---------------- ASDF_filter_set: ---------------- This file ASDF_filter_set.txt is the counterpart to the schematic drawing ASDF_filter_set.png an improved jumper selectable Sound Card anti-aliasing filter first outlined in the documentation files of my ASDF audio compression system, written last century. That said, the reasons for the use of things outlined in the documentation is still true regardless of the compression type used, or not used as may be the case of dubbing Phonograph records to Audio CD for instance. In both cases you're still facing all of the pit falls of connecting Stereo gear often capacitive coupled at half the line voltage, carrying with it 50% of what ever RF hash is riding on the power line, which your Sound Card will alias against the sample rate, producing chirps whistles whines and the like in the sampled stream of data, yet won't be heard over the loud speakers while recording, but is very present when you play it back. ---------- More Info: ---------- The above type of material is covered in detail in the two files asdf_filters, and Recording_Read-Me.txt, that come with program source code tarball asdf-0.5.tgz -------------------- Direction of Signal: -------------------- Referring to the schematic ASDF_filter_set.png signals generally travel from left (input) to right (output) The pin jack on the far left end is the input to this filter assembly, and usually it is plugged into some "Line-out" signal source, your reciever, a tape deck, or pre-amplifier. The pin jack on the far right hand end of this schematic is the output from this filter assembly, and usually is plugged into the Sound Card, as well as the "GND LUG" bolted to a grounding metal surface of the PCs case. Since ground loops are a significant factor, this design imposes a mecca point signal ground. Rather than running extra heavy wire along the mecca point, you should instead try to flywire all such connections to the same solderable connection if reasonable. The point here is keep wires short, by carefully laying out componets to minimize lead length, remember wires are not only resistive, but at megahertz frequencies their inductive reactance can be several ohms. ------------------------------------------ Protecting the delicate Sound Card inputs: and How this situation ever came about : ------------------------------------------ The mecca extends to the PC cabnet ground. If you want you can ground the stereo to the PC, as an additional measure, but the mecca Ground LUG MUST connect directly to the PC cabnet. Note: do not connect pin jacks to Sound Card input, before hooking up the Grounding LUG, and by extension do not remove the Ground LUG without first disconnecting the pin jack from the Sound Card input. This should be obvious but the Clamp Zeners offer no protection unless the Ground LUG is in place. This is particularly true in an analog differentiator based design, where the Sound Card input employs a virtual ground to avoid ground loop problems, creating problems of their own. This is to protect the input to the sound card! A stereo cap coupled to half the line voltage can permanently damage the sound card if plugged directly in to the stereo's line out. The reason this is true has to do with the evolution of the two worlds, stereos came from Mono-HiFi standards, where reducing electric shock to human beings was more important than saving equipment, and the signal input grids of Vacuum Tubes could withstand tens of volts, almost indefinitely. Then came stereo, and solid state, which were frequently modular, meaning you needed to, as an end user, plug you stereo components together, often with RCA cable plugs. Solid state designs merely set the input impedance high enough to limit input currents, at tens of volts, we're talking 65 volts AC here, to safe levels. Safe that is for the input transistors. Later this protocol was grandfathered in when Integrated Circuit designs came in, but Computers have always been grounded. --------------------------- Some historical background: --------------------------- To find evidence of open loop in that world, you have to go all the way back to the days of Short Wave Radio interfaces to Teletype machines, through what is called a 20 milliampere current loop. The 20 Mill loop was a standard in widespread use in Telephony from about 1915 through to almost modern times, when ESS Electronic Switched Systems came in replacing the old CrossBar, and Stroger Switch based direct dialed phone systems. All phone system technology was closed to the public, at least in the US, until the breakup of AT&T forced manufacturers to "get along" with each other, "by law" "or else" Meanwhile the computer world was openly hostile to the phone system, as was the phone system to the computer world. For a long time connecting the two was flat out illegal! If you wanted to connect your computer to another computer via a pair of modems, if the phone company found out, they could fine you, charge you all sorts of fees, and permanently bar you from ever having a telephone in your name again! You wern't allowed to legally connect anything to phone wires. Oh people did it, folks break various laws today too, and most of the time they get away with it, but once in a while, they make an example of some poor schmuck, to put the Fear Of Police in the majority of all those potential evildoing common folk. Large corporations did need computers connected with modems, some of them leased private lines from the phone company, costing tens of thousands of dollars a month, so