There are several different publically-available crossfeed circuits. They all sound different, they have different levels of complexity, and some make demands on the circuits they’re used with. I’ve settled on Chu Moy’s Modified Linkwitz Crossfeed for amps I build due to its low complexity, good sound, and low requirements on the surrounding circuit. I’m happy enough with this circuit that I’ve made PCBs for it. Both the PIMETA v2 and PPA v2 amplifier boards have mounting holes that line up with corresponding holes on the crossfeed boards.
|
A populated Modified Linkwitz Crossfeed PCB (v1.2)
You will notice that two of the four C2 positions are left unpopulated. See below for the reason.
The current version of this circuit board is 1.85" × 1.6". The outer pair of mounting holes are 1.4" apart, and the inner pair are 0.5" apart.
I won’t go into much detail about this circuit, since Chu Moy’s article has all the info you need if you want to tweak the circuit. I do have a version of the schematic here.
The circuit uses four capacitors and 10-12 resistors, depending on whether you want single-level crossfeed or a high and low crossfeed setting.
In Chu Moy’s article, he describes two different sets of part values, one for a low-impedance (“low-Z”) version of the circuit, and one for a high-Z version. I prefer the high-Z version because I think the crossfeed circuit works better between a source and a headphone amplifier. The low-Z version is for use directly ahead of the headphones, a poor place for a crossfeed circuit, in my opinion. A bonus of using the high-Z version of the circuit is that its capacitor values are lower, allowing higher quality capacitor types to be used. If you really wanted try the low-Z version of the circuit, you would need to compromise on capacitor quality, since more compact capacitors generally sound worse. I will use the high-Z values exclusively below.
Not all capacitor lines have 0.12 µF capacitors in them, so C2 is split into 2 board positions in parallel, allowing you to use 0.1 µF and 0.022 µF to get 0.12 µF total. Since C1 is 0.022 µF, you need caps of that value anyway. (Pedants, take note: 0.1 µF plus 0.022 µF is not 0.122 µF, if you’re dealing with tolerances of 2% or wider.)
There are two capacitor footprint sizes on the board. The C2x1 caps are 0.7" × 0.3", or approximately 18 mm × 7.5 mm. The other four caps are 0.5" × 0.2", or 12.5 mm × 5 mm. There is a bit of space around each of these, so you can get caps a smidge larger and be okay, particularly if they aren’t box caps, and thus don’t need to sit right down on the board.
The best sort of caps for this application are polypropylene film-and-foil. The trick is finding ones which are small enough to fit the board and available. Panasonic used to offer the ECQP line, which fit perfectly, but they have discontinued them. Wima’s FKP-3 series looks like it should work, but I don’t know where to get these in hobbyist quantities. The best widely-available polypropylene film-and-foil cap line is the Orange Drop 715P series, but they're large and have an unfortunate lead crimping, so you have to be willing to re-bend the leads and cope with the cap bodies hanging over the board edge.
If you’re willing to step a bit down the quality curve, metallized polypropylene caps are much more widely available. Both Wima’s MKP-10 series and BC’s MKP 416 series will work with this board, for example.
If you want to cheap out a bit or you want to squeeze this board into a tight place, the board will accept polyester box caps with standard 5 mm pin spacing.
The resistors have 0.3" pin spacing, so standard 1⁄4 W and some specialty 1⁄8 W resistors will fit. For the resistor values, see Chu Moy’s article.
The IR, IL, and IG pads are right channel input, left channel input, and input jack(s) ground. I’m sure you can figure out what OR, OL, and OG are for. :)
The remaining pads are for the switches used to control the crossfeed. The switch pad naming scheme is:
|
The first part tells you the switch, the second the pad function, and the third the channel.
The normal switch configuration for the crossfeed uses two DPDT switches. (See below for an alternate arrangement using a rotary switch.) S1 picks the crossfeed level, and S2 is for bypassing or enabling the crossfeed circuit.
If you only want one crossfeed setting, you can leave out the R1As and jumper from each S1C to the corresponding S1L.
If you use a standard DPDT toggle switch, the pad functions will line up so that you can put a “rainbow” of wires between the switch’s solder lugs and the pads on the board. That is, the bottom-most lugs on the switch will be wired to the pads at the edge of the board, the top lugs to the pads farthest from the edge, and the middle lugs to the pads in between. See below for a picture showing this arrangement.
The functions on S1 are High crossfeed, Common, and Low crossfeed. The functions on S2 are Enable, Common, and Bypass.
