Nico,
I'm not sure people WILL accept what I say, and nor should they!! I was wrong once before, 1963 I think it was.
KT,
The virtual earth is the junction of C1/C13, and the input signal earth should be taken to this point, and the two VE caps should at the least have resistors of 2k2 strung across them to precisely fix the mid-point of the 24V unipolar power supply.
John,
Really it is a single rail supply, but we are simply creating a virtual earth to improve psrr since ripple from the top rail cancels equal ripple from the bottom rail at the mid-point. This could be a great advantage in this circuit since cans are highly noise sensitive, much more than speakers I guess.
Nico,
You are a bloody wizard. Thanks mate......
Hugh
I'm not sure people WILL accept what I say, and nor should they!! I was wrong once before, 1963 I think it was.
KT,
The virtual earth is the junction of C1/C13, and the input signal earth should be taken to this point, and the two VE caps should at the least have resistors of 2k2 strung across them to precisely fix the mid-point of the 24V unipolar power supply.
John,
Really it is a single rail supply, but we are simply creating a virtual earth to improve psrr since ripple from the top rail cancels equal ripple from the bottom rail at the mid-point. This could be a great advantage in this circuit since cans are highly noise sensitive, much more than speakers I guess.
Nico,
You are a bloody wizard. Thanks mate......
Hugh
Hi Hugh,
No explanation necessary.
I don't have a problem with either, split or single works for me. My point is the basic amp schematic is not yet nailed down - it's getting to be a bit confusing.
I am sure you guys have used it before not knowing it. Think about an op-amp using single supply, you derive a V/2 or 0V refenece for it exactly this way way. Normally you string two resistors together, but you are limited by the current being passed. By bypassing the resistors with capacitors you create a low impedance path for ac.
Yeah, you've got me pegged. I have a congenital problem with doing things the conventional way. It's a character flaw that sometimes works, sometimes gets me into trouble. I like this little project, because it's simple enough that I can follow the thought processes behind it. I like Hugh's approach too, of taking basic circuits and refining them to make good sound.
I also like self balancing/canceling sort of stuff, like this virtual earth. Has an elegance that seems natural to me.
BTW. I don't have any philosophical objection to more complex stuff (I built Mihai's amp), I just can't follow everything in those cases.
Sheldon
Approaching the end
Guys, this is what I have so far. I would like to push for finalization.
Some capacitors may not be the absolute correct value because they do not yet exist in my library.
I have added the Virtual Earth for completeness, but it is still a decision required by the group.
All the components have to be defined by manufacturer and part number so I can create a library part for them and every one is building the same amp with the same characteristics. Before I have an agreed Bill of Materials I won't start the PCB design.
For the cross feed I used two small 12V DIL relays and an indicator LED, you can decide if it should be a simple switch or the configuration shown.
I have also added a PCB link selector (J9 & J8) for the feed back take off point of the bootstrap because not unlike myself, John also strives for the lowest possible distortion.
I am not sure if there should be a regulated power supply per amplifier, this is for you to debate.
I am also not yet convinced that the CMC that you have chosen would do the job, for the frequency range that it should work in. It must be an iron core and would be the size of a small PCB mounted transformer. I guess one can use a transformer with two separate 6V windings but the PCB size grows.
Do not direct questions at me, I am just going to follow instructions as by group agreement and posted by Gareth, there must be no ambiguity of what is to go on the PCB I am not going to sit in a tight loop changing things on a whim, I do not have the time for that.
If I missed something on the schematic, Gareth it remains your call.
Kind regards
Nico
Guys, this is what I have so far. I would like to push for finalization.
Some capacitors may not be the absolute correct value because they do not yet exist in my library.
I have added the Virtual Earth for completeness, but it is still a decision required by the group.
All the components have to be defined by manufacturer and part number so I can create a library part for them and every one is building the same amp with the same characteristics. Before I have an agreed Bill of Materials I won't start the PCB design.
For the cross feed I used two small 12V DIL relays and an indicator LED, you can decide if it should be a simple switch or the configuration shown.
I have also added a PCB link selector (J9 & J8) for the feed back take off point of the bootstrap because not unlike myself, John also strives for the lowest possible distortion.
I am not sure if there should be a regulated power supply per amplifier, this is for you to debate.
I am also not yet convinced that the CMC that you have chosen would do the job, for the frequency range that it should work in. It must be an iron core and would be the size of a small PCB mounted transformer. I guess one can use a transformer with two separate 6V windings but the PCB size grows.
Do not direct questions at me, I am just going to follow instructions as by group agreement and posted by Gareth, there must be no ambiguity of what is to go on the PCB I am not going to sit in a tight loop changing things on a whim, I do not have the time for that.
If I missed something on the schematic, Gareth it remains your call.
Kind regards
Nico
Member
Joined 2009
Paid Member
Nico,
I'm still thinking about this Virutal Earth. It seems to me that we've just added two more active devices and some passives in order to implement it. And if we are going to do that, why don't we just make a dual rail supply using a centre tapped transformer ?????
