[rickcr42]

I will say this one time and only one time.No discussion,no comments here it is :

Do NOT direct posts in my direction if you do not want me to come back into this thread and post my feelngs on this topic.I tried to let it die but some of you are far too anal and can not let go.You bring me in and you will have to deal with me.Simple.
So don't go there or put on your asbestos suit.Cast a stone in my direction and I will come back with a truckload of rocks so chill out.

Quote:

Another point that I would like to add, is that most of those "manufacturers", became manufacturers years after they were here, in headwize, and in many other forums and sites, helping people and offering his free DIY support, the case of ppl, tangent, voodoochile, JMT, Meier, and so many others, of course after seeing that many of other "not so helpful people" are making money out of their knowledge (and even out of others knowledge) well why not them??? Inside the rules or not, with a commnercial point of view or not, what I feel at the end, is that they are helping many people to get out, of their problems....

You need to think before you type.This alone if I so decided could be if I followed the letter of the rules is grounds for banning so walk easy.I say to you right here and directly.Do NOT carry this.
Let it go right here and right now or it will get ugly and neither of us want to go there.I will also say if you ever meet me in person,please stick your tongue out at me.I would enjoy that.

Finally-do NOT respond to this post.Keep the damn thread ON TOPIC because if I check my email and I see a bunch of off topic posts I will not only delete them but will take other steps as well since this is a very clear warning to cease.Consider it formal.

thanks

rickmonster

 

[amb]

Quote:

Originally Posted by grasshpr What I'm saying hear is using a low pass filter to provide smooth transitioning into the proper audio signal. A low pass filter is merely a stable state-space system that has an input and spits out an output. If the input is a quasi-periodic signal, then the output would simply be the same quasi-periodic signal with a phase shift. Only difference would be that this system would have to transition from some initial condition (which would be ground, or the way you have it which is in disconnection mode).

I see what you're trying to accomplish. Phase shift the signal so that when we reconnect the headphones it occurs at zero-crossing time to avoid the small discontinuity in waveform. Good thought, but...

If you're trying to make a phase shift in a uniform way with respective to frequency then you need an all-pass filter, not a low-pass filter. But more to the point, you've got to remember that a relay is a mechanical device and it takes a small amount of time for the contact to close after energizing the coil. To do what you suggest here assumes that we know exactly when that is going to happen and, we don't. And even if we did, transitioning from the phase-shifted version of the audio to the non-shifted version will still incur a small discontinuity in the waveform without doing something really fancy, and another relay is not going to cut it. And relay contacts do "bounce" when closing, how would that affect the signal?

Quote:

However, you still have your ground signal connected to ground (which can supply some current to the headphone right?).

Actually, no. Trust me. Do a little measurement on your own. Hook up your DMM in AC milliamps mode in series with the connected ground, play a test tone and watch your meter and you'll get a reading of some amount of current. Now, disconnect the positive side of both outputs. Still got a reading?

Quote:

Same principle as static discharge, when you touch ground on a metal surface and your body is at another potential.

Static discharge is tens of thousands of volts or higher, arcing across air. Headphone amp DC offset is much smaller and cannot do that. Have a look at the schematics of almost any piece of audio equipment, and you'll find that the power switch only switches one side of the AC line and there is no need to switch both. Same principle -- it breaks the circuit, and we're talking about AC mains here, which is much higher voltage than any DC offset that this protection circuit is going to encounter.

At any rate, I think you're obsessing. What you suggest is far, far beyond what this circuit is intended to do.

 

[grasshpr]

Quote:

Originally Posted by amb You've got to remember that a relay is a mechanical device and it takes a small amount of time for the contact to close after energizing the coil. To do what you suggest here assumes that we know exactly when that is going to happen and we don't. And even if we did, transitioning from the phase-shifted version of the audio to the non-shifted version will still incur a small discontinuity in the waveform without doing something really fancy, and another relay is not going to cut it.

You are right! Thats why I mentioned building a low-pass filter with minimal phase shift, this would allow you the reduced transition from minimally phase shift signal to the actual signal. This discontinuity will be much much less (depending on the phase shift) compared with transitioning from open to some audio signal directly (which could be large depending on how your potential on the open end is). At worst, you will hear a minuet crackle (which in my opinion is better than a pop. I guess I'm sounding like Rice Crispies).

Quote:

Originally Posted by amb Actually, no. Trust me. Do a little measurement on your own. Hook up your DMM in AC milliamps mode in series with the connected ground, play a test tone and watch your meter and you'll get a reading of some amount of current. Now, disconnect the positive side of both outputs. Still got a reading?

You are right again, however, I was not referring to only normal conditions where the open end will not effect the output to the headphones. I was talking more about cases like when the air is very ionized (usually due to high moisture content in air) where the voltage difference between ground and your open side will measure slight voltages. Given, these values will not be very large, but they exist. Test for yourself, measure ground and leave the other end free from a DMM and see if you get some stray voltage.

Quote:

Originally Posted by amb Static discharge is tens of thousands of volts or higher, arcing across air. Headphone amp DC offset is much smaller and cannot do that.

Of course I was exaggerating this situation with the point on static discharge (and was in no way trying to compare the magnitude of current flow compared with what you have in the headphone end). I was just pointing out that current may flow in/out of ground if you do have one end open.

Quote:

Originally Posted by amb Have a look at the schematics of almost any piece of audio equipment, and you'll find that the power switch only switches one side of the AC line and there is no need to switch both. Same principle -- it breaks the circuit, and we're talking about AC mains here, which is much higher voltage than any DC offset that this protection circuit is going to encounter.

Yes, I use the same principle for my amp, however, power supplies are different compared with outputting audio signals to headphones (where the sensitivity to stray voltages, whether it be ever so slight, could be annoying). What makese power supply circuits different from headphones is that power supplies usually have additional regulation components where noise will not effect the output much. Headphones don't have that luxury.

