Sorry that my post was a bit rambly, and without a simple answer, but glad you got something which works for you. Thanks to everyone who defended my ramblings. Although there is no "one size fits all" answer, with hindsight, I could have given better guidance.Elopid wrote:Wasn't quite the outcome I was expecting from a simple question lol.
In terms of source / input impedance, if you're connecting the filter directly after a modern, line level source, it should be low enough to have little influence. My concern here, was that you might be connecting after a level control, which if it was a relatively high level pot, could have significant influence. Regarding output impedance, my main concern was that I have seen quite a lot of driver boards, which have relatively high input impedance by design. On the other hand, if you found that pre-emphasis was noticeably lacking, for that reason, you could always add a resistor to ground; between the RC filter and the input of the driver board. I should perhaps have mentioned that. This will lower the level, but that's the point. It's passive, it has no gain, but output needs to rise by almost 14dB by 15KHz. This apparent gain can only be achieved through loss, which is achieved from a potential divider effect, defined by the ratio of R to RL (where RL represents the load impedance - assumed to be resistive, and still assuming RS to be insignificant).
The RC values which Mr Woodstarr gave, need to work into load impedance of about 1k. BW used these values, and fixed it at about 1k with resistors. Using RC values of 12k and 4.7nf gives a TC of 60us (instead of the textbook 50us), so response will start to rise a little early, but it stays pretty close to the 50us curve (difference between 60us and 50us is pretty small anyway). There is a second Fc (which defines the top end roll off), which comes from the time constant of RL*C. With RL at 1k, and C at 4.7nF, this is 4.7us; giving an upper Fc around 34KHz (we've run out of apparent gain, and the rise has almost flattened out by this point). You can look at this as 60us high boost, and 4.7us low cut - working together, and interaction of the two will make both slightly shallower than textbook first order filters. Response has started to roll off very slightly by the time we get to 15KHz, but as we started off slightly above the textbook line (due to the initial 60us eq), we're still in good shape (still assuming negligible source impedance), and the high frequency roll off - albeit shallow, falls in a pretty good place. Using 33k with 1.5nF into 1k load will also give good pre-emphasis, but level will be lower, and the high frequency roll off won't start as early - slightly different balance of compromises.
