Many "high end" speakers have special connections so that their woofers and tweeters can be connected to dedicated amplifiers. These speakers are called "biamp" capable, meaning that two stereo amplifiers can be used, one for the woofers and one for the tweeters, or one amplifier for one speaker and one amplifier for the other. Such speakers have two sets of inputs on the back, and these can be shorted together for use with only one amplifier. The problem is that for biamplifying to work the amplification (or "gain") of each stereo amplifier must be identical. Problem- there is no standard for gain and most amplifiers don't specify their gain or even have a gain control. Moreover, most preamplifiers have only a single set of outputs so cannot drive two amplifiers without special connectors or special circuitry. To address this problem, this project is a buffer that takes an input line level signal from a preamplifer, and passes it into dual stereo outputs-- the kicker is that one set of stereo outputs (the higher level one) has a gain/attenuation control so that it may be matched exactly with the other set of outputs.
Thus this project allows for biamping speakers (that are biamp capable- having two sets of inputs) using any pair of amplifiers driven by any preamplifier. The preamplifier may be single ended or differential/balanced, and each amplifier may itself be single ended or differential/ balanced. Got a balanced pre and singel ended amps? No problem! Got a single ended pre and 1 balanced amp and 1 single ended amp? No problem!
This buffer uses two cermet trimpots per channel for attenuation, they anc be switched out of the signal path in favor of a selected high quality resistor. I chose to buffer using dual opamps in the circuit since there are many varieties of opamps avialable and because if you feed a differential singal into a dual opamp, the common mode noise of the opamp will cancel out at the opamp outputs.
The topology is an "Input Buffer" followed by its output connected to two stages- "Fixed" and "Variable." The "fixed" output goes to another buffer and to the differential outputs. The "Variable" output goes to a multiturn cermet trimpot and then is buffered again before feeding the differential output.I built this into a chassis from a discarded piece of HP test equipment. I wanted to keep the original look and even as much of the original wiring/parts as I could. I used "ultra-low" noise voltage regulators from Linear Technologies in the power supply. I even built my own printed circuit boards (PCBs).
UPDATE (May, 2009)- I replaced the on/off switch and the neon lamp, both of which failed. Though careful searching I found Mouser part #36NE004 appeared to be a direct lamp replacement. Its data sheet indicates a 30K series resistor should be used and the HP case already includes a 31.6K resistor. Also the "Indicon" rocker switches from CW Industries (Mouser #629-AA211B11) fit exactly, they slide right into the on/off cutout.
Step 1: Circuit and PCB Board Design
Notes: The circuit gives you a lot of options. Switch #1 selects between a balanced/ differential input and a signle ended input. Either way the outpt is differential because if the SE input is selected the signal goes through DRV134 which splits it into a differential output.
The concept behind Switch #2 is that you adjust the gain of the 2nd output to match the first and then "probe" the board to measure the serices resistance of the trimpot. You then substitute a fixed resistor and flip the switch to remove the cermet trimpot from the signal path and instead use the high quality fixed resistor.
OK you "purists" out there will note that there are 2 switches in the signal path. These are very good NKK switches that I have used before with great results. However, note that there is a "unbuffered" output on the rear panel which feeds the input directly to the output with just silver/teflon wire, no switches.
Step 2: Having the Boards Made
Notes: Here are the boards, looking pretty good. I used Eagle PCB software and had the boards made by CustomPCB.com. The signal boards are double sided and the PS Board is an old one that is single sided.
Step 3: Making the Front Panel
Notes: Got a couple of HP chassis' cheap from the surplus store. I like the look but wanted to remove some background and label silk screening from the many green LEDs as the labels are not relevant. So I tried sugar cubes. You use them like sandpaper, but keep them dry! I found that they are just abrasive enough to wear off the silk screening but not abrasive enough to scratch the underlying paint. Be careful! Sugar cubes are abrasive enough to actually wear away the paint and you see that I wore the paint down to the metal in a few places. Whoopsie! Sorry about that, but its good enough for me! Better luck with the 2nd one?
Step 4: Creating the Rear Panel
(screenshot of panel and selection of countersunk hole)
Step 5: Assembling And Testing
Notes: This works great, the variable vol is easily adjusted, the output is clean and noise is very low. DC offset can be barely measured. There are exactly 7mA going through each LED. The LEDs are fed from +/- VDC bridge rectified, +/- VDC after the CRC filter, and +/- VDC regulated, using trimpots to set the current (and brightness) accurately.