Copyright © 2008 by Rick Chinn. All Rights Reserved.
A passive preamp is really a misnomer since it is just an attenuator connected in series with the audio signal on its way to a power amp. This is not a lot different than the level control found on the front panel of many power amplifiers. In fact, aside from physical location, it is the same thing.
There's not a lot to it, and (as usual), the devil is in the details. The basic recipe:
If you use a common-ordinary potentiometer, section to section tracking is a potshot at best. The implication here is if the sections aren't closely matched, then a mono signal (or something panned center) drifts in the stereo field as you change the setting of the control. You can buy closely matched ganged potentiometers from Penny & Giles or from Shallco but they are expensive; on the order of $75/section or so for the P&G parts.
The value of the potentiometer matters. It affects how much cable can come between the pot and the amplifier. Higher values reduce the allowable cable length, lower values become harder for the upstream gear to drive. A potentiometer is nothing more than a voltage divider whose output impedance is the parallel combination of the two resistors making up the divider. The worst-case occurs at 6-dB attenuation, where the two halves of the potentiometer (pot) are equal in value. At this point, the output impedance is highest: it's the pot value divided by 4. The output impedance works with the cable capacitance, making a first-order lowpass filter.
The cable becomes an issue because of its inherent capacitance. Most audio cables are 50pF/foot, and a 10k pot can drive 318pf for the frequency response to be -0.1dB @ 20kHz. That translates to 6.37 feet of cable. But wait... that's just the cable, and it doesn't include the input capacitance of the amplifier, which gets subtracted from the 318pf first, and then the remainder divided by 50pF to get feet. If the amplifier has several hundred pF of input capacitance, then there isn't much left for the cable capacitance. Yes, you can use low-capacitance cable.
A 5k pot is a better deal because at 6dB attenuation it's 1.25k output impedance means it can drive 637pf of cable/input capacitance for the same frequency response. 5k is still a reasonable load to expect pro gear to drive. But 5k is a harder value to find, especially with an audio taper, and needing 6 gangs (sections) probably takes the level of difficulty to the 6th power or more.
You can make a stepped pot using a rotary switch. The advantage is that you determine the steps, and the section to section matching is *very* good. The disadvantage is it's a fair amount of work, and the six-gang switch will be spendy, though probably less than a precision six-gang potentiometer. The other difficulty is the number of positions; 22 is probably a workable minimum. Step sizes can be 1.5dB, which is just barely audible. It's a nice value for a mixer pot. For a monitor level control, perhaps 2-3 dB would work out. Typically, these controls start out with small steps for the first 20dB, then the steps get bigger for the next 10, and they get really big, and the last step is a step off of a bridge. I have a computer program for computing the resistors. It is available (not for Mac).
As far as I know, nobody makes a six-gang commercial-grade rotary potentiometer. That is, one available off the shelf, and at a price that won't cause your wallet to involuntarily contract. Four seems to be the magic number here. Close tracking between sections is hit-and-miss. You could probably have Alps (Japan) make something, but the minimum order would be several thousand (THOUSAND) pieces, but they will make exactly what you want. "Off the shelf" is not a part of their vocabulary.
Shallco still makes rotary attenuators, and they'll make whatever you want to order, in whatever resistance you want. It won't be cheap, and there will be leadtime. They also make precision switches, so they're an option for the DIY route. Do sit down when talking with them. http://www.shallco.com/attenuators.html
If you make your own attenuators, the resistors can actually be 5% carbon film units. The value is close enough, these units are pretty precise from the start, and they're cheap. If you go thru the math, you'll see that additional precision doesn't really buy you anything in terms of accuracy of attenutation. The really critical resistor is the last one (highest value of attenuation) in the chain. Sure, you can buy 1% Dale metal-film parts, which will cost more. Sound better? I dunno. I think this stuff is down below the level of picking flyshit out of black pepper.
For switches, there are only a few possibilities. Grayhill, Electroswitch, Shallco, and Elma. You want gold contacts for this. I think I would also add a 10dB toggle switch too, to reduce the level by 10dB all at once. This can be used as a DIM switch for when the phone rings, or when you need to monitor quietly since it changes where you are setting the control (and the stepsize associated with that position).
Are commercial units overpriced? I think so. Considering that none that I know of use a P&G rotary pot for their control, I think they're charging a lot for inflated puffery and audiophoolery. But then, that's what I think of the entire high-end hifi business.
A schematic diagram will appear here eventually. No time right now.