[Solrighal]

I can't remember what it was that's supposed to be difficult with the AKG's. I'll be able to let you know soon enough

 

[Solrighal]

Does the O2 benefit headphones such as the V-Moda M-100? I wouldn't have thought so myself but what do I know?

 

[miceblue]

Does the O2 benefit headphones such as the V-Moda M-100? I wouldn't have thought so myself but what do I know?

Whatever sound the O2 has, it benefits the M-100 from my experience; the soundstage opens up a bit. I don't use the O2 with the M-100 though since it's too loud for me, but if you lower the digital volume it should be fine.

 

[Solrighal]

Nah, I doubt I'll use them much. I just wondered since, although my phone can push the M's loud enough they do benefit from the kick in the backside the E18 gives them. Same volume but more drive and attack and everything just feels bigger.

 

[SilverEars]

what about built in sound card?  You have no choice but to use the software volume control.  Is it software or is it software controlling analog like the Dragonfly?

 

[agdr]

I can't remember what it was that's supposed to be difficult with the AKG's. I'll be able to let you know soon enough

It is the sensitivity, which is the other thing along with impedance that figures into how hard a set of headphones is to drive.
 
The current AKG701's have a sensitivity of 105dB/V from AKG's specification sheet.  That means with 1V(rms) into the headphones you will get a sound pressure level (SPL) of 105dB. 
 
A rule of thumb for headphone amps is to be able to hit 120dB, which is just as ear-splittingly loud as most folks would ever want to get headphones.  The O2 headamp can muster 7V(rms) of output swing, which would give:
 
SPL = 105dB(SPL)/V + 20log[7V(rms)] = 121dB, which hits the mark.
 
BUT... then you have to also check the current draw at that maximum voltage swing and make sure the amp can sink/source that, too. 
 
At 7V(rms) into 62 ohms we get:
 
Iout(rms) = 7V(rms) / 62 = 113mA, and that is per channel. 
 
The output chips in the O2 can sink/source around 120mA per channel maximum, but that is if you have a transformer that is big enough to supply that much current to the output chips in the first place.  As I've posted earlier in this thread the half wave voltage doubler power supply in the O2 results in large current spikes through the transformer when the diodes conduct, such that the transformer secondary needs to be rated at about 3 times as large as the DC (rms) value being supplied.  For the details see the second picture down on the left here, half wave capacitor input power supply.  Idc = Irms = .28 * secondary rating, or use a secondary rated at 1/.28 = 3.5x the Irms value:
 
http://www.hammondmfg.com/pdf/5c007.pdf (opens PDF)
 
So for 113mA x 2channels, plus 25mA of quiescent current in the O2 chips, plus 25mA more for the O2's internal battery charging, that gives a current draw on AC of:
 
I(max, rms) = 113mA + 113mA + 25mA + 25mA = 276mA.
 
Your transformer secondary would then need to be rated at least 3x that, or 3 * 276mA = 828mA, which means round up to the next standard value of 1A.  The Triad WAU16-1000 (Mouser #553-WAU16-1000 for $11) would be perfect.
 
For a lot of high sensitivity headphones, like my AKG-K550's at 115dB/V, the output voltage level needed is so low [about 60mV(rms) = 0.06V(rms) at nromal listening levels]  that the smaller 400mA-rated transformer secondaries are just fine.  But the 701's are one of those harder-to-drive lower sensitivity, lower impedance, headphones that need the larger voltage swing and the associated higher current drive.
 
What happens if you use the smaller 400mA transformer secondary with the 701's?  After all, the maximum DC current draw we calculated was 276mA total, which would seem to be within the 400mA rating of the smaller WAU16-400 transformer, at least ignoring that half-wave rectifier current spike behavior for now. Well first off that 113mA per channel was all the way up at 120dB.  For most listening you might only be consuming 70mA or 80mA per channel or so, peak.   Then when you figure in "music power", where the peak levels of music are 3x or so the average, that may drop the average current draw down even further to 40-50mA per channel.
 
Then what happens if the transformer rating is too small is the iron core saturates, which can create distortion in the voltage and current output, and the transformer heats up more than normal.  Whether that AC power input distortion makes it to the O2 amp output would depend on a lot of factors with the regulator chips and the op amp PSRR.  You might never hear it.  But for $3 more or so between the 400mA WAU16-400 and 1A WAU16-1000, might as well get the 1A secondary if you are ordering that O2 transformer for the first time and have a harder to drive pair like the 701.

 

 

[r010159]

  It is the sensitivity, which is the other thing along with impedance that figures into how hard a set of headphones is to drive.
 
The current AKG701's have a sensitivity of 105dB/V from AKG's specification sheet.  That means with 1V(rms) into the headphones you will get a sound pressure level (SPL) of 105dB. 
 
A rule of thumb for headphone amps is to be able to hit 120dB, which is just as ear-splittingly loud as most folks would ever want to get headphones.  The O2 headamp can muster 7V(rms) of output swing, which would give:
 
SPL = 105dB(SPL)/V + 20log[7V(rms)] = 121dB, which hits the mark.
 
BUT... then you have to check the current draw at that voltage swing and make sure the amp can sink/source that, too. 
 
[deleted part of an excellent analysis]
 
Then what happens if the transformer rating is too small is the iron core saturates, which can create distortion in the voltage and current output, and the transformer heats up more than normal.  Whether that AC power input distortion makes it to the O2 amp output would depend on a lot of factors with the regulator chips and the op amp PSRR.  You might never hear it.  But for $3 more or so between the 400mA WAU16-400 and 1A WAU16-1000, might as well get the 1A secondary if you are ordering that O2 transformer for the first time and have a harder to drive pair like the 701.