This being a stereo circuit, there are two channels. It happens that channel 1 is the left and channel 2 the right, but most of the time it isn’t important to know this.
Although the modified Linkwitz circuit was envisioned to use a pair of DPDT toggle switches, Scott Lindeman came up with two arrangements that each let you use a single 4P3T rotary switch. The wiring is a bit more complex in each case and a rotary switch takes more room than a pair of toggles, but you may prefer the look of a crossfeed selector knob to that of a pair of toggle switches. The rotary setup also ensures that you can only pick one of 3 settings. The toggle switch method allows a fourth setting: with the crossfeed bypassed, toggling the high/low switch does change the sound, because the high setting attenuates both channels a bit.
The switches I’ve used from C&K and E-Switch have the “common” solder lugs in the center of the switch, labeled A thru D. Then there’s a ring of 12 other solder lugs labeled 1 thru 12 surrounding the common lugs. Lug A connects to lug 1, 2 or 3 depending on the knob position, B connects to 4-6, etc. Here’s what gets connected to what:
|
Simple Way:
|
Silent Way:
|
The Simple configuration is electrically the same as using two DPDT toggle switches. The only trick is that some of the solder lugs on the rotary switch are connected to their neighboring lugs.
The Silent configuration is a variation on this which eliminates the click you get in the headphones when changing between crossfeed settings. It does this by ensuring that there is always some resistance between input and output; in the other configurations, this connection is momentarily interrupted when changing the crossfeed level, giving an audible click. The switch lugs which are not mentioned in the table are not connected. Also, you must tie S1H1 to S1C1 and tie S1H2 to S1C2 on the crossfeed board and change the R1B resistors to 6 kΩ, assuming you use the high-impedance values elsewhere in the circuit. The low crossfeed setting is then this 6 kΩ resistance in parallel with the 2 kΩ high resistance, which gives a 1.5 kΩ overall resistance.
The mounting holes on the crossfeed board are spaced to line up in two different ways with corresponding holes in the PPA and PIMETA v2 amp boards. Don't expect alignment with three or more of the holes.
With the PIMETA board, there isn't much space around the screws, so be careful about the diameter of the standoffs you use. Nylon spacers are probably too bulky to use, for example, which is why I include two 3⁄4" Keystone #4 aluminum spacers with the board. (Mouser stock #534-418) You may want to get a different length, depending on your space requirements. For instance, if you use PCM-5 box caps, you can probably mount the board upside down over your amp board without collisions and still get away with a 1⁄2" spacer, making for a very compact arrangement.
Also pay attention to the size of the screw head. I find that typical round-head #4-40 screws stay inside the design bounds while securing the board securely.
|
| Crossfeed mounted on PIMETA v1 board |
The PPA has more space around the crossfeed mounting holes, so you don't have to be as careful about standoffs and screw head sizes as with the PIMETA.
I first build the amp without crossfeed, and I make the wires from the input jacks to the board a little long. Once the amp is working well, I cut the input wires roughly in half. Then I hook the wire halves going to the input jacks to the crossfeed board’s “I” pads, and the remaining wires to the crossfeed’s “O” pads. If the amp doesn’t work right after this surgery, you know the amp itself is working, so the problem must be on the crossfeed board.
Depending on the enclosure setup, I sometimes don’t do the toggle switch wiring until after I have the board mounted. This way, I can cut the wires exactly to the right length. If you wire the toggles to the board before you mount the crossfeed, you can end up with hookup wires that are too short or too long.
Some things about the circuit’s behavior that you might want to know:
The Modified Linkwitz crossfeed circuit attenuates the signal going through it by roughly 6-10 dB, depending on the source you use. (The variance is probably due to different source output impedances.) You need to set the amp’s gain 2 to 3× higher than you would if there were no crossfeed in the circuit to counteract this attenuation. Try 2× first, and only go up to 3× if the amp still won’t drive the headphones loudly enough in your setup.
The term “bypass” is misleading, since the crossfeed isn’t completely bypassed. If it were, the amp’s output voltage would jump by 2-3× when you flip the switch into the bypass setting. Instead, what the bypass setting does is just makes the impedance between the channels much higher than the path that connects the each channel’s input to the output. Therefore, the signal “prefers” the direct path from input to output. There is still a tiny bit of channel mixing, but it’s a necessary compromise to avoid the volume jump.
I offer the board plus standoffs in my Parts Shop.
Sorry, I do not offer a kit for this any longer.
| Updated Thu May 26 2011 19:21 MDT | Go back to Audiologica | Go to my home page |