Aspen Headphone Amp

[GK]

Originally Posted by AKSA

Here's the Bode plot from LTSpice:

Looks like unity gain is at 11.6MHz, 3dB point is at 1.6MHz.

I noticed GK's plot showed a gain of near on 34dB, which is very different to the CLG of 15dB.

?

I plotted the loop gain and phase. A closed loop bode plot tells very little about a simulated amplifiers stability margins and you consistently apply the wrong technical terms to the wrong things.

[Bigun]

Originally Posted by MJL21193

Maybe a variation on the Jung super reg. Make things a bit more challenging for us older hands.

That would be an interesting learning process. But how would you simplify it - the Jung super reg. looks too complicated for the purposes of this simple (but great sounding) headphone amp ?

[There is that simple feed-fwd shunt regulator that the tube guys are playing with, works providing that the current levels are low]

[Bigun]

Originally Posted by G.K.

I plotted the loop gain and phase. A closed loop bode plot tells very little about a simulated amplifiers stability margins

Good tip, I rarely look at the amplifier performance open loop, which misses a lot. But still, I like to know how it looks with the loop closed - any signs of instability there would be bad...

[andy_c]

Bigun,

For more info on the loop gain concept, see the posts starting here in the frugalamp thread. In this post, I talked a little bit about the theory of why one should look at the loop gain (rather than the closed-loop gain as Hugh did above) to determine stability. Note that I originally made an error, and told OS to simultate V(b)/V(a), when it should have been -V(b)/V(a).

[Ipanema]

Hi GK,

Can you post your simulation netlist? I would like to know how you simulate the loop gain.

Thanks.

[ikoflexer]

For those who use ltspice, load the following file which is part of the ltspice distribution:
C:\Program Files/LTC\LTspiceIV\examples\Educational\LoopGain2.asc

And run the simulation.

Explanations about where that comes from can be found here
http://sites.google.com/site/frankwiedmann/loopgain

HTH

[ikoflexer]

Originally Posted by Bigun

That would be an interesting learning process. But how would you simplify it - the Jung super reg. looks too complicated for the purposes of this simple (but great sounding) headphone amp ?

[There is that simple feed-fwd shunt regulator that the tube guys are playing with, works providing that the current levels are low]

Here's simple and great performance.

http://www.diyaudio.com/forums/power-supplies/143693-simplistic-salas-low-voltage-shunt-regulator-post1928071.html

Hot too, it's a shunt design. Don't look back.

[MJL21193]

Originally Posted by Bigun

That would be an interesting learning process. But how would you simplify it - the Jung super reg. looks too complicated for the purposes of this simple (but great sounding) headphone amp ?

It isn't as complex as it seems and my sims say it hard to beat for this application.
I wouldn't make any real changes, just use an ordinary opamp to reduce cost.

[ostripper]

by andy c. - and told OS to simultate V(b)/V(a), when it should have been -V(b)/V(a).

And it works soooo goood. No oscillators YET !!

I prepared the .asc (in the aksaHFamp.zip folder) and it works nice , 775khz unity gain with 94db margin (attachment 1), this would be very stable.

Hello , Hugh ... looks good , I "dug up" the japanese models , but used the mje15032/33's for OP's. With the "super " reg ,or even some heatsinked 317/337 devices , this should sound golden. I balanced out the input diff. by tinkerin' with the tail current and chose a 10 pf lag comp to assure margin on the attached .ASC. Nice work !
OS

[GK]

Originally Posted by Ipanema

Hi GK,

Can you post your simulation netlist? I would like to know how you simulate the loop gain.

Thanks.

Hi

I just used the simple method shown in the LTspice example file; LTspiceIV/examples/Educational/LoopGain.asc.

This just involves putting an AC voltage source (set to '1') in series with the output and then plotting one terminal over the other.

Follow Andy's links above to a more complete explanation.