You mean it's too low and loading the op amp? That's possible I suppose, especially as with my 300 ohm headphones I'm running at unusually high signal voltages ... though I doubt they ever break much above 500 mV. I regularely put similar loads in my phono stages, since the RIAA network impedance gets very low at high frequencies.
Ok, here's more of a challenge. Sapphire3 vs. the TPA6120A2 output stage in the Xonar Essence STX sound card.
This puts the Sapphire3 at a large fundamental disadvantage as the TI headphone amplifier replaces the final op amp of the DAC's IV/filter stage. So this huge chunk of line level output - the op amp, interconnects, volume control, and (crucially) the very low level analog signal after the volume control leading up to the Sapphire's op amp - is removed from the signal path. Using the built-in headphone stage the volume control is done in software by the ASUS controller IC, after the signal has been upsampled to 24/92 but before it even reaches the DAC. In theory that's a far superior method to doing it in the analog domain post conversion, even with the fancy Goldpoint attenuator I recently fitted to my Sapphire3 amp.
To cut to the chase: it's a very close thing. The TI headphone IC sounds very listenable. In many ways it's like a baby brother Sapphire3: soundstage and overall presence dialed down from a 10 to an 8, but with a sneaking suspicion that the native Xonar headphone output is fractionally cleaner.
The highest complement I can pay it is to note that without looking down to see if the headphones are plugged into the Sapphire3 amp or the computer front panel, I cannot immediately tell from just a few seconds of listening which it is I'm listening to.
It's only on the more complex, difficult musical passages that the additional stability and drive of the Sapphire3 make it clear that the IC, despite all native advantages, is not going to win this.
Interestingly, the background seems blacker with the Sapphire3 though it should be expected to have a lower S/N.
This puts the Sapphire3 at a large fundamental disadvantage as the TI headphone amplifier replaces the final op amp of the DAC's IV/filter stage. So this huge chunk of line level output - the op amp, interconnects, volume control, and (crucially) the very low level analog signal after the volume control leading up to the Sapphire's op amp - is removed from the signal path. Using the built-in headphone stage the volume control is done in software by the ASUS controller IC, after the signal has been upsampled to 24/92 but before it even reaches the DAC. In theory that's a far superior method to doing it in the analog domain post conversion, even with the fancy Goldpoint attenuator I recently fitted to my Sapphire3 amp.
To cut to the chase: it's a very close thing. The TI headphone IC sounds very listenable. In many ways it's like a baby brother Sapphire3: soundstage and overall presence dialed down from a 10 to an 8, but with a sneaking suspicion that the native Xonar headphone output is fractionally cleaner.
The highest complement I can pay it is to note that without looking down to see if the headphones are plugged into the Sapphire3 amp or the computer front panel, I cannot immediately tell from just a few seconds of listening which it is I'm listening to.
It's only on the more complex, difficult musical passages that the additional stability and drive of the Sapphire3 make it clear that the IC, despite all native advantages, is not going to win this.
Interestingly, the background seems blacker with the Sapphire3 though it should be expected to have a lower S/N.
Attachments
Last edited:
tried out more opamps tonight and the AD744JNZ are very nice in the v2 boards. clean, tight, open with vocals being projected very well. highs are clean and instruments like brass and strings have a real bite to them like they should. kick drum and bass guitar are well defined. think i'll be keeping them in for a while.
Just a guess based on the TPA6120A2 datasheet and your descriptions of the sound, if the Sapphire is "voiced" with a LT1357 opamp, it will sonically match the Xonar Essence.
As for the blackness - it's more black with most of the DOAs. Given the technical specs of opamps like the AD797, this should not be so or even audible. It could be due to DOAs running their input and output stages in class A mode over a wider range?
I have 2 Asus STX, I'd be quite disappointed that sapphire doesn't outputperform by a large margin the HO of the Asus.
Low impedance headphones sounds average on the Asus, a black background is a nice start, but expect a little loosy bass, cripsing highs. High impedance beyerdynamics sounds better.
Both line out and headphone out is plagued by an average soundstage definition & layering, image lacks precision. I stoped using both of my Stx... In my humble opinion, Sapphire quality must be dac limited in your tests.
About sapphire design, I wondered if it would be possible to have a switch to (truly) bypass the whole voltage gain stage to use only the buffer when voltage gain is unnecessary?
Is it possible to optimize it for a lower impedance load (25ohms)?
Low impedance headphones sounds average on the Asus, a black background is a nice start, but expect a little loosy bass, cripsing highs. High impedance beyerdynamics sounds better.