 

[Garbz]

Acutally isn't it common to use DPST switches to switch both L and N in the powersupply? This would be for safety purposes more then anything, especially in european countries like Austria where the power plugs can be inserted in either direction.

Also this would allow you to use a balanced powersupply where both wires are live, which was mentioned in the power filtering thread here last week.

 

[guzzler]

Here's a little single layer layout (2 wire links). Only for 24V supplies, as I've missed out the little regulation stage. I've replaced the very expensive BUF634 with just a simple transistor pair, save you a few pennies there

Also, I've replaced the pair of 2N3904s in the Darlington configuration with a single Darlington transistor. Looking at a ZTX605 in an E-line package (shown on board as a TO-92) for that position. Costs about 2.5x as much as 2 x 2N3904, but saves on board space. The whole board measures 66x24mm

2D Layout

Eagle3D Image: (not rendered relay, pot, or diodes)

Eagle Layout File: here

Should be correct I think, haven't religiously checked it yet. Not built it yet, may etch one in the next few weeks. Enjoy...

g

 

[AtomBoy]

Guzzler,

Neat neat neat! Could you post the schematic (.sch)?

 

[guzzler]

It's just the original schematic, with the two transistors, the zener diode and two resistors missed out in the power supply. The transistor pair is just 2N3904/6 with emitter resistors and common base.

g

PS, if you can wait a while, I'll try and get one made so you can see what it's doing

 

[AtomBoy]

Thanks, Guzzler. Yes, I can wait. I guess.

I'm still learning Eagle, but I've done a version of the schematic here:

http://www.swcp.com/~atomboy/amps/e12.sch

My parts are sized a little different from yours. I've got an order going at Mouser that I've been working from. I could post it if there's interest.

 

[amb]

Quote:

Originally Posted by AtomBoy I'm still learning Eagle, but I've done a version of the schematic here: http://www.swcp.com/~atomboy/amps/e12.sch

Your C6 is missing a connection to the base of Q3, and Q4 is wired wrongly (reversed collector and emitter).

 

[AtomBoy]

Quote:

Originally Posted by amb Your C6 is missing a connection to the base of Q3, and Q4 is wired wrongly (reversed collector and emitter).

Thanks, amb! I've fixed it now (I found the mirror button for Q4

). Updated .sch posted to original location. I almost had a .brd layout I wasn't embarrased to post. Maybe in a couple of days.

Do you have any thoughts on replacing the BUF634 with the transistor pair ?

 

[amb]

Quote:

Originally Posted by AtomBoy Do you have any thoughts on replacing the BUF634 with the transistor pair ?

Looks usable, I think I'll plop those parts onto my breadboarded prototype and see how well it works.

 

[amb]

I checked guzzler's dual-transistor rail splitter on the breadboard. Here is the schematic of just that portion:

While this splitter basically works, there are some issues.

1. Since the transistor pair has a rather low input impedance, and the current flow into the transistors' bases is significant, it impacts the accuracy of the splitter operation. I tried reducing the voltage divider resistors from 10K to 1K and it helped a little but still insufficient. Also the lower value resistors increase the current draw of the circuit greatly.
2. Due to the open-loop nature of this buffer, the emitter resistors increase the output impedance of the buffer, which also affect the splitter's accuracy. I tried dispensing with the emitter resistors and just connected the transistors directly, and again that helped, but not enough.

The net affect of these shortcomings is that the virtual ground potential changes significantly between just after power-up and when the relay kicks in. I measured a ~800mV "shift". Now, since the circuit is supposed to detect as low as 70mV offset, this large shift in the virtual ground would cause start-up problems with the circuit (i.e., the relay may not close at all after power up due to the skewed "ground", or it may click in and out in an oscillatory manner, depending on how the zero-reference trimpot is adjusted).

Given this, I did a bit more testing and determined that the BUF634, while better than the dual-transistor splitter, also suffers a bit of virtual ground shift (probably also because it's open loop).

I replaced the BUF634 with a OPA551 opamp hooked up as a voltage follower and the situation is much better. This is clear evidence that high input impedance, low output impedance, and closed-loop feedback all contributes to improved accuracy of the splitter, which is required for optimal operation in this circuit.

I have changed the revision of this circuit to 1.1 and the following is the new schematic showing the switch from BUF634 to OPA551. The OPA551 is also less expensive than the BUF634 (8-DIP version $3.94 vs. $6.10 from Digikey).

 

[guzzler]

To lose even more money, at the expense of a couple of extra parts, would it be plausible to use a cheap opamp, say another TL071 (or use one TL072 for the whole thing), and the transistor pair method (thanks for the schematic there amb), and wrap those up inside the feedback loop of the opamp, to gain the high input impedance needed, with the low cost of that method? I’ll try and knock up a quick layout (and schematic, for once

) later on today

g

 

[guzzler]

Well, here's my altered version. I don't like it as much as the first one, but nothing comes for free

Dimensions are 52x37mm and leaves plenty of room for some mounting screws etc... edit: estimated cost is £5 including a high quality relay

So, layout

**WRONG**

Eagle3D

Revised portion of supply splitter:

Eagle .brd file

g

 

[amb]

guzzler, I played with your revised splitter a bit on the breadboard. The two transistors basically operate as a class B output stage, where there is a dead zone in the middle where neither transistor is on. This makes the virtual ground potential "touchy" and a bit unpredictable. I suggest the following:

1. Eliminate the emitter resistors and tie the two emitters together.
2. Add a 47 ohm resistor between the bases and the emitters of the transistors. This makes the opamp do a little of the work in that dead zone to keep things stable.

Edit: this is not good enough, see post 97 below.