The Kuartlotron - keantoken's simple error-correction superbuffer

[nattawa]

Kean, I don't really get why using a printed inductor that takes a chunk in the layout. I tried to simulate the performance in Spice on different inductor parameters, and it seems the buffer is quite tolerant to the parasitic of the inductor. I had a 2-ohm series resistance, 47pF parallel capacitance, and a 10K parallel resistence to make it a really crappy inductor, and the THD at 20KHz and 100KHz remained unchanged. Do you think we could then get away with a hand-wound multi-turn inductor with fine magnetic wire around a 1/4w, 100k resistor as a bobbin? Or maybe an off-the-shelf inductor?

[Destroyer OS]

Are you going to wind mine?

[nattawa]

Of course, if you pick up at my place, with some booze if you don't mind.

[keantoken]

You're right, almost any inductor will do. The important frequencies are in the MHz region. I've used a 1.6uH ferrite bead and it was fine. I will add a solder bridge so the inductor is optional in case you want to use your own. Or I may eliminate the inductor entirely, but I don't see why a self-wound inductor would be superior.

[Project16]

Congratulations Kentoken for this schematic!

I made my listening tests this morning and I was very surprised by the sound quality and what still surprises most is the lack of breath or noise.
The sound is very transparent but brings a kind of dynamic, basse punchy and very fine treble but it is really impossible for me to describe the quality of listening and of course no distortion.
In summary it is a success this circuit and it's really nice of you to share.

Next to the post that contains a .zip my version.

Regard's!

[keantoken]

Thank you.

The layout is still materializing, slowly...

[Destroyer OS]

I'll buy it!

That looks easy to work with. That spacing on the inductor Vias might be tight. I think two SMD pads are nice for solder bridge. One can also insert a tiny piece of wire into solder bridge if it's tough. That way it's easy to remove as opposed to vias. Vias, round, also are hard to if not impossible to bridge.

Also, it looks like you haven't bypassed inductor.

[jls001]

Originally Posted by keantoken

You could make R8=100k. The reason it's 10k is for DC setups since I was asked for a DC version. However this might make your volume pot uneven.

None of your options can damage anything (unless your system is very sensitive to offset). If I were you, I would try them all by listening. After all, every amp responds to source impedance differently, every source responds to load impedance differently. I can only give rough guidelines on what will lower distortion the most. But low distortion doesn't guarantee good sound.

My suggestions:

For driving a power amp: use the pot before the buffer, so the amp benefits from low source impedance.

For buffering sources needing a light load: Use R8=100k. If driving a power amp, a 200k pot before the buffer may be the best option, if it is statically shielded.

To buffer sources needing constant load impedance: Use a 10k or even 5k pot after the buffer.

Thanks, Keantoken.

I changed the CITEC pot (oops, it was a 50K log. I thought I had used the 10K log) to an ALPS Blue Velvet 50K log. I didn't hear much sonic differences but the pot curve on the ALPS seem to be more logarithmic. 6 o'clock to 9 o'clock is gradual and doesn't get loud. On turning it up further, 9:30 starts to get louder, and by 10 it is about comfortable from the listening chair. On the CITEC, the change in SPL between 10 to 12 o'clock wasn't much and it used to sound like there wasn't much change especially when lowering the volume. I like the "action" of the ALPS more.

In the interest of finding out the best value and position, I will also try 10K log at the current position (before buffer) and also after. I am still waiting for the DACT clone pots.

[keantoken]

I'm all for the Salas regs, except if I have to match Jfets I may never build my own PCB...

Right now I'm using 5mm pin 10mm dia lytic footprints. Are there any special caps people would want to use that wouldn't fit?

[Destroyer OS]

Lead spacing is perfect. Nichicon caps in KZ are 330/470 at 12.5/16mm

I'd assume going 330uf at 12.5mm would be fine, to not make the board overly large. Most caps will be 10mm.