Both line out and headphone out is plagued by an average soundstage definition & layering, image lacks precision. I stoped using both of my Stx... In my humble opinion, Sapphire quality must be dac limited in your tests.
About sapphire design, I wondered if it would be possible to have a switch to (truly) bypass the whole voltage gain stage to use only the buffer when voltage gain is unnecessary?
Is it possible to optimize it for a lower impedance load (25ohms)?
We are definitely on the same page here. I agree that the soundstage / definition is a couple of steps off the pace ... but since I find it basically musical and enjoyable to listen to I'm willing to trade off the last word in precision.
Do you have a better solution for computer-based audio? What do you use now instead of the STX?
I was evaluating the headphone stage using 300 ohm headphones. The basic quality is not too different from the line out, just a bit smaller and weaker than when I use the Sapphire. DAC limited? Yes, ultimately, I think so. Apparently lots of people find the quality drops off with lower impedance loads though. Curious whether that's a function of the TI chip or some less than ideal power supply design by ASUS.
As for your question about the Sapphire:
It's not as convenient as a a switch, but it is easy to drive the buffer directly from the rev. 31n boards: just remove the op amp and attach the input to the OA_OUT pad. The circuit is now DC coupled. Careful!
An op amp bypass switch would be a little complicated to implement, as both the input and output of the op amp would have to be reconnected.
The Sapphire is already optimized for 60 ohms. There isn't that much to change - or that much to be gained - from reconfiguring the buffer for 25 ohm headphones. You'd want to double the bias current and halve the supply voltage.
I suggest you change the Zener diodes to 9V, use a transformer with 2x9VAC secondaries, and increase R10,R9 to 15 ohms. That will get you most of the way there without throwing a wrench into the circuit and messing everything up.
Attachments
Last edited:
Why not wind your own ferrite cored output transformer? You'll probably find it'll make your low impedances phones sound more like 600ohm Beyerdynamics.
I don't know... I've never seen solid state headphone amplifier with transformer. it's usually used to circumvent problems of tubes output.
I refered to higher ouput impedance beyerdynamics in my previous post because they sound quite ok on the Asus Stx. not because I want my headphone to sound like them. I would not exchange my headphones for any Beyerdynamics. I'm a happy owner of a Fostex TH-900 (recabled to XLR4). A sensible high end of closed cans that seems easy to power, but is in fact hard to rightly pair.
rjm said:
After the STX, I used an Audiolab M-Dac for 3 years, which had a very good XLR output, and an ok RCA output (I did serious comparison on outputs, M-Dac should be used on XLR, RCA is plagued by 5/6mv DC). It was quite an upgrade from the STX, Although I did not like its HO. I've sold it recently (~380€) after I upgraded to the Pioneer U-05-S, with a balanced headphones out that pairs very nicely with the Fostex. The Pioneer is maybe a little cold sounding on the Dac output, but not on the HO. I liked the Pioneer so much that I bought a used second unit to leave at work
rjm said:
The STX HO impedance is about 10 ohms... Many electrodynamics headphone doesn't like this. It also doesn't have enough current to control orthodynamics headphones.
---
Just to explain why I would want to do some exotics mods to the Sapphire, here is the background for what seems like a strange idea:
I have tested many dacs, headphones ampfliers, cans, even some of the best commercials systems (other's) money can buy, (but that my money will never be able to afford
).I'm "forced" to used closed cans (at home because of WAF, at work because of co-workers), after testing all the high end usual suspects, I've found that, in my very humble opinion and for my very own tastes, the Fostex TH-900 is my preferred closed cans. So my goal is to build a system tailored to its needs, not a versatile headphone amp. (I had a Violectric HPA-V200 for that, sold it too).
My current planned build is a Soekris dam1021 raw ouput with an opc "the wire" BAL-BAL, input grounded to do the SE to BAL conversion, at unity gain.
But I'm already thinking about what next to try, and would like a low Z-out (< 2 ohms) high current buffer with no voltage gain (because I don't want potentiometer, just the nice ladder resistor of the DAC) and no AOP. So I though that a "stripped down" Sapphire could fit the bill nicely. I asked if it was possible for a switch, but no switch isn't a problem.
What may be the ultimate plans is to use a dual dam1021 in balanced configuration and 4 buffer boards.
DC coupled should not be problematic with a dam1021, especially with no voltage gain. TH-900 maximum output is rated at 3000mW, much more than my ears can handle.
