Neurochrome HP-1

Who knows. Some folks like some stuff. Some don't. It is what it is.

In my mind it is possible perhaps that an amp would produce distortion large enough to mask recording nasties. Which in turn may sound "good" 0.3413% of the time, and horrid 99.6587% of the time. I don't think any of Schitt's amps do this "magic". I dunno.

One likes the amps one likes, and that's it. It may be cuz one is shinny and the other one is not.

As far as what @Psalmanazar is trying to say, I got no idea. But I know he hates the O2, and while NwAvGuy is not my role model, I don't find his amp so bad (at low gain).

(Note I'm audiophile level 30, while @Psalmanazar is proly level 41)

 

I'll check it out eventually. Some folks that I do not consider deaf seem to disagree with you. In all likelihood this means one thing for sure: The HP-1 is not for @Psalmanazar.

 

I do agree having a current buffer only mode (1x gain) would be useful. Particularly with sensitive cans indeed. But 6 dB and you are getting killed with HD6x0?! What DAC are you using?

I'll check this out. I do have HD600, and I do have sensitive POS cans.

 

The claimed BW of the amp is 71 kHz, and things look pretty flat to 20 kHz. Maybe he is using a high order low pass filter, but I doubt it. AFAIK DACs typically have to roll of the BW sooner and may probably contribute more phase distortion (higher order low pass filters).

But dunno.

 

I believe phase distortion has a lot to do with bandwidth because of how close to the cut-off you are, and because most every analog domain filter is effectively an IIR.

Your magnitude response might be wonderful all the way to the hundreds of kHz. But your phase starts to give up much earlier. Say you put your low pass corner frequency at 1 MHz (or somewhat close to, can't remember). You might be off by a few degrees in phase at 20 kHz.

What this means is that putting the cut-off too close will result in "phase aligning" going down the drain for treble frequencies (or lower, depending on how low that low pass cut off is).

Everything is connected @borrego. Such is the force.

EDIT: It's all about trade offs, you could put in a Bessel low-pass, and end up with as decent a phase response as possible, but almost no rejection. You could put an Elliptic low-pass and have shit loads of rejection, but horrid phase response. Most serious folks use Butterworth and similar trade offs apply. Most every other rando use a single pole low pass filter which will still f**k up the phase as I described. The devices themselves are not infinite BW and depending will f**k up the phase as well w/o help of a low pass filter (not too bad IMO though).

 

@gbeast, before you edited your post you mentioned issues with the O2 in terms of attack and stuff.

I actually don't think you are crazy there. The O2 uses some op amps that have somewhat limited slew rate. One of them IMO dangerously so. I think is the current buffer (3 V/us). With some voltage swings, I think that may be problematic. Hell, I think Doug Self himself (NwAvGuy's hero) said that 3 V/us is uncomfortable.

I don't think the HP-1 is SR limited though given the op amps used, unless something funky is going on.

 

One more think folks. I dunno why many make a distinction between time domain and frequency domain (linear) distortions.

Time domain and Frequency domain are two ways of looking at the same data.

Frequency domain has TWO components: Magnitude and PHASE. The fact that nobody seems to publish phase response vs frequency does not mean it does not exist.

 

DC Block and HF rejection

You will need a capacitor IMO if you want to keep things safe from giving their magic smoke. If you want to use a servo, you will add a capacitor there (HP-1 uses a servo from what is claimed).

It may also depend on the devices you use. And your upstream equipment.

If folks are thinking servos, I know some folks think a servo take the capacitor out of the equation. IMO it does not. It can allow one to use smaller caps and bigger resistors for the DC block, but the caps are still there. Furthermore, they are limited in how much DC you can block. IMO it's risky.

With out input coupling caps (low pass configurations) DC offsets will go through the gain stage and may damage your driver. On top of that, most decent amps use an RF filter which might come in handy if you are going to listen to things that also couple RF noise (high pass).

If there are other ways to block DC w/o blocking caps and servo let me know. I'll be all over it.

(You may think not-bipolar and support only low offset DACs, but then folks will complain about other stuff and may not use their 100 mV DC offset favorite weirdo semi-broken DAC)

NFB stuff

As far as global vs local feedback usage, I think it depends on the devices you are using which one is better. I believe the HP-1 uses global feedback BTW. O2 is local.

Current feedback vs voltage feedback, I prefer voltage. Current feedback devices I've seen can give you a lot of current drive, but their BW depends on the feedback resistor AFAIK, so they are not as versatile. It depends on what you are trying to build.

For our phase distortion discussion, current vs voltage feedback, and local vs global feedback probably maters little, and it's more how you implement things. It's not just this or that thing in isolation. Be careful about absolutes. Dark side of the force there.

 

Also a servo only (no coupling caps) might also reduce 1/f noise I think. But again, it is risky. And I think HP-1 uses a servo anyways so dunno. Putting a sufficiently small resistor with the coupling cap might alleviate the 1/f noise issue, among other things.

Many designs use both (input coupling cap / output servo). The problem being that coupling caps at the output are proly not a good idea given the load impedance being somewhat of a random variable. So for inputs they use coupling caps, and for outputs they use servo.

O2 only uses input coupling caps. The output goes out wide open because the dc offset of the current buffers is awesome according to NwAvGuy. I think that is risky, and IMO a flaw in his design (given worst case specs). It will also limit how many devices he can parallel. He is using bipolar op-amps which tend to be low noise, but have relative high bias currents. His favorite op-amps are no exception.

In the O2, I think the high frequencies are limited by the gain stage compensation caps and the RF blocking at the inputs. I got no idea on the HP-1, but maybe things are similarish.

 

Disagree on the inevitability of local feedback. Here is the LME49600 datasheet:

http://www.ti.com/lit/ds/symlink/lme49600.pdf

Figure 28 describes a global feedback architecture with integrated servo for headphone amplification. Basically a "The Wire" deal. Lot's of designs based in that application me thinks.

Linear Technologies is littered with app notes on global feedback designs as well. Hard to keep them stable at unity gain if I remember correctly. Maybe that is one reason the HP-1 does not offer 1x gain? Again, dunno.

Also, I agree the OPA1611/1612 are very interesting (little SMT thingys).