I thought the Virtual Earth approach was a response to the constraint of a single supply voltage, such as a battery for example. But we don't have that constraint unless we want to restrict operation to a Wallwart.
So it might still boil down to the question of whether the amplifier with a single rail is quiet enough ?
Does this sound logical to you ?
I'm still thinking about this Virutal Earth. It seems to me that we've just added two more active devices and some passives in order to implement it. And if we are going to do that, why don't we just make a dual rail supply using a centre tapped transformer ?????
I thought the Virtual Earth approach was a response to the constraint of a single supply voltage, such as a battery for example. But we don't have that constraint unless we want to restrict operation to a Wallwart.
So it might still boil down to the question of whether the amplifier with a single rail is quiet enough ?
Does this sound logical to you ?
Gareth,
Virtual Earth is an advancement from a center tapped transformer power supply. This is one up, not one down.
Besided I read here that people want to use a wall wart. Thus the transformer would drop off this drawing and replaced by a socket for a wall wart.
I have seen 24VAC wall wart in the UK some time ago. I think someone has to do a search to see how plentiful and easily obtainable they are, else we have to go the internal PCB mount transformer route.
Virtual Earth is an advancement from a center tapped transformer power supply. This is one up, not one down.
Besided I read here that people want to use a wall wart. Thus the transformer would drop off this drawing and replaced by a socket for a wall wart.
I have seen 24VAC wall wart in the UK some time ago. I think someone has to do a search to see how plentiful and easily obtainable they are, else we have to go the internal PCB mount transformer route.
http://www.action-electronics.com/acadapter.htm#Ac found this, never purchased from them, but looks like they may have what we need, 24-VAC 2.1-Amp is the largest I could find. here's another http://www.mouser.com/Power/Plug-In-AC-Adapters/_/N-wp53?P=1z0wd6bZ1z0y71gZ1z0wd9m
This is 100-240 VAC http://www.mouser.com/Power/Plug-In-AC-Adapters/_/N-wp53?P=1z0y71gZ1z0wd9mZ1z0wd9r
This is 100-240 VAC http://www.mouser.com/Power/Plug-In-AC-Adapters/_/N-wp53?P=1z0y71gZ1z0wd9mZ1z0wd9r
Last edited:
Interesting, this seems to be a little switch mode supply. I have tried switch mode and they do open up the sound to a very large degree. I do not know why more enthusiasts don't use switch mode.
One hears 100Hz mains hum a lot easier than 100 kHz hum. Besides it is a lot easier to filter and the reservoir caps can be much smaller.
One hears 100Hz mains hum a lot easier than 100 kHz hum. Besides it is a lot easier to filter and the reservoir caps can be much smaller.
Member
Joined 2009
Paid Member
Why is that ? - it's not obvious to me.
Also, with the single rail supply, you tried your test of putting the headphones to the rail via a capacitor and you said it was 'clean' - which I thought meant that single rail was OK ?
I hope a bit of skepticism is OK
I'm with Nico. He is the dude from Delphi! I say we go virtual earth, 24Vdc switch mode supply identified by TinZ, and move forward with this design.
Gareth, the idea of virtual earth has at its core the notion that connecting a quiet star earth to the centre tap of a transformer is NOT the quietest way to make a bipolar supply. Why connect a quiet earth to a rip snorting AC device like a trafo? This technique isolates ground utterly from the trafo. I use something related in my power amps, with each rail created by its own bridge rectifier, and then a CMC to separate the two filter caps. Star earth does not connect to any trafo winding, and thus the supply is far quieter.
John,
No one here doubts your apprehension of matters tech. I apologise if it looked like I was trying to teach Granny to suck eggs.....
Hugh
Gareth, the idea of virtual earth has at its core the notion that connecting a quiet star earth to the centre tap of a transformer is NOT the quietest way to make a bipolar supply. Why connect a quiet earth to a rip snorting AC device like a trafo? This technique isolates ground utterly from the trafo. I use something related in my power amps, with each rail created by its own bridge rectifier, and then a CMC to separate the two filter caps. Star earth does not connect to any trafo winding, and thus the supply is far quieter.
John,
No one here doubts your apprehension of matters tech. I apologise if it looked like I was trying to teach Granny to suck eggs.....
Hugh
Okay Gareth, the power supply tested was the first discreet PSU hugh offered. It was clean I agree. I tested it with two load resistors (50 & 100 Ohm). But I also said that the Amp PSRR rejection was only 17 dB, and this is not so good.
If the PSU was to start driving heavier loads then there will be ripple, but this would be masked by the signal and you would be reasonably happy, but the signal is modulated by the ripple.
Secondly the amp has a 15 dB gain, and try as you will using resistive bias ripple will apear at the input of the amp, being through the negative which is the common for the signal.
If you create a new reference point that is not the negative rail somewhere in between, using simply two resistors, as a voltage divider of say V/2 this junction or new reference point will not contain any line related noise.
