Aspen Headphone Amp

[Nico Ras]

Hi Geatan,
just check wether the miller cap value works for that transistor, you may have to adjust it to get optimum rsults.

[gaetan8888]

Originally Posted by Nico Ras

Hi Geatan,
just check wether the miller cap value works for that transistor, you may have to adjust it to get optimum rsults.

Hello Nico

From past use of the KSC3503, I know that I may need to raise a bit the value of the miller cap and phase lead cap.

The 2SC3423 have a Cob of 1.8 pF , the KSC3503 have a Cob of 2.6 pF

I will check with a 10k hz square wave on a scope.

Btw, I will do my pcb, so it may be usefull to see an image of the pcb routes.

Thank

Bye

Gaetan

[mlloyd1]

delurkin again ...

lookin' good!
i see light at the end of this tunnel and it doesn't look like it will be a train wreck!


mlloyd1

Originally Posted by AKSA

The references are different..... but you clearly have worked out why it's been done. So in truth, a little thought, seeing the obvious intent of it all, and you've reached an epiphany......

Try this schematic!

Hugh

[jcx]

series RC in parallel with R1?

I've found a input high frequency termination "Zobel" can be useful, adding phase margin, reducing source/vol pot setting dependency

looking at the feedback loop signal path the input Q1 likes a low Z HF AC gnd at its base

also can fliter some RF from the input


I know Hugh has some ideas on feedback and amplifier voicing but I hate to see all of the compensation bypassing the output stage - "TMC" or 2-pole compensation both would allow loop gain to be maintained to a higher frequency around the output stage



earlier output Zobel design was questioned - the main load difficulty is cable capacitance

due to the desire for small flexible cable headphones can have higher cable C - the amp should be stable into a few hundred pF to at least a few nF

while sim will not be definitive due to modeling limitations a look at output behavior with a step current appied to the output or a V step in series with the modeled cable Cload can show some useful features - AC analysis is less useful than .tran based sims

[Nico Ras]

Hi jcx,

What you say is valid, Could this be the reason for the "standard 120 ohm" series resistance in the output of the headphone amplifier. I simulated with and without. Without the series resistance there is most definately instability if the leads are seen as very capacitive.

[Nico Ras]

JCX,

The cable C plus speaker L forms a tank circuit, lowering its Q significantly with added series resistance will dampen excitation.

[sandyK]

Originally Posted by Nico Ras

Hugh, please decouple the input from ground by your usual method 10 OHMS..., you still have an ground loop issue even thought it is called virtual ground, unless you are planning one VG per amp.

Nico

Hugh

I agree with Nico.
You don't need any more Professors !

SandyK

[Bigun]

Guys, it's looking really good. I'm not sure that I was able to 'direct the thread' or whether the thread was able to direct me but so far it's turned out very interesting and in my view a truly team effort

And with the design phase coming to a close it's perhaps a good moment to let you know that I have some business trips and other things that will reduce my available time for awhile. But if all goes well I expect I'll have some pcb's from Hugh and in my 'inventory' soon enough !

[Nico Ras]

Hi Sandy, do you agree with my comment above your last post - it seems logical

[keantoken]

Another nitpick, I scrutinize the value of C3. The graph shows the AC current through R9 (bootstrap source) in mA, for values of 100uF, 220uF, and 470uF (I believe the AC analysis assumes an input voltage of 1V). Normally it is very low, but has risen considerably at 20Hz (lower is better, the bootstrap is supposed to be a CCS). What I don't know is what effect this has on the actual sound.

This should probably wait until someone prototypes it, I just want to point this out. I've been uneasy about the descending value of C3 over time.

- keantoken