Originally Posted by Madebynick
Many thanks for the reply sir. You have set a few things straight in my head which is a big help. On your advice, I took delivery of some 75ohm resistors and will integrate them into the new cable I'm building. I do think I need to do some reading around this subject however as there are many terms thrown around which are still somewhat confusing to me!
Perhaps you could clear up a couple of other things for me too if thats ok?
You mention that the new TI chip you're using in the SkeletonDAC can now be used as a headphone amp and as I understand it, the headphones would simply be connected to the same output one would use when connecting the DAC to an amp (assuming standard setup with flat frequency response). With this in mind, what makes the new TI chip more suitable for driving headphones from this output compared to the original chip? If the answer is outside the scope of this forum, please tell me. I won't mind.
Secondly, and this is probably basic knowledge that I should have already accrued before I venture to these forums, is to do with those frequency response curves you posted. It's taken a lot of web reading to try to explain to myself exactly what these graphs mean and now I think I understand. Please feel free to send a rolling eye emoji and point me towards electrical engineering nursery school if I'm wrong.
If I'm correct, those graphs are showing (very simply put) the volume of the frequency range indicated by the graph. These curves are the direct result of altering the combination of resistor and capacitor values and suggest that with a larger capacitance value, the volume of lower frequencies will be higher and therefore sound output will appear more bassy? (Apologies if I sound like a bass head here. I don't mean to!) What makes me think that I may be wrong however is the shape of the frequency curve around 1300Hz. Is this curvature due to a filter built into the circuits design that ensure higher frequencies cannot be heard? I may guess that the effect of increasing the load applied to the DAC as a result of the 75ohm resistor is the controlling factor dictating where this curve plateaus? Am I correct in thinking that if this flat response was no seen at 1300Hz and above, that the higher frequencies would be clearly heard as a background hiss as I have experienced in the past?
I do hope these questions won't cement my reputation for being an audio electronic novice to the point where my head will be permanently hung in shame.
Thanks in advance,
1. The PCM2704/5/6/7/ chips all have a built-in headphone amplifier. The chip that it replaced was the PCM2702. It had no amplifier and the output was taken directly from the Digital-Analog converted output. In the case of the PCM2702, it was a voltage output, but very small with practically no current output. IOW, you could actually damage the PCM2702 chip if you attempted to feed headphones directly from the DAC. With the built-in headphone amplifier of the PCM2704/5 used in the SkeletonDAC, you can actually hook headphones up and generally speaking - not worry about blowing up the DAC chip.
All that said, the quality in driving ability of the DAC's amp is not super. So, the types of headphones, their efficiency, and the resulting high-pass filter circuit resulting from your parts choice will dictate how well it drives a pair of headphones.
2. You have this backwards, actually. The "0 dB" line on those graphs (the X-axis) is the point of normal volume, all things being equal. So, what the curves are telling you is how much the bass is decreased by the resulting RC-circuit high-pass filter. The RC-circuit high-pass filter is a natural phenomenon of the electronic circuits and loads in play. You have no choice in the matter on whether it will occur - only that you have a choice in selecting parts that will minimize the bass loss. With low-impedance phones, a very large value capacitorr (470uf to 1000uf) will result in almost no bass reduction at all. However, the caps will be very, very big (in most instances). So, what I'm trying to portray is that you can come up with a compromise between bass lost and inconvenient capacitor size. That's all.
I'm not sure what you mean by the 1300 Hz. I suppose if you looked with a magnifying glass at the first graph up there that's based on 47uf output capacitors, one might say that the bass attenuation begins at about 1300 Hz with a 16 ohm impedence headphone load - and bass attenuation begins at about 1100Hz with 32 ohm headphones, 1000 Hz with 60 ohm headphones and so on and so forth. But - that's all you can say about it. The whole point is that is the theoretical result with 47uf output coupling caps. As you can see in the second graph, the frequencies at which bass begins to be attenuated is quite a bit lower with 220uf output coupling caps.
The 75 ohm resistors were suggested to raise the "R" value in the fundamental relationship of the RC circuit. R is multiplied times C (and also modified by other factors). So, if we increase R or increase C - the result is that bass attenuation begins at a much lower frequency. The difference is that big resistors are much cheaper and easier to obtain and install in the circuit than big capacitors. At some point, higher value resistors will have a noticeable impact on the audio sound themselves, but if given a choice between increasing impedance at 16 ohms or selecting some outrageously large and expensive capacitors, the 75 ohm resistor is a no-brainer. This has a long history of precedence, because that's exactly what a 75 ohm adapter is intended to do with IEMs and small PDP's. Small PDP's cannot afford to include physically large output capacitors, so the resistor adapter was developed, instead (R times C exists with PDP's and their headphone connections). Output coupling capacitors may not exist with all PDP's, but usually a line-out adapter bypasses everything that was intended to cut down on the DC offset. So, to protect your headphones, DIY line-out adapters usually involve output coupling capacitors of some type. This would also occur in the SkeletonDAC (and the Bantam/Alien before it): if output coupling capacitors were not used, your headphones would see 2.5V on output at zero volume. That's why the capacitors are used, but the physics of the resulting electrical circuit has some drawbacks - mainly that an RC circuit is formed with the "R" of the headphones. This means that if you are not careful in parts selection with the capacitors (or some resistor adapter), then you will lose bass in the audible frequency band (what the curves in the graphs above show).
EDIT/PS - After posting the first response above, I realized that one neat solution is to simply use the 560uf organic polymer caps - used in the GrubDAC and SkeletonDAC as power caps - as the output coupling caps on the Skeleton DAC [for direct-connected headphones]. I haven't tested how they sound, but others seem to think that OsCons (another organic polymer type of capacitor) sounded pretty good in the signal path. The nice thing is that those 560 uf's are very little bigger than the pads already on the PCB - and height is not an issue, even with the little Hammond plastic case (1551HTBU).
Edited by tomb - 11/25/11 at 4:37pm