Aspen Headphone Amp

[keantoken]

I think there some things regarding grounding that need to be agreed upon.

1: Is our ground connected to the chassis, which is basically a huge antenna unless connected to earth ground? Some houses, mine included, have no ground connection! So perhaps we should treat the chassis as an RF shield rather than absorber. Does someone have experience to offer?

2: How are our signal ground and earth ground isolated, or are they at all?

We need to come up with our grounding scheme. Then we can decide whether we need virtual earth and how to implement it.

This is what I suggest:

1: Signal ground connected to negative rail.

2: Power ground connected to capacitive virtual earth as in schematic. (I deleted the top capacitor in the schematic since it improves PSRR)

I don't know if signal ground and power ground being connected is a problem for a headphone amp. I see nothing wrong with this. I think we could still get better PSRR if we removed decoupling from the input and made it so that it followed the virtual earth.

In any case I like my virtual earth solution better than the EF virtual eartth because it has lower impedance and is more cost effective, and think it should be used if we decide on a virtual earth grounding scheme.

PS. We should hold an official vote, we really should (possibly through the poll system?). Things to include: relay-switches, multiple headphone outputs, multiple headphone impedances, onboard or outboard supply, etc. I'm sure there are plenty of silent viewers who would vote.

Carlos, that is a great schematic. What program do you use?

- keantoken

[AKSA]

Nico,

Would you like to advise KT why his approach is compromised, or should I?

Hugh

[Nico Ras]

Hi Hugh, why don't you maybe my way of explaining is not that great.

[keantoken]

Does it have to do with the common mode chokes?

- keantoken

[Nico Ras]

Hi KT,

Since this design was earmarked as a group design, a vote on each issue would be a good idea. I suggest that you coordinate this with Gareth, he is after-all project leader.

B.T.W. I was not aware that Carlos offered a schematic as well, please share it with us.

[keantoken]

I was referring to you, Nico, not Carlos! Got it mixed up somehow!

Okay, I will PM Gareth about the vote and depending on his response, we'll have something going.

- keantoken

[AKSA]

Hi KT,

1: Is our ground connected to the chassis, which is basically a huge antenna unless connected to earth ground? Some houses, mine included, have no ground connection! So perhaps we should treat the chassis as an RF shield rather than absorber. Does someone have experience to offer?

I would be quite OK to connect virtual ground to chassis but NOT to IEC ground. This would also scotch any RF if it is induced.

2: How are our signal ground and earth ground isolated, or are they at all?

You could join signal and power ground, but keep IEC ground from the power socket connected only to the chassis.

This is what I suggest:

1: Signal ground connected to negative rail.

2: Power ground connected to capacitive virtual earth as in schematic. (I deleted the top capacitor in the schematic since it improves PSRR)

I don't know if signal ground and power ground being connected is a problem for a headphone amp. I see nothing wrong with this. I think we could still get better PSRR if we removed decoupling from the input and made it so that it followed the virtual earth.

In any case I like my virtual earth solution better than the EF virtual eartth because it has lower impedance and is more cost effective, and think it should be used if we decide on a virtual earth grounding scheme.

I will go with Nico's suggestion, KT, because it has been used before with great success and the transistor virtual earth also offers a DC as well as an AC path which has to be good for bias currents.

Both transistors can be TO126 BD139/BD140, very cheap, while the complexity is higher, it's a belt and braces approach and covers all current bases.

You should have a cap on both VE resistors, not just one, and the signal and power earth could then be taken off the negative rail and placed on the centre, Virtual Earth as originally suggested. This will deliver the current nulling for very quiet noise outcome, and ensure that there is no asymmetrical switch on pulse.

I do not believe a vote is appropriate; we are seeking the right answer, and truth is not always amenable to a vote! In any event, I'm knackered, so my answer is short.

Cheers,

Hugh

[keantoken]

So your reasoning is that the socket ground is polluted with all sort of garbage? I was wondering why most plug-ins only have 2 prongs...

You are right about voting... In this case it should be fine to bring up a list of things to ask you about, and if anyone has a reasonable argument changes will be made if necessary. I won't have time until 4:00 (GMT-6), but someone will have to read through posts a few or more pages back and write up a list.

I'm not sure whether this is significant:

What does it matter that we connect chassis to either power ground or signal ground? We could also connect the chassis to the positive rail, but what affect would this have on the circuit, or is it significant at all?

- keantoken

[Nico Ras]

Hi KT,

I actually thought Carlos was lurking in the shadows somewhere not sharing his thoughts. (I am just kidding, I wanted to see your response)

Anyway, the program that I use for schematic capture and PCB design is called Easy-PC from No.1 Systems (Number One Systems :: PCB CAD Software :: Home of Easy-PC PCB Design)

And I use MicroCAP9 (Main Redirect) for simulations.

Hugh is right regarding the voting, I support voting for specific features but voting on how Hugh should design the circuit is not an option, it his his call.

Kind regards

Nico

[Nico Ras]

Whew! where did those links come from???? Never-the-less they actually work.