I've built a handful of headphone amps during my short time in DIY audio including the TangentSoft CMoy, GS DAK Novo and Beezar SSMH as well as a pair of Chip Amp mono power amps (still under construction) with associated ESP soft starts, AMB Labs E24s and ESP DC offset protection. Now I'd like to build something from the ground up.

I'd like to make something simple but a step up from the TS CMoy to drive my RS1s. So, I've been browsing the data sheets to come up with the basis of a design and stumbled across a schematic for a high performance headphone driver in the BUF634 datasheet (fig 5, page 9). I have reproduced this schematic below, adding a pot along the way.


High Performance Headphone Driver

Well, I guess that TI/BB must know what they're on about, right? So what do you think of the above schematic?

I'm assuming that the gain is altered by changing the relative resistance of R2 to R3. I make the gain in the above circuit 20 (R3/R2), although BB report the gain as 21. Ideally, I'd prefer a gain of over 5 but less than 10.

I'm surprised to see that there isn't a resistor between between the opamp and buffer. The TS CMoy certainly had one after the opamp (R5) to reduce noise....although it was optional. I also notice that the M3 also has one between the opamp and discrete buffer.

In fact, (to my inexperienced eyes) there would appear to be some similarities between the simple circuit above and the M3 / PPA circuits except that the above uses an IC for the buffer whereas the others use discrete components. Am I right?

Once the circuit for a single channel is sorted, I intend to use it implement a three channel design. Then I'll consider the power circuit.

Any thoughts, pearls of wisdom or recommendations greatly appreciated.

Place a 10k resistor from 6 to 4 on the opamp for class A biasing. OPA627 respond well to that, lowering the distortion without dropping load impedance too much for a nicer sound.

Also, replace the 5k feedback resistor with 2.5-ish for a gain of 11. Then you can use the OPA637 instead of the 627 which is a nice upgrade.

Maybe a couple o' Ohms from opamp to buffer, but whatever. The less resistors in the signal path the better, although a small one is insignificant. It's more about elegance.

Add a really nice PSU and you're done

Nothing new though...

Thanks Slash47. Grateful for the reply. I'll include your suggestions.

Not trying to change the world, just do something nice and simple.

It's a good design: tried, tested and loved by many. I wasn't hating

Hmmm. Just googled "opamp class a biasing" and Tangent's site came top of the returned results. A quick scan reveals a number of ways to implement the biasing. Perhaps I'll take a read and consider something more complex than a single resistor to bias the opamp.....I sense an increase in complexity coming on!

Holly came from Miami, F.L.A.
Hitch-hiked her way across the USA
Plucked her eyebrows on the way
Shaved her legs and then he was a she
She says, Hey babe
Take a walk on the wild side
Hey honey
Take a walk on the wild side





se

Please note that you'd really just be doing a PPA in that case. You might as well just do that! Nice and easy, with a ready made PCB.

I personally don't like the lack of space on some ready made PCB's, though. I like expensive caps when I DIY

(bigger and shinier is always better, right?)

EDIT: Cool design with the transformer, but do you think it'd sound nice? I'm not sure... It'd prefer a good 1uF film cap and incorporate the buffer in a low noise opamp's feedback loop.

It certainly earns some cool point, though, which is very important.

Yeah, but the simplicity of the BUF634 over the discrete buffer on the PPA has its attractions. And the JFET cascode doesn't look too complex does it?

Quote:



Radical! Have you actually tried this?

No, it's very easy. I just don't think it sounds better then a large value resistor.

I might've meant the Pimeta, BTW. I confuse this stuff all the time, sorry.

If you use the opamp for line driving and want to bias it, you'd have to use ~2.5k for biasing which might kill some dynamics. In that case something else might be nicer, like JFETs.

Quote:



I've every reason to believe it would sound wonderful, provided certain input criteria were met (i.e. no DC offset, 200 ohm source impedance or less, no more than 2 VRMS peak).

I've been designing using high quality microphone step-up transformers for voltage gain for years and when used within their limits, I'll take a good quality transformer over any active solution.

Quote:



Hey, whatever works best for you.

Just thought I'd toss out something a little more out of the ordinary is all.

Quote:



Thanks. Though cool points only go so far. Still have to satisfy on other levels.

se

Quote:



Not that specific transformer with that specific buffer IC, but as I mentioned previously, I've been using this same basic approach for preamps and amps for years.

Haven't previously done a dedicated headphone amp yet, but this is what I'm in the process of working on toward that end:



Once I get it fully fleshed out I'll post all the details of it here in the DIY forum if anyone wants to build their own.

se

PJPro, try this one

author site (in russian)

Quote:



Steve?

Quote:



...Dad?



Yeah, it's me.

se