Originally Posted by
LairdWilliams
Ok - time for a lesson on line-level audio interconnects, which is what you create when you connect an audio output from one piece of gear to the audio input on another.
Speakers are not line level...and the "goals" for a speaker connection to an amp are VERY different from the goals for connecting two line-level devices.
People get confused between the rules for speakers and for interconnects...that's ok.
I apologize in advance if this is too much detail - so I will save you some trouble. It is VERY hard to have too HIGH an input impedance. If you don't want any more information, then stop here and you should be all set. Otherwise - some basic concepts may help you in the future:
Ok -so most of this is about REALLY basic electronics. You need to think voltage dividers and RC-filters.
Voltage Dividers
When you string two impedances (or resistances) in a row and take a signal from in-between, you have a voltage divider. A voltage divider is just an attenuator - e.g. it will drop the voltage by a certain amount. The new voltage is a simple function of both the first and second impedance. If the first impedance is I1 and the second is I2, then Vnew = Vold * I2 / (I1 + I2). In the case of an audio interconnect, I1 is the output impedance of the sending device, and I2 is the input impedance of the receiving device.
So look at this function for a sec. What is I2 is really big and I1 is tiny? The I2/(I1+I2) would be really close to 1. REALLY close to "no attenuation". But if I1 were really big and I2 were small, then the new voltage would be a very small fraction of the original voltage, and you get "tons" of attenuation. If I1 and I2 are exactly equal then you get exactly 1/2 the voltage out that you put in, which is exactly -6dB by the way.
There is more to it than this, since the impedances actually vary with frequency. But as it turns out - the manner in which frequency impacts the whole process makes the following MORE true, and not less:
If you have a low output impedance feeding a nice, high input impedance then you get very little attenuation (at ANY frequency), which is what you want. This is because the voltage divider that you create by connecting the equipment has a ratio of darn close to 1.
Laird-speak: "
Low output impedance good, high input impedance good"
in context
Soooooo - this is why I said that connecting the 100-ohm output impedance of the 105 to a 23.5k input impedance on an amp should not have any audible problems stemming from the resulting voltage divider. The attenuation due to the voltage divider in this case is 0.42% ,which is about -0.04dB...which is essentially nothing.
RC Filters
And R-C filter is just a passive filter created by joining an impedance (or resistance) and a capacitance. RC networks get used all the time for crossovers. You can create both high-pass and low-pass filters using one resistance and one capacitance, just depending on the order in which you string them together.
Why does this matter? Well, your shielded cable has a capacitance set up between the signal wire and the (grounded) shield. The amount of capacitance is a function of the materials used in the cable and its length, with longer cables having higher capacitances than shorter ones. So now you have an output impedance followed by a "cap-to-ground", which is the textbook picture of a low-pass filter. A low-pass filter blocks higher frequencies and permits lower frequencies to "pass".
It is again a simple math function to determine the frequency at which the highs start to roll off. This is the 3dB-down point, and it is computed easily as f = 1/(2*pi)*R*C, where pi is everyone's favorite oddball mathematical constant.
[In Context]
Ok - so let's turn the question around though. "What capacitance would I need, with the Oppo BDP-105's 100 ohm output impedance, to result in roll off starting at the theoretical human hearing max frequency of 22000 Hz?" (Never mind that the vast majority of folks over 20 years old can't hear much above 16000Hz, and that as you get older it gets worse, and that if you play in a rock band you are darn lucky if you can hear past 12000Hz.)
Anyway - this results in a value of 0.072uF (micro-farads). Wow! That's not much right?! sorry - wrong. "Normal" cables run about 120pF per Meter (that's PICO-Farads, which are 1000 times smaller than NANO-Farads, which are STILL 1000 times smaller than micro-farads). Some higher-priced cables do better.
So we are talking about 72,000pF * 1meter/120pF = around 600 meters of cable.
So you'd have to have nearly 2000ft of audio cable running between your 105 and an amp, and some VERY unrealistically good ears, to even HOPE to hear the tone roll-off resulting from the RC filter.
If you have a nice low output impedance, you pretty much never have to worry about tone roll-off due to the RC filter created by your output impedance and the capacitance of your cable.
Laird-speak: "
Low output impedance good (again)!"
Loads
So now we get to the last major factor impacting the electrical signal that crosses your audio interconnects, loads. This one is more complicated, so I will avoid any mathematical treatment. Your downstream device can actually rob your upstream device of power. If you have a small amplifier circuit connected to your output (which is
usually the case), then the receiving device can "draw" so hard on the signal that it robs your output device of power - like if the sending device can't provide enough power for the receiving device. The tonal effects of this problem can be
really pronounced. I run into this problem all the time with guitar amps that have unbuffered FX loops and passive loop level controls.
There are several ways that a receiving device can "load" a sending device....but one is that there is a path to ground through the input impedance at the receiving end. Losing current here is bad news, as that can contribute substantially to the load on the sending end. So, guess what you need. I high input impedance means that LESS current will travel to ground than it would with a low input impedance. High input impedance wins again!
Laird-speak: "
High input impedance good (again)!"
That's 2 votes for having a high input impedance to zero for having a low input impedance for highest fidelity.
That's 2 votes for having a low output impedance to zero for having a high output impedance for highest fidelity.
I'd say we know who the winners are. Oppo can't control the input impedance of the device to which the 105 is connected, nor can they control the quality or length of the cable you choose to use. What they CAN control is output impedance - and they did so very well. In doing so, they made the cable largely-irrelevant (other than mechanical considerations). They also, apparently, provided a recommendation for load impedance. If they recommend a 47K input impedance for a load, just about anything at or over 47K will do.
I would be worried about anything much lower than that.
(For reference, In the guitar world, we routinely deal with input impedances of 100k-ohms to 1,000k-ohms once we reach the point where we want to preserve audio fidelity.)