they could legally connect two computers together. All of this gave rise to two new industries. The DAA and the DDC basically the phone company would rent you, in much the way they rented you your telephone, a DAA device, that reproduced the phone company voltages and signals, over a pair of beams of light. If you shorted 10,000 volts into such a phone connection, the DAA device would be reduced to a smoldering ruin but the precious phone company equipment would in theory remain unharmed. Other than that the DAA effectively did nothing, it was a technical solution to a political problem. Likewise each computer manufacturer made their own DDC device, which interfaced grounded computer equipment to the phone company's DAA so that computer signals could now go over telephone lines, and do so legally. All of this was allowed to happen because the phones were considered a vital national resource, that needed to be run as a single cohesive system. To that end AT&T was allowed to operate a legal monopoly under US law until the 1970s, when the government had been bringing pressure to divest for some time, partly due to abuses by AT&T, and AT&T had been fighting these moves. But part of the deal that gave AT&T a monopoly, also disallowed AT&T to enter the Data Processing market. Mostly for that reason AT&T suddenly reversed course to get into the Computer market. The rest is history, AT&T was broken up, to form the "Baby Bells" protocols were opened up, suddenly anybody could make a phone, or a device that behaved like one, like modems that connected directly to phone company equipment. But take apart any modem, and look closely at the wiring and circuit foil layout around the RJ11 jacks and you see high isolation componets, transformers, optical couplers, and the like. The mecca Ground LUG that connects directly to the PC cabnet clamps any voltage greater than plus or minus 5.8 Volts, shunting any excess current it carries, via the two pairs of Zener Diodes, off to the computer's own ground, to protect the input to the sound card. There are some clever designs that attempt a differential reconstruction of the Left, and Right, channel voltages with respect to the computer's ground, by actually making the Sound Card's input return (ground) not a real ground at all, but as I said, part of a differential reconstruction of the two signals. Perhaps this is a good idea but almost always leads to a fried Sound Card. You see to make this scheme work, now there is no ground at all, no isolation transformer either, what these designs do, is set the impedance very high, and allow all three input wires, the Left, Right, and what the stereo reciever calls ground, which is actually 65 volts AC, to enter the Sound Card, where it is Analog subtracted to reassemble the tiny Left, and Right signals. Can you guess what happens when any of these three wires come in contact with a chunk of metal carrying a static charge? Poof is the operative phrase here. Another scenario is what happens if you plug the stereo into power while the audio cable is plugged into both the Sound Card, and the stereo. If the divider capacitors on the power mains of the stereo were at a high minus 179 volts when unplugged and the sine wave of the power mains was at the opposite phase when the stereo was plugged back in, for just an instant you deliver up to 358 volts to the inputs of the Sound Card! Poof again. The real weakness here is that using resistors all three wires, for later reconstruction, is far too demanding of an analog differentiator, line-in voltage is typically about one volt, so the differentiator has to accurately reconstruct that one volt signal riding on a potentially 358 volt noise source. So that scheme is also doomed due to the awfull fidelity it would make. And these offerings are the best the engineers of the Sound Cards industry wide are able to bring to market. If this stuff were easy or cheap, it would have been done by now, and so far I'm only talking about catastrophic failure modes, the things that doides in this design solve. --------------------- Designing the filter: --------------------- In this design I placed the coils and surrounding componets, resistors, and capacitors, as a set of starting conditions, as usually one doesn't have as many coils to choose from, tweaking the values of resistors and capacitors makes much more sense. Then with a scope, and a generator, I homed in on a design that was repeatable. I used a Sprague 1:1 toroid core transformer as a coil in the low frequency filter, one each for left, and right, channel respectively. Each winding of the transformer measured 820 uh, and the combined series inductance measured 1.6 mh, so apparently the mutual inductance was insignificant. I mention this in case you have to substitute using a single 1.6 mh coil, in each low frequency filter. The Zener diodes are shown with cathodes connected together, however if it makes a difference, ease of layout for instance, you can turn them around if need be, connecting them anode to anode. All Zener diodes are 400 mw or better, all resistors are 1/4 watt or better, Capacitors are 25 volt or more, monolithic type, or some other equally low ESR rated type. ----------- In closing: ----------- Do read those other two documents, regardless of encoding / compression technique you plan to use, most of what they say is applicable to any Sound Card recording. Also remember you cannot hear aliasing through the monitor speaker, it only shows up in the resultant digitized recording during playback.