What do we think of this headphone amp with replaceable opamp

[ef37a]

I find Amazon have them at £44.00 Hmm, tempted but...MOT next month!

I'm not sure that there's any point buying one Dave. I'll bet you've probably got something lying around that will work just as well.

I have James, an HA400 but I doubt the specc' is as good as the U3? Also the wee Behrry needs a 12V wall rat whereas USB power is much more convenient.

In any case I was thinking of my son.

Dave.

[James Perrett]

I have James, an HA400 but I doubt the specc' is as good as the U3?

You are right, the HA400 is a much simpler design compared to the U3 and omits the output transistors. The headphone output comes straight from the output of an op-amp via 47 ohm resistors. They're great for non-critical applications but don't sound as good as the headphone output on my Zoom U-44 or RME Digiface USB.

[Hugh Robjohns]

The argument has raged for ever whether we need amps to go much beyond say 30kHz and has never, to my knowledge ever been resolved? I don't think there is any other area of "control" engineering where systems are made to respond to signals way outside what they will experience?

I agree. The old design adage is, 'The wider open the window, the more muck flies in'.

But... I was reading only yesterday about Rupert Neve's philosophy of maintaining an internal bandwidth to 200kHz. (Im still not clear if its just the internal bandwidth, which wouldn't be an issue for me, or the input bandwidth... which would).

The loony 'record everything at 192kHz' brigade argue it is necessary because most instruments produce ultrasonic harmonics — which is true — and they are obviously critical to the listening experience.

I'm not convinced... but they claim Japanese experiments prove humans can perceive ultrasonics... again, I'm not convinced.

I've also read that those ultrasonic harmonics can interact to produce modulation artefacts below 20kHz — which is true — and that proves you need to record them...

But surely, if they interact in the (non-linearities of) air to produce audible frequencies, won't they be recorded by a 44.1kHz system anyway ?

And if the interaction occurs after recording doesn't that imply unwanted non-linearities in the recording / mixing system?

More practically, the number of mics capable of capturing harmonics up to 100kHz can be counted on the fingers of half a hand! And none of them are in the list of the world's most desirable mics!

So, my philosophy is that products should have their bandwidth intentionally restricted to around 40kHz at most — an octave above 20kHz. And likewise — exclude anything below 10Hz.

I record challenging acoustic stuff at 96kHz and 24 bit. Everything else at 44.1/48 24bit.

None of my recreational listening is ever higher than 44.1/16....

[Tomás Mulcahy]

Well that escalated quickly That's way more info than I had hoped for. Thank you all, very illuminating. Given the faffing about it would take to hook this up to my little rig (because all the i/o is doing other stuff already via a patchbay) the benefits do not outweigh the effort.

I suspect that's probably more down to the fact that the 2nd gen solo is bus powered from a type-B connection. My first gen 8i6 will happily deliver ear-damaging volumes into 250 Ohm Beyers.

Ah thank you. That's reassuring. I'm using a first gen 6i6- a thing that refuses to break! I might try DT-880 so...

[Martin Walker]

None of my recreational listening is ever higher than 44.1/16....

Ditto, and if I buy any digital download albums that turn out to be 48 or 96kHz sample rate and 24 or 32-bit, I down convert them to 44.1/16 anyway.

[Folderol]

At the risk of severe criticism, I still have (and very occasionally use) A tiny headphone amplifier I designed and built a {cough} few years ago. I uses two 7611 CMOS Op Amps (one for each channel) that then feed an AC126/AC127 pair wired as emitter followers with their bases wired directly together i.e. no bias. This all runs from a single PP3 battery and is suitable for medium impedance phones.

In spite of the unbiased transistors, distortion is surprisingly low. It's a long time since I actually measured it but in my notes (I don't throw anything away) I have THD at less than 0.1% at 400mW into 15 ohms and rising to < 0.4% at 10mW. Frequency response is -1dB at 50Hz and 15kHz.

The reason you can get this distortion performance is that germanium transistors don't so much have a pedestal voltage, but a 'soft' knee that starts at about 0.05V rising to 0.2V and 100% feedback takes care of the rest.

[ajay_m]

Though to be fair, any half way decent opamp driving a pair of common silicon transistors (rather than germanium) with an open loop gain of 1000 such as the NE5532 (and that's at 10KHz, it's way higher at 1KHz) is gonna give you excellent THD (assuming the feedback loop is from the actual output of course). Because after all that's what negative feedback does. Traditionally you also of course fed some quiescent current through the output pair to minimise the crossover knee point where one of the pair transitions from conducting to non-conducting or vice-versa, but this was because your primary amplification stages prior to that likely had considerably less open loop gain than an opamp, back when discrete amp circuitry was the norm.

Don't forget that the output stage in this scenario also has gain, hFE for a decent small signal silicon transistor ought to be 50 or more at a collector current of 200mA say. So then assuming the closed loop gain is maybe 10 or so, you can see that regardless of the non linearity of the output stage, the very high gain of the opamp will almost completely cancel this out via negative feedback.

[Nazard]

But... I was reading only yesterday about Rupert Neve's philosophy of maintaining an internal bandwidth to 200kHz. (Im still not clear if its just the internal bandwidth, which wouldn't be an issue for me, or the input bandwidth... which would).

Timothy de Paravicini had a similar philosophy and advocated an extended recording bandwidth so as to include all mixing products. I found his thoughts and writing on this quite convincing and still do, (although you can never include all possible mixing products).

As an organist, I often create unharmonically related sounds using various stop combinations and it strikes me that on many commercial recordings the full 'character' of similar stop combinations is often not fully realised. Could this be related to the restriction of mixing prducts in the recording chain? Possibly and it is certainly an interesting area, but one which is very difficult to draw any absolute conclusions upon.

[ef37a]

"Don't forget that the output stage in this scenario also has gain," Not voltage gain though Ajay? The vast majority of power amp circuits employ the output devices as emitter/source followers with slightly less than unity voltage gain.
Self (again!) has written that some designs used OP stages with up to 6dB of gain but his opinion is that parasitic oscillation is hard enough to keep at bay with no gain and excellent results can be had with followers anyway.

On the bandwidth debacle we have to come back to transducers. Hugh has said that very few mics get to 100kHz. How many speakers have much of an output past 25kHz or so? What is the point of amplifying something you cannot reproduce (and probably can't capture anyway!)

When I entered the electronics world I had about 5 years grace then they invented the Silicon Planar Transistor! I spent the next 20 years keeping RF out of audio kit! Make the bandwidth fit the transducers...plus a bit for jazz.

Dave.

[ajay_m]

True, a single complementary pair of transistors will have slightly under unity voltage gain but significant current gain, which for small signal transistors is somewhat higher than more high powered devices. Not enough coffee this morning when I wrote that