Last little word about the STX. The audiolab M-DAC can display real frequency rate, my STX did 44095/44096hz instead of 44100... even an onboard realtek toslink ouput doesn't deviate more than 1hz from the 44100hz goal. I think than the clock was another fix on the rev2.
Many thanks RJM for your tweaking tips, I will continue to think about it.
The Pioneer U-05? That's in the same ballpark as the TEAC UD-501 I was looking at earlier. Also the Marantz DAC1. I really need to go out and hear some of these.
Thanks for sharing your experiences. In headphone amps it really does seem that electrodynamics vs. orthodynamics is a thing, in terms of component matching. I haven't looked at this, I probably should, with respect to the Sapphire circuit. It is simply made to put x amount of current in y load, with low (>2 ohm) output impedance.
Thanks for sharing your experiences. In headphone amps it really does seem that electrodynamics vs. orthodynamics is a thing, in terms of component matching. I haven't looked at this, I probably should, with respect to the Sapphire circuit. It is simply made to put x amount of current in y load, with low (>2 ohm) output impedance.
I've built one, I'm in the process of building a second one. Some details of both are on my blog. I don't really understand why they're not used with SS electronics as they solve two issues rather nicely. Firstly that the vast majority of SS amps don't have adequate PSRR, you always get to hear the PSU. And secondly they allow easy adaptation to a very wide range of impedances of headphones without the complexities of variable bias and variable supply voltage.
Last edited:
get yourself a line spooler Line Winders / Spoolers to wind the coils. you can get them with counters as well.
I don't see any cost or difficulty in achieving that kind of bandwidth. Admittedly I don't design mine to give full power at 20Hz as I don't think I have any music that demands that, but even if I did it would at worst just double the cost. Seeing as the parts cost is a couple of $ (plus a few hours of DIY time), that's to my way of thinking cheap for the advantages conveyed.
@Bibio - thanks for the suggestion, I've been looking on Taobao for something similar but not quite found anything totally suitable so far.
@Bibio - thanks for the suggestion, I've been looking on Taobao for something similar but not quite found anything totally suitable so far.
So, say, for a 600 ohm : 60 ohm, 3.2:1 turns ratio transformer what are we looking at in terms of bandwidth and size?
I suppose since they don't have to pass DC current (well, for the sake of argument we'll call it < 1mA) so that relaxes the the design parameters somewhat. I'm imagining something like a 5 cm EI frame. Is that about right?
I suppose since they don't have to pass DC current (well, for the sake of argument we'll call it < 1mA) so that relaxes the the design parameters somewhat. I'm imagining something like a 5 cm EI frame. Is that about right?
The size is partly determined by what output impedance your 'phones want to see. Me, I'm happy with having at least 10X lower than the 'phones, actually I aim for about 5% of their impedance but don't worry much unless I get to 10%.
Assuming then your 60ohm phones didn't mind a 3ohm source impedance then the next determinant of the size is the LF extension. I'm happy to handle full power down to 40Hz, at 20Hz I'll be -6dB.
Given those two provisos you can build a trafo around a PQ32/30 ferrite core which is (worst case) 31mm high by 34mm across, 23mm deep. With 5cm to play with you can go all the way to 20Hz at full power I reckon.
Top end bandwidth I've not measured but I'm sure it goes way into the ultrasonics as trafos don't struggle with higher frequencies (especially not ferrite ones which have very low losses), its LF which gives them grief.
Here's a pic (from my blog) of my most recent trafo build - its a PQ32/30 step down 37 : 1+1+1+1 : http://www.diyaudio.com/forums/blog...ing-best-bang-buck-headphone-amp-p1010035.jpg
Assuming then your 60ohm phones didn't mind a 3ohm source impedance then the next determinant of the size is the LF extension. I'm happy to handle full power down to 40Hz, at 20Hz I'll be -6dB.
Given those two provisos you can build a trafo around a PQ32/30 ferrite core which is (worst case) 31mm high by 34mm across, 23mm deep. With 5cm to play with you can go all the way to 20Hz at full power I reckon.
Top end bandwidth I've not measured but I'm sure it goes way into the ultrasonics as trafos don't struggle with higher frequencies (especially not ferrite ones which have very low losses), its LF which gives them grief.
Here's a pic (from my blog) of my most recent trafo build - its a PQ32/30 step down 37 : 1+1+1+1 : http://www.diyaudio.com/forums/blog...ing-best-bang-buck-headphone-amp-p1010035.jpg
Last edited: