This article came about through a curious combination of circumstances. I was reviewing the Peavey VMP-2 mic preamp (7/98), an excellent design. In the course of the review I compared the Peavey to a preamp section from a popular low-cost mixing board and to the solid-state reference preamp I always use for microphone reviews…and a lot of recording jobs.
To put it crudely, both the Peavey and the reference preamp beat the pants off the cheapie. They didn’t sound exactly like one another, but they were pretty close—and they both sounded excellent.
I decided to write up the reference preamp. It’s not too hard to build—all the tough work was done during the design process—and it offers a level of performance superior to anything I’ve heard on the commercial market at a comparable price (about $750). Except for the Peavey.
So why not simply go out and buy a Peavey? I’m not saying you shouldn’t, by any means; it’s a fine preamp and a remarkable design achievement. But there are a few reasons why you might want the reference preamp:
1) The Peavey uses tubes, which generate a lot of heat and use a lot of power. They also burn out eventually and need to be replaced.†
2) The reference preamp is designed to produce optimum results when connecting to +4 balanced inputs. The Peavey, as I reported in my review, does a stellar job with unbalanced inputs, but if you’re running balanced you may find the reference preamp preferable.
3) The two preamps do sound different—good, but different. Having a choice of preamps gives you more flexibility in a session; another color for your palette, if you will. The reference preamp has tighter bass and more precise imaging, while the Peavey has a lusher midrange.
4) Finally, you’re building it yourself. Building a piece of gear gives you an intimate acquaintance with its character and possibilities you can’t get from a store-bought product. You can also tailor it to your own needs, habits, and preferences.
Yes, it’s made with ICs
We all know that integrated circuits are the favorite whipping boys of audiophiles and of many circuit designers. It’s true that bad ICs sound terrible; so do good ones if they’re used badly.
Unfortunately they’re used badly most of the time. Usually to save money, designers take good ICs and place them in circuits where they can’t possibly perform well. As a result many engineers believe categorically that ICs stink and that discrete solid-state (or tube) designs are the only road to high-quality audio.
I beg to differ. To me as a home constructor, discrete solid-state designs aren’t worth the effort, bulk, heat generation, and expense. Building them properly requires testing and matching many components, allowing for greater heat dissipation, budgeting more cubic inches, and spending a lot more money. Frankly, if I’m going to go through that, I’d just as soon go whole hog and build a preamp with tubes. (In fact, I have; an article about that one is in the works.)
Using this preamp (and its predecessors) for the last 12 years has persuaded me that there are many valid roads to good sound, and that integrated circuit designs can in fact sound very, very good—if done well.†
How do you do them well?
The radio perplex
Good integrated circuits are wideband devices, and most of them are extremely susceptible to radio frequency interference (RFI for short). RFI leaks in via the inputs (or other portals—I’ll get there in a minute). Even if you don’t hear disk jockeys and taxi dispatchers in your monitors, the RFI intermodulates with audio to create distortion—the classic “solid-state harshness” so often associated with IC-based equipment.
RFI can also ride the DC voltages that power the device, entering the audio chip via the power supply pins. ICs are fairly good at rejecting audio frequency crud from the power supply, but they provide an open door at radio frequencies.
It is therefore necessary to provide an extremely clean, tightly regulated, low-impedance power supply—and to make sure these characteristics are true at radio frequencies as well as in the audio band. The 3-terminal regulator chips usually found powering recording gear won’t do it.
The supply must also be designed carefully to minimize the generation of its own high frequency garbage. When rectifier diodes switch on and off (as they do in a typical supply) they generate bursts of radio-frequency crud that can leak into audio circuits and really stink up a room.
In this preamp I’ve used individual regulator circuits on each card to minimize interaction between channels. The cards are powered by a pre-regulator, which is driven in turn by a heavy-duty outboard power supply designed specifically to minimize RFI generation. Overkill? Maybe—but it sounds very good.
Drive me, baby
Most integrated circuits have puny output stages. I’m not going to get into a long discussion of “Class-A” and “Class-AB” operation—that’s a whole ’nother article—but suffice to say that to my ears, integrated circuits operating in Class-A can sound clear, clean, sweet, and dynamic (everything ICs aren’t supposed to do) whereas ICs operating in Class-AB have that characteristic “crispy,” “crunchy” sound so prevalent in cheap consoles.
It’s possible to design circuits that run Class-A all the time. One way is to incorporate high-current “buffer circuits” into the outputs, using additional ICs or discrete transistors. Unfortunately this increases bulk and cost, and can sometimes cause instability (nothing like having your preamp turn into a tone generator when you plug in certain cables). Besides, I like to use as few components in the signal circuits as possible; all audio devices add noise and distortion to some degree, and I’d rather avoid that.
