Snuggles

[widea]

This headphone amplifier is based on the Tori amplifier, which name seems to come from an agressive dog.
I hope to have tamed that dog, hence the name 'Snuggles', Simulating the amplifier in LTspice was not easy for me to get right. I probably should have called this amp 'Struggles' instead of 'Snuggles'.
Anyway, I am planning to build it using a NE5532 or similar, but what do you think? Does it have a chance of success?

 

[Mooly]

It looks workable

Not sure R3 does much tbh and R1 (and using a resistor is fine) could perhaps be improved with a simple transistor current sink instead. As the amp is AC coupled the DC balance is not so important but would be an issue if you converted it to a dual rail supply.

 

[widea]

Hi Mooly, thanks for your quick reply.
R3 actually did the trick to make it work in LTSpice, but like you I am not sure why it is needed. I will not be surprised if it will just work without it, like the original Tori amp did.
My space here is limited and I only have a single rail supply. That is why I made it AC coupled. Apart from I do not hear the difference anyway

 

[Mooly]

Are you using one of the opamp models in LT or just a generic symbol type model? I don't see a type unless you have hidden it.

 

[widea]

I used a generic model. I do not have a model for a NE5532 available.

 

[Mooly]

That sim is not behaving as expected...

Try an LT1056A opamp (it is like a TL072) and a 2N2222 model for the transistor. Add a cap in the feedback return.

R1 is not allowing enough current into the load on negative half cycles. Reduce the value or add a current sink.

 

[widea]

Will try that, thanks Mooly!

 

[widea]

Snuggles rev.1 looks good ) although the 2N2222 will be close to it's max. dissipation of 625mW, will try D44VH10 for that, still have a few of those laying around).

 

[Mooly]

The great thing with simulation is that overloads don't burn a hole in the screen

Try this...

 

[widea]

Two bones in one, stabilizing the emitter current by replacing the 50 ohm resistor and sharing the load over two transistors instead of one
Now Snuggles is able to fetch the current I want to hear the result, burning a hole or not

 

[Mooly]

You can alter the emitter resistor to get the current you want. If you are getting the volume you need then without distortion then increase its value a little.

 

[widea]

Ok, I will keep that in mind. The values you choose I expect the volume to be sufficient, using the 80 ohm headphones that I have.
First I will build something, diy style. I am very curious how it will sound.

 

[Mooly]

It should sound good. There are a few things you can experiment with as well such as taking the output transistor out of the feedback loop.

 

[widea]

taking the output transistor out of the feedback loop

The harmonics this causes may sound interesting and if so I will make that 'shortcut' switchable, if it does not cause oscillation.

 

[Mooly]

The harmonics this causes may sound interesting and if so I will make that 'shortcut' switchable, if it does not cause oscillation.

That's a good idea. I can't see any stability issues with a design like this.

You can also use an LED for the current sink which saves a transistor. Resistor value depends on the LED colour (different colours have different forward volt drops)

If you make the signal caps '1' in the sim (1 Farad) it effectively takes the time constants out of the sim and you will see a more accurate FFT.

 

[widea]

You can also use an LED for the current sink which saves a transistor. Resistor value depends on the LED colour (different colours have different forward volt drops)

Which is also a good idea!

If you make the signal caps '1' in the sim (1 Farad) it effectively takes the time constants out of the sim and you will see a more accurate FFT.

And I thought the .asc file was corrupted I learn a lot Mooly!

 

[widea]

Finally, after some virtual experimentation I decided I will build the following circuit, output transistor switchable inside and outside the feedback loop, using a LM317 as a constant current source.
I expect the switch between Q1 inside and outside the feedback loop to be interesting, soundwise.

Q1 outside the feedback loop:

Q1 inside the feedback loop:

 

[Mooly]

Would only add that C7 at the input really needs a resistor adding in series with the input to define the response. As it is now it just adds a cap across whatever the input source is and that is not ideal. Try a 2k2.

 

[widea]

Hi Mooly,

Previously in an other circuit I noticed that just using C7 solved a problem with a crackly volume pot, appearently caused by DC. That is why I added it as a precaution. But I will keep your suggestion in mind of course, thanks.