Originally Posted by philodox
Then again, I'm a stepped attenuator kind of guy, so take that with a grain of salt.
Either way, looks like a good update to the DA10. Nice work.
So let’s talk about step attenuators. The advantage is clear, you are using passive parts (resistors), and that enables one to set reasonably precise attenuation levels.
Initially it sounds very appealing because there is no active device in the circuit, but that is an illusion. Sooner or later, you will need to tap the signal with some active circuit, be it a headphone amplifier, an input to a power amp, some mixer or what not.
There are 2 problems with the passive attenuator:
The first one I already mentioned – the signal has to pass through mechanical contacts, thus wear and tear becomes an issue.
The second problem is bigger. The output resistance of a passive attenuator changes from position to position (attenuation setting). In fact in most positions the resistance is very high, it can be many KOhms. That is not a good thing. The output of the attenuator is “connected” to the next device by a cable, and a cable with high impedance at both ends is very susceptible to picking up electromagnetic noise (from ac line to radio station to an electric drill next door….). This is a fact of life, and that is one of the reasons that driving cables is best done with low impedance.
One can solve the issue by using a very short cable between the passive attenuator and the next device (which is typically at least a few KOhms). Alternatively, one can put an active circuit to “tap” the high impedance point and drive the cable via low impedance source resistance (impedance conversion). Typical impedance for driving a cable is less then 100 Ohms, not many KOhms. Of course putting an active circuit makes it into an active, not passive attenuator.
Many passive attenuators are over 10KHm to 20KOhms at the middle position. Again, you will end up needing to do such active impedance conversion anyway (the only question is where), and it is best to do it right next to the resistive attenuator tap.
In addition, having a high output resistance (passive attenuator) driving the destination device can really screw up the frequency response. Say you have an active driver with 10KOhms resistance, and a 10 foot cable with say 100pF per foot (pretty typical value). That will limit the bandwidth to around 15KHz! (3dB loss at 15KHz is terrible) In fact, you are already 1dB down at 8KHz!
But with an output resistance of 100 Ohms, the response of the same cable capaitance is flat to better then .001dB at 20KHz, and .02dB loss at 100KHz. In other words, low source resistance is very important to overcome capacitance. Again, short cable helps a passive attenuator. In fact putting that active stage right at the tap of the passive attenuator is ideal, because you eliminate the cable capacity loading effect.
To summarize, driving a cable with high source resistance with an active circuit at the end of a long cable is real bad practice, it makes for susceptibility to external noise, and it screws up the frequency response (loss of highs). The illusion of less active circuitry is just that, an illusion:
One stage of active circuit is very little in the context of the audio signal path (from recording to playback). The mic pre at the recording space, the analog mixer circuits, the AD analog circuits, the DA analog, the power amp or headphone amps and possibly more circuits amount to dozens or more of active circuits. Saving one stage while asking for the trouble I pointed out is not wise…
Well, my DA11 attenuator is in fact a precision resistive passive circuit, just the way you like it (I also like it), followed by internal impedance converter stage to overcome the issues I pointed out – minimize noise susceptibility and have a good and predictable cable independent and load independent flat frequency response…
And while at it, most passives are single ended (unbalanced outputs). My output stage offers both balanced and unbalanced low impedance output drive.