CMOY with some questions

[k00zk0]

Before going out to make a discrete SS amp out of transistors, I had these LT1358's and random components so I wanted to replace this noisy circuit I've been using from a pair of desktop computer speakers for a while.
 
Well, it works, but I've a few questions; perhaps this is all because of this software (LTspice). I got LTspice because it had this opamp in it.
 
http://img9.imageshack.us/img9/9852/cmoycircuit.png
 
This is exactly my circuit. The sines on the left represent an input signal and the 24 ohm is my headphones. The power source is literally four Li-ion cells tapped in the center for ground. The wacky numbers are just to simulate one side of voltage being applied before the other due to a DPST switch. Amp is in unity gain. No potentiometer, just line level input.
 
First issue: there is a DC offset that collects on the output jack when no headphones are connected. Once I short the pins or connect something, it disappears; is this harmful to the phones? It comes up when I worked on the amp with nothing connected (seems to be from a few on-off cycles with no headphones).
 
Next: Very loud on/off pop. I think this is due to the v+ and v- connecting at slightly different times. LTspice shows this to be 2V, I assume this is okay?
 
Third thing:
 
http://img42.imageshack.us/img42/328/cmoyresponse.png
 
Green is the input signal, blue is output across the 24 ohm resistor. I wish I just had an o-scope!
 
This distortion only happens when the signal is very low, such as at the voltages you see from the input on the above pic: 0.00001Vac. Is this normal for op-amps, can I hear this, or is it just the software rounding incorrectly or something?
 
Is this what is termed "crossover distortion"? I'd like to make it class A but that would seem to leave 15mA for drive current.
 
Last thing; I am connecting the ground of the headphone to the V-. Bringing it to ground gives a motorboat noise that changes f based on input. v- is the only stable configuration. Everything okay with that?
 
Despite it being done so shoddily with the rail voltages fluctuating by .2 of a volt and whatnot, it sounds so much better than this last amp which I shorted out all over the place with my hands (bare board on the desk) with noisy TEA2025 opamp. It's a good quick temporary fix. The new issue now is Dell's output being noisy as the processor changes power states, the "whine" that many users of mobile processors report. Need to DIY a DAC...

 

[Avro_Arrow]

You need to go here and learn some basics.
Your amplifier circuit is all wrong.

 

[k00zk0]

I've read most of that and before that, understood a bit about physics.
 
http://tangentsoft.net/audio/cmoy-tutorial/misc/cmoy-tangent-sch.pdf
 
That is pretty much my circuit, except it is unity-gain as the datasheet of this opamp shows can be done with just a short from output to -in, and a 1Mohm instead of a 100k for the early resistor because they're the only ones I had. Instead of the input cap (the source is DC-free) I have an output cap s suggested in figure A1:
 
http://gilmore2.chem.northwestern.edu/projects/showfile.php?file=cmoy2_prj.htm
 
I don't want to add the load resistor because that would cut the available current in half.
 
I don't have the parts for the power supply so I am tapping batteries midway. It is working and sounds good compared to only two other references - this noisy amp which it trumps hands down, and this other laptop's output jack which it is comparable to, if I ignore the noise coming from the CPU power state cycling.
 
I knew there was this issue with how to connect grounds - I don't have a chassis, but still understood there's some issue with keeping things on virtual ground separate from the input ground. I looked and looked and there wasn't a lot of info on my application, just stuff with a dedicated headphone return current collecting part of the amp. Tying all the grounds together, including the center of the four 3V cells, which I would assume would just power the v+ and v- properly in reference to the ground, gives me a motorboat noise that goes away if I bring the headphones ground to the negative, which due to the cap (6.6v across it) keeps the headphones at 0vdc. It works in LTspice, it works IRL.
 
What specifically do I have wrong? Just tell me. I've read that stuff and am following it as far as I can see, besides some mods. Others on some forums also said they tied their headphone return to v- which worked.
 
Thing is, the instability is not shown by LTspice if I tie all the grounds together. Apparently, I need a capacitor somewhere, because this is not a FET input opamp. I don't have any and if I am going to an electronics store I will just get everything for a proper unity gain discrete amp.
 
Just wanted to know, whether there is anything wrong with the way I am using this now (will damage phones or output device?) and if that distortion there was an artifact of the simulation or if that is actually happening?

 

[Avro_Arrow]

OK, a few things.
 
1. While not technically wrong, using a center tapped
battery is considered bad practice. If one side of the battery
discharges faster (or is not matched well) then you can develop
a DC offset equal to the difference in battery voltage. Not all
op amps are affected by this but some are. At the very least
it will cause unequal voltage swings.
 
2. Referencing to two different points. Your input ground is
referenced to the battery center tap while your output ground
is referenced to battery negative. Your output ground should
also return to the battery center tap.
 
3. Excessively large input resistance. All op amps require some
current on their input pins for biasing. FET op amps require an
order of magnitude less current that Bipolar op amp (on average).
This current is drawn through the input resistor. With such a large
resistor and a Bipolar op amp, the DC offset must be very large.
Reduce the input resistor to just a few K will reduce the DC
offset to a reasonable level.
 
4. Output capacitor. A design with a center tapped power supply
does not need an output capacitor. You also have no charging
resistor on the capacitor forcing it to charge when you plug in
your headphones (resulting in the loud POP you mentioned).
Polarized electrolytic capacitors need a DC voltage across
them to sound good. The voltage needs to be at lest 1/2 to
3/4 of the capacitors rating.
 
Address these points and your amp will sound dramatically better.