 

 
This reminds me of the test problems in one of my EE courses that I took and forgot eons ago. This is a very nice analysis. I have a question. I purchased my O2/ODAC already preassembled. How difficult would it be to replace the existing transformer with the one that you mentioned? A simple re-soldering job? This may become a nice side project for me to get me out of the doldrums.
 
EDIT: Let me ask a stupid question. How simply can the amp be redesigned to moderate the current spikes that develop when the diodes start to conduct? I am a little embarrassed to say what I am thinking is the solution.
 
BG

 

[Solrighal]

I feel like I've stumbled into some kind of alternate reality. I now know exactly why I'm driving a van for a living. This is way over my head guys. I've re-read that twice now and I still don't know if my new amp and headphones are gonna work out.

 

[r010159]

I feel like I've stumbled into some kind of alternate reality. I now know exactly why I'm driving a van for a living. This is way over my head guys. I've re-read that twice now and I still don't know if my new amp and headphones are gonna work out.

 
Sorry. Let this commercial break end for the original program to resume. 

 
If I understand the tech part well enough, the short of it is the O2 should be able to power the AKGs. But the O2 is not in the optimal configuration for those phones. There is room for improvement. I will let the original author elaborate on this.
 
BG

 

[jring]

   
How difficult would it be to replace the existing transformer with the one that you mentioned? A simple re-soldering job?

 
Literally plug'n'play - you just get the 1A wall wart, plug that and are ready to go.
 
Joachim

 

[SilverEars]

   
Hi,
 
when 16 bit audio data is mapped to 24 bit words this means multiplying by 256 or shifting left by 8, which is the same. The missing bits on the right hand side are set to 0.
 
EDIT: This multiplication is necessary in order to keep the volume equal for 16 bit and 24 bit mode. I one just kept the same numeric value in 24 bit as in 16 bit, the music would be inaudibly quiet. Thanks to user r010159 for pointing out that I was a bit quick here.
 
 
16 bit word:
 
              1001100110011001
 
24 bit word - high 16 bits same as before followed by 8 zero bits:
 
100110011001100100000000
 
If you're then halving the volume all words are divided by 2 or shifted to the right by 1 - left side is padded with zero bits again:
 
010011001100110010000000
 
So what you loose is one of the zero bits on the right hand side which were added by converting 16 to 24 bit.
 
 
In the case of native 24 bit material the case is a little bit different but not much:
 
100110011001100110011001
 
now we half the volume by dividing by 2 or shifting right 1 place - left side is padded with zero bits again:
 
010011001100110011001100
 
So we effectively lost the least significant bit on the right hand side which can be ignored since
 
 
a) the effective resolution of the best 24 bit DACs is only 21 bits and often less - so an error in the least significant bit will not be measurable
 
b) because human hearing can only resolve 16 bits of dynamic range anyways
 
EDIT: Please note that this argument is a bit simplified by assuming unsigned integers. In reality we need signed numbers and in 2's complement notation multiplication and left shifts are not equivalent. But in essence it's the same in 2's complement - namely for data scaled from 16 to 24 bits no data is lost by setting software volume to 50% and for 24 bit data the least significant bit is lost.
 
Joachim

You sound like you know a bit about digital volume levels.  

  Can you explain how it ties into percentage of the software volume levels on the computer?  

 

[Solrighal]

Literally plug'n'play - you just get the 1A wall wart, plug that and are ready to go.

Joachim

Are you seriously saying that all that ^up there^ boils down to using a different plug?

 

[Xenophon]

FWIW:  I use the Epiphany acoustics O2/DAC to drive my HE-500's in my office setup.  These are lower sensitivity (about 89 dB) 38 Ohm cans.  Not ideal with the O2, it simply lacks oomph for these.  Tried changing the stock wall wart to a higher grade one but to my ears it makes no difference.  I'm ok listening at low/moderate volume to classical but if I dial in gain and crank up the volume quality suffers big time.  I applaud agdr's explanation and design efforts but if I were him and with his obvious technical knowledge:  build something new from the ground up and market it.  I'm not big on incremental development to adjust a design, better in general to start from a white page imo.

 

[Solrighal]

FWIW:  I use the Epiphany acoustics O2/DAC to drive my HE-500's in my office setup.  These are lower sensitivity (about 89 dB) 38 Ohm cans.  Not ideal with the O2, it simply lacks oomph for these.  Tried changing the stock wall wart to a higher grade one but to my ears it makes no difference.  I'm ok listening at low/moderate volume to classical but if I dial in gain and crank up the volume quality suffers big time.  I applaud agdr's explanation and design efforts but if I were him and with his obvious technical knowledge:  build something new from the ground up and market it.  I'm not big on incremental development to adjust a design, better in general to start from a white page imo.

I would have the thought the O2 could deal with those easily.

 

[Xenophon]

I would have the thought the O2 could deal with those easily.

Depends on how you define 'deal with' I guess.  Don't want to start a long argument but if I compare the sound quality (not 'volume') out of the O2 to what I get connecting them to my 2.7 W into 50 Ohm V200 or my 1.5 WPC tube amp then there's a clear difference.  It's ok at low gain/lowish volume though.  And the comparison is of course not 100% honest from a price point view (V200/V800 setup is >2k USD) and comparing a tube amp to the O2 more or less negates what the O2 stands for.