Instead I’ve chosen to jack up the idle current of the ICs to a level that guarantees Class-A operation at all times. This is a trick I picked up from Walter Jung, an engineer with Analog Devices with a remarkable gift for creative design and lateral thinking.
A few other things
Integrated circuits need adequate speed. Over two decades ago, Walter Jung formulated a rule of thumb about audio ICs: for each peak volt they need to put out, the “slew rate” (a measure of how quickly the chip can jump from one voltage to another) must be 1 volt per ĶS (V/ĶS). Thus if a chip is putting out 5V peak, its slew rate must be at least 5V/ĶS.†
I used fast chips.
The passive components used in a preamp need to be high-grade. Since I’ve written three articles on the subject (‘Clean Up Your Gear,’ 5,6,7/96) I won’t belabor the point, except to bring up an issue associated with ICs. Since the resistances in mic preamps need to be fairly low (higher resistances are noisier) the coupling capacitors associated with them need to be big. This usually means aluminum electrolytic or tantalum caps—the worst-sounding varieties. Much of the harshness and brittleness often associated with ICs may be due instead to the cheap capacitors usually used with them.
I’ve designed this preamp to use as few coupling capacitors as possible. The input stage is coupled to the main fader by small polypropylene caps for bass rolloff; a large polypropylene is switched in for flat response, or the amplifier may be “nulled” to eliminate DC at its output. In the latter case there is no coupling capacitor for the “flat” setting; the amplifier runs directly into the fader.
The second stage has another polypropylene cap at its input to keep the amplifier’s bias current from making the fader noisy. Other than that, all the chips in the preamp are “direct-coupled”—no capacitors at all. (The best sounding capacitor, after all, is no capacitor. Zen.)
The resistors are all audiophile-grade metal film units—including most of those in the power supply. No, I don’t think this difference will be audible, but 20 years hence they’ll still be close to their original values, while cheap carbon composition resistors will have drifted far away.
In fact, the overall design of this preamp owes a good deal to high-end audiophile design theory. To my ears the results justify the effort; the preamp sounds clean, clear, dynamic, and neutral.
Finally, a word about another favorite whipping boy: transformers. Unlike many contemporary audio designers, I like them (well, good ones). Here’s why.
A high-quality audio transformer (like the Jensens I use) offers a properly balanced input well up into the megahertz region. Many transformerless inputs offer good balance and noise rejection at audio frequencies, but most fall down badly at the radio frequencies that cause problems; while tricks such as “common-mode inductors” (which are really a type of transformer) and ferrite beads certainly help, they still remain vulnerable.†
Let me quote from my colleague and fellow Recording writer Scott Dorsey, who (like me) makes a lot of live recordings:
“When I went from the [transformerless preamp] to the [transformer-coupled preamp], I found myself spending a whole lot less time fiddling around with cable placement to avoid noise pick-up from dimmer systems and the like, which gave me more time to worry about actual recording.” (5/98, review of the Great River mic preamp)
Just so. I’ve been using my preamp and its predecessors for a dozen years in all sorts of situations (including ten feet from an 18kW FM transmitter) and have never—not once—had problems with RFI, dimmer buzz, or any similar phenomenon. I rest my case.
Of course, they need to be good transformers—and for me that translates to Jensen. Deane Jensen managed to design transformers with excellent transient response (not an easy task), low losses, clean bass, and beautifully transparent midrange and treble. For the technically inclined, the JE-115K-E that I use in this design has a “Bessel” response, which means its transient response at high frequencies is close to perfect. This avoids the “hashy” sound prevalent in less well designed transformers.
Enough self-justification; let’s look at the design.
Words of warning and encouragement
This preamp is a fairly straightforward project, but it’s not particularly simple. There are a lot of components to mount on the four boards (preamp cards, pre-regulator and raw supply), two boxes to drill out and punch (the preamp’s cabinet and an outboard box for the power supply), and parts to order from a few sources. You’ll also need to label the fader settings (tedious, but at least they’ll be very accurate), build an umbilical cable to connect the preamp to the power supply box, and perhaps zero out a DC voltage or two.
So this probably shouldn’t be the first construction project you try. On the other hand, please don’t be intimidated; it’s not all that hard either. You’ll need a 30-35W soldering iron (a soldering gun won’t work), needle-nose pliers, wire cutters and a wire stripper, and the usual assortment of screwdrivers. An electric drill will do, but a drill press will do better; you’ll also need a set of high-speed drill bits, a center-punch, and a 1/2" bit for deburring. Nut drivers are nice but not essential.