This is because the symetrical artefacts appearing on the power rails are summed at the junction and the result is zero. Therefor the common mode rejection ratio of the amp is improved from 17 dB manyfold.
If the two resistors are equal the summing point voltage will be half that of the supply, regardles what the supply voltage may be and you can call this point zero or common and everything gets its reference from this point.
Now this is not entirely true if the current drawn out of this common point is not zero because as soon as you draw current then an imbalance of the summing point is created and it does not perform the function as it should.
What one does is to use a transistor to generate the zero volts so that the impedance of the summing point is lowered in the direction current is drawn.
Now should we modulate this point with ac like connecting a speaker to it, it would totally screw up the theory that I just put forward which applies only to dc.
The solution to that is to place two capacitors from the zero or reference point, one to the positive rail and one to the negative rail. The purpose of the capcitors is not only to keep the dc voltage at the summing point stable but to offer a low impedance shunt for any ac to either the positive or negative rail.
Gareth, I have a hard time explaining this, I am sure a guy like Andrew with his way of words would do a far better job than I did.
I all has to do with the geometric summing of common mode signals.
If the PSU was to start driving heavier loads then there will be ripple, but this would be masked by the signal and you would be reasonably happy, but the signal is modulated by the ripple.
Secondly the amp has a 15 dB gain, and try as you will using resistive bias ripple will apear at the input of the amp, being through the negative which is the common for the signal.
If you create a new reference point that is not the negative rail somewhere in between, using simply two resistors, as a voltage divider of say V/2 this junction or new reference point will not contain any line related noise.
This is because the symetrical artefacts appearing on the power rails are summed at the junction and the result is zero. Therefor the common mode rejection ratio of the amp is improved from 17 dB manyfold.
If the two resistors are equal the summing point voltage will be half that of the supply, regardles what the supply voltage may be and you can call this point zero or common and everything gets its reference from this point.
Now this is not entirely true if the current drawn out of this common point is not zero because as soon as you draw current then an imbalance of the summing point is created and it does not perform the function as it should.
What one does is to use a transistor to generate the zero volts so that the impedance of the summing point is lowered in the direction current is drawn.
Now should we modulate this point with ac like connecting a speaker to it, it would totally screw up the theory that I just put forward which applies only to dc.
The solution to that is to place two capacitors from the zero or reference point, one to the positive rail and one to the negative rail. The purpose of the capcitors is not only to keep the dc voltage at the summing point stable but to offer a low impedance shunt for any ac to either the positive or negative rail.
Gareth, I have a hard time explaining this, I am sure a guy like Andrew with his way of words would do a far better job than I did.
I all has to do with the geometric summing of common mode signals.
Sorry Hugh, I opened a can of worms here. But I believe that in you offering something here to the blokes, it must follow in your trend of designs, by being sonically superior to whatever else is available.
If there is no sonic superiority between this and a commercially available Headphone amp, then why bother in the first place.
Kind regards
Nico
If there is no sonic superiority between this and a commercially available Headphone amp, then why bother in the first place.
Kind regards
Nico
Member
Joined 2009
Paid Member
I think I know how the Virtual Earth works, but I'm not sure I understand why it's cleaner than a proper dual rail supply. The idea that the Star earth is cleaner only makes sense if it prevents currents flowing where they shouldn't be.
However, I've noticed that the issue of power supply design has been the main item over the course of this thread. It has also attracted a lot of different input and different opinions. It's quite fascinating how it's developed. There have been simulations and references to prior work etc. And at the end of the day we are now talking about subjective impressions based on listening experience to headphone amps that have used SMPS - a good topic to ignite further discussion
I'm not sure it should be my decision how to proceed but I think we've converged on the 'diplomatic' solution already. In other words, we will take the psu off board (and potentially out of the box). Put the necessary filtering on the board, virtual earth option included, to ensure good psrr. Then builders can make their own choices of power supply including the use of a SMPS wall-wart.
That works for me
p.s. Nico - no apologies for opening cans of worms, I think that was the right thing to do, it's what we wanted from this thread. And your input may have saved us from being 2nd best.
However, I've noticed that the issue of power supply design has been the main item over the course of this thread. It has also attracted a lot of different input and different opinions. It's quite fascinating how it's developed. There have been simulations and references to prior work etc. And at the end of the day we are now talking about subjective impressions based on listening experience to headphone amps that have used SMPS - a good topic to ignite further discussion
I'm not sure it should be my decision how to proceed but I think we've converged on the 'diplomatic' solution already. In other words, we will take the psu off board (and potentially out of the box). Put the necessary filtering on the board, virtual earth option included, to ensure good psrr. Then builders can make their own choices of power supply including the use of a SMPS wall-wart.
That works for me
p.s. Nico - no apologies for opening cans of worms, I think that was the right thing to do, it's what we wanted from this thread. And your input may have saved us from being 2nd best.
Last edited:
Member
Joined 2009
Paid Member