Aspen Headphone Amp

[ikoflexer]

Does anyone know first hand how it sounds? Has anyone built a prototype, or, is anyone planning to do it before the group buy?

[keantoken]

Originally Posted by forr

Hi Keantoken,

---Also, how about you try bootstrapping to Q7's emitter instead of to the output? This will make the bootstrap much more accurate. One transistor works more, but this should theoretically encourage even harmonics.---

Would doubling this with a second electrolytic connected to Q10 emitter be worth for linearity ?

I just checked in the simulator. I found very little affect (slightly positive) from changing between bootstraps to Q7 and/or Q10's emitter. In other amp simulations, I have noted that the harmonic spectrum changes significantly depending on whether you bootstrap to Q10 or Q7's emitter. I think this is worth experimenting with.

Also, the circuit has begun uncontrollably oscillating after I added series inductance to the caps and decreased the timestep. Increasing the miller cap didn't help, so it seems the problem is in the C-multiplier. I get the same oscillation whether or not I use the CFP c-multiplier. Just added some smaller bypass caps...

Okay, now look at this. The new bootstrap reference helps! 3rd harmonic is about 25db below the 2nd, whereas with normal bootstrapping it was 20db (I measured this at 20KHz since it gave a better FFT). (THD itself actually increases, but I prefer this because the spectrum is, by Hugh's logic, more benign). (I added a .four command to the schematic, results of which you can see in the error log)

Attached are old and new FFT's, and the .asc file used.

PS. I know the Jfet is funny and it can be done without (I agree with Hugh about bias decoupling). Also, the Vout node is before the output cap, since this gives better FFT. And for the record, the CFP C-multiplier can be done without, because the capacitors take care of PSU impedance in the audio range much better (CFP output impedance is about .078 ohms). I selected the caps by clicking "select capacitor" in the cap menu and then edited in my own estimated inductance values.

- keantoken

[Bigun]

Hugh,

I like the extra diode, from my thermal sims this is a safer option and ensures that the bias current will have a -ve thermal coefficient.

Good addition of the discharge resistor to the output cap - I was just about to suggest this (I remember it from NPs Zen amp) and of course as usual you are a leap ahead...

keantoken,

I'm disturbed by the problems with the cap multiplier. We should discard this if there is a risk. The psu needs to be clean and without regulators we need to ensure that channel-channel x-talk is kept low so that may mean dual supplies with passive filtering.

I need to look at those FFTs....

[keantoken]

Well, we need to test the multiplier in real life. I think I'm doing this ESL thing wrong. Is it standard for capacitance multipliers to oscillate!? (by definition they should not, because they have no NFB. Of course, ESL adds a whole new dynamic to this)

I think we only need regulators to keep hum out, otherwise we can't get much better than the caps for load rejection. Why not power both channels from the same supply but with separate LC filters?

We could try boostrapping the LTP source in order to increase PSRR if we go without regulators.

- keantoken

[tessier]

Hi keantoken

You've got a PM

Thanx

Paul

[keantoken]

Hey Tess, sorry for being late. I have actually been putting together a reply. I had to reset my computer too, because the space bar was doing weird things...

- keantoken

[MJL21193]

Originally Posted by Bigun

Hugh,


I need to look at those FFTs....

Here's what I get for Hugh's single supply model:

[Bigun]

I like the change in bootstrap, it keeps a bit H2 in there even at lower frequencies. This is something it would be nice to prototype, but my vote is in favour of your change.

Took me a second to remember you'd moved the OUTPUT to the 'wrong' side of the output cap - frequency response was great at 10Hz compared with my original sim

I've read headphones are pretty good at low frequencies, we should revisit the cap values.

For the cap multiplier, if we want to work that angle further, perhaps we should stick with the non compound version, keep it simple ?

Single supply still preferred. For my last project I used soft recovery diodes instead of the usual one-piece rectifier and CLC filters. No complaints about the results so far!

[tessier]

Originally Posted by keantoken

Hey Tess, sorry for being late. I have actually been putting together a reply. I had to reset my computer too, because the space bar was doing weird things...

- keantoken

Hi

No problem, I was just not sure that you've readed your PM messages.

Thanx

Paul

[Bigun]

Originally Posted by MJL21193

Here's what I get for Hugh's single supply model:

Looks to me like the THD is too low, we don't want this to sound like an opamp do we !? - let's get that H2 up...