A digital multimeter is best for testing, but you can get along with a VOM from Radio Shack. You’ll also need a couple of chassis punches: 15/16" and 3/4", for mounting female and male XLR connectors. (You can order them from Allied Electronics along with parts.)
In other words, you need the usual moderately equipped workshop of an amateur audio constructor. You can fabricate the circuit boards yourself if you like, or I’ll be glad to sell them to you at a reasonable price, along with matched resistors for the phantom-power circuits—but I’m getting ahead of myself.
The input stage
Take a look at Figure 1, the first stage of the preamp. We begin with a pair of matched resistors, R101-102. These provide phantom power to microphones that need it; in order to maintain the preamp’s noise rejection they must be matched very closely. You can buy a couple of pairs from me. (I won’t gouge you; I’m not in retail anyway.) Or if you have a digital multimeter, you can buy ten resistors and select matched pairs to within 0.1%.†
The phantom power voltage is regulated on each card by an IC-based regulator; we’ll talk about that when we discuss the power supply.
The transformer comes next. Its turns ratio is 1:10, so it provides a 20 dB voltage step-up. This transformer, the Jensen JT-115K-E, comes in several models with single, double, or triple shielding for use in moderate, bad, or appalling hum fields. I’ve had little difficulty with hum pickup, but if you plan to mount the preamp in a rack with other equipment, you may want to spend a few extra bucks on the heavily shielded model. (Aside from hum pickup they sound the same.)
The transformer mounts to the board using a pair of screws; the primary and secondary wires run through two holes in the board and connect to the appropriate terminals. It’s a good idea to twist the primary wires together gently before you solder them and the secondary wires as well. Be sure you keep the colors straight, and treat the wires gently—they’re fragile.
Many mic preamps and mixing consoles use variable-gain amplifiers at the input stages. These work by changing the resistances around the amplifier—the “feedback network”—usually with a variable resistor or sometimes a switch.
I don’t like this. In the first place, the “reverse-log taper” variable resistors (pots) used in this application are hard to find in small quantities. They also need big coupling capacitors—which usually means electrolytics or tantalums. Feh.
More subtly, changing the feedback around an amplifier circuit (integrated or discrete) changes the level of distortion it produces and often changes the character of the distortion as well. The “harmonic recipe” of the distortion can be altered, which means you may have two channels with very different sound even though you’re ostensibly using the same preamp. I prefer to use two fixed-gain amplifiers with a level control between them.†
How much gain should each amplifier have?†
In 1973 Russell Hamm published a landmark paper, “Tubes Versus Transistors: Is There an Audible Difference?” (Journal of the Audio Engineering Society 21:267, May 1973). He presented a chart of microphone output voltages in typical recording situations, and concluded that the hottest signal likely to be found was 0 dBu, generated by a Neumann U87 microphone six inches from a loud yell. Taking Hamm’s data as representative for the moment, let’s look at what this implies for my mixer design.
I run the amplifiers with hotter than usual supply rails: Ī21.5V rather than the usual Ī15V. Using these voltages, the ICs will put out a maximum of about 17V peak, or +24 dBu. Since the transformers’ voltage gain is 20 dB, that means the input amplifier’s gain must be no more than 4 dB to avoid clipping. So that’s what it is.
Of course Hamm’s data are 25 years old, and there are mics on the market with hotter outputs than the U87. I debated lowering the input stage gain but found the result would compromise the preamp’s noise performance. Instead, should you find yourself clipping the inputs, I suggest using an input pad. You could incorporate a switchable pad into the preamp, but I’ve been burned several times by switches in mic-level circuits, and hate them. (One of them almost got me fired, back when I worked in public TV; it shot craps in the middle of a live fund raiser, just when the string quartet started playing. Pfui.)
Instead I recommend carrying a few in-line fixed pads in your tool kit, and I’ll present a design for one at the end of this series.
The amplifier for the input stage is a high quality, low distortion FET input IC from Burr-Brown, the OPA-604. I have a prejudice in favor of FETs, especially in circuits that see the outside world (even through a transformer). They are far more immune to RFI than standard bipolar transistors, and they tend to be cleaner as well. The OPA-604 is a newish design, expressly intended for high quality audio.
There is a “trimpot” (screwdriver-adjust variable resistor) directly below the IC in the diagram. If you have a digital multimeter you can use this to “zero-out” the DC on the output of the IC. (We’ll discuss how to do this at the end of the article.) With the DC zeroed out you can omit C108 (a big polypropylene) and replace it with a piece of wire. This will sound better and cost less.
