND and PGND ground planes

Gerber view.

However not quite finished, trying to the the THD-20 numbers a little better preferably below 0.001% THD-20.

Board is 90x90mm and can be used with the same supply board and no thump board used with my Thor head amp boards.

Nice!

/U.

Very nice and I like the pcb layout, too! Getting close to my ideal amp performance. -RNM

Thanks, I stole a few layout ideas from you. ;) You made some interesting posts in the thread about your headamp. Especially the few posts about grounding. Although I am using the star point as the ground point for my TRS output jack instead of the SGND ground plane.

Most of the points you set forth for good headamp design I agree with.

But I am using a DC servo, hard to get parts(2SK213-216 and 2SJ76-79) and the amp is really not happy about driving 30 Ohm loads, the THD gets too high.

Then again, I design for my own use and in this case, performance at 250 Ohm is the most important, great performance down to 150 Ohm and up to 600 Ohm is just a bonus.

Performance is actually a great deal better at 1VRMS than 5VRMS, then we are talking about 0.0000XX% THD-1 into a 250 Ohm load and 0.000XXX% THD-20 into a 250 Ohm load.

With input filter frequency response is -3dB down around 240kHz, without input filter we are talking about -3dB down at 3 MHz, with output all the way out to 5.7 MHz, so it is a very wideband amplifier. LF cutoff is, well, almost down to DC.

Ofcourse it is all simulated, but peformance should not be too far off in the real world.

Compared to you and many other people on this forum I am just an amateur.

I work with electroncis as my day job but in another field, so designing amplifier circuits is something where I learn as I go along.

I joined this forum only to teach from my broad experience and so you didnt steal anything. It is offered to all. In fact, I wish more people would ask questions but it is a Build oriented forum and not so much an intellectual curiousity forum. Good luck on your project and learn from all the old timers because it will get you further down the road faster without doing our part over again. Thats how we make progress.

Neutrality, for low-impedance phones, the Thor is more suitable, right?

Short answer, yes.

Long answer, Mjølner would, according to simulation, happily drive a 32 Ohm load up to around 100-150 mW. Above that level distoriton gets too close to the 0.01% THD-20 area for my liking.

Would it work decently into 32 Ohm headphones? It probably would, but if it does it is just a bonus.

If I build it it will be a +/- 30VDC version

That is 2666mW into 150 Ohm, 1600mW into 250 Ohm and 666mW into 600 Ohm. In short, extreme headroom and power reserves as well as very low distortion.

However, since I just spent my last "hobby" money for this month on a pair of Beyerdynamic 250 Ohm DT 880 PRO headphones I wont be building anything until next month. Have to order PCB's and components for Mjølner HAB MKI as well as PCB's for Thor Supply Board MKII.

Simulation with +/-30VDC rails.

20kHz/1VRMS :

150 Ohm : 0.000229% THD
250 Ohm : 0.000082% THD
600 Ohm : 0.000020% THD

20kHz/5VRMS :

150 Ohm : 0.001214% THD
250 Ohm : 0.000494% THD
600 Ohm : 0.000231% THD

20kHz/10VRMS :

150 Ohm : 0.003147% THD
250 Ohm : 0.001533% THD
600 Ohm : 0.000947% THD

More than enough low distortion output to drive more or less any 150-600 Ohm headphone on the market.

Done some modifications to the signal/shield ground plane.

The signal part of the signal/shield ground plane now has its little own isolated island only connecting to the signal/shield ground plane at one point, which is the GND connector going to the audio ground star.