C104-105 are there to shunt radio frequency crap away from the amplifiers. I use these “decoupling” caps on every IC package—not just one pair per card like some cost-cutting manufacturers. Eternal vigilance against RFI is the price of audio quality.
Finally, check out the transistor (Q101), resistors (R107-108), and diodes (D101-102) at the bottom center of the diagram. These constitute a “current source,” which biases the output transistors of the IC for full-time class-A operation.
Caps and such
The last parts in the diagram are caps and resistors surrounding the main channel fader, VR103. When S101 is in its center (off) position C106 couples the amplifier directly to the gain control; since the capacitor is small, the bass rolls off at 100Hz. This provides compensation for microphone “proximity effect,” which can heavy up the bass when a mic is used close to a sound source.
S101 can also switch additional capacitors in parallel with C106. When C107 is switched in the capacitance is increased and the bass rolloff frequency is lowered to 30Hz. This provides less rejection of proximity effect, but can be useful in filtering out room noise, leakage, P-pops, etc. When C108 is switched in the response is flat down to 1.6 Hz, and I feel justified in labeling this switch position simply as ‘Flat.’ Of course, if you zero out the IC and replace C108 with a piece of wire (you can delete R109 in that case, incidentally) the circuit will be really flat—down to 0Hz.†
R109-110 are there to prevent popping when you switch t he coupling caps in and out.
Time to fade
The main gain control, or fader, is VR102. You’ll be using this a lot, and I recommend spending some extra bucks and buying a really good one. Penny and Giles make the best, but the price is way high. I think they’re probably worth it, and plan to buy a couple—one of these days.
Meanwhile, some audiophile parts stores (listed in the parts lists) sell high-quality Alps and Noble faders that also work very well. At more reasonable prices, Bourns conductive plastic pots (audio taper) sound good and seem to be reliable.
Buy some cute knobs; I like the retro look and bought some that wouldn’t look out of place in a World War II submarine flick. Make sure your knobs are plastic, though; metal knobs can conduct RFI from your fingers into the chassis. No joke—I’ve measured it.
C301-302 are the input coupling caps for the second stage. Its IC uses bipolar transistors at the input, and they generate a significant bias current that can make the main fader noisy very quickly. By using these (high quality) caps, the fader’s integrity is safe.
I decided not to use a coupling capacitor on the output of the second stage. Instead I used a “servo amplifier,” a separate chip that zeroes out any DC that might appear on the main amplifier. It acts like a large coupling cap, but by some electronic sleight of hand it does so using smaller, higher-quality capacitors that are also much cheaper than big polypropylenes.
The only coupling cap in the entire preamp is C302, a high-quality polypropylene that keeps currents from the second stage away from the main fader. (Which in turn keeps the fader from getting noisy and scratchy. I almost said “itchy and scratchy.”) It’s hidden in the lower right corner of Figure 1; I put it there because it’s connected directly to the main fader rather than living on the printed circuit board.
Check out Figure 2. This is the second stage proper. It’s a straightforward design; the amplifier at the top (facing backwards) is the servo amp, while the bottom amplifier is the one that carries the signal. The main IC is the LT-1028A from Linear Technology—an extremely low-noise chip with low distortion.
Again I’ve jacked up the output stage with a current source (the transistor and associated components) to ensure Class-A operation into normal loads. In practice, the second stage will drive either a typical digital multitrack (10K input impedance) or the eq stage (about 6K input impedance, worst case) while operating in Class-A.
Eq (more or less)
I confess that I was ambivalent about including eq in this preamp design. I’m something of a minimalist; I prefer to control tonality by choice of microphones and positions when possible, rather than diddling with eq controls. After all, every additional stage adds some noise and distortion.
When I do use eq, I prefer a high quality, fully parametric unit when available. But there are times you need a little quick-and-dirty eq on the spot, so I included some in the preamp. It’s rudimentary—gentle shelving in the bass and treble, and a peak and dip control at 1.8 kHz to back off a characteristic squawk in Neumann KM84s—but it’s fully bypassable. With the switch in the “Bypass” position, the preamp connects to an unbalanced recorder with only two amplifiers in the audio chain—about 1/10 the number in a typical console. (Is the difference audible? You bet.)†
Real fanatics can hardwire a jack directly to the output of stage two, avoiding even the possibility of distortion from the switch contacts. (Yes, they can distort.)
Figure 3 shows the eq stage—or stages, since there are two of them. The first handles the 1.8 kHz peak and dip control, while the second takes care of bass and treble shelving eqs, with turnover frequencies of 200 Hz and 2 kHz respectively. Each stage inverts the polarity of the signal—positive-going voltages become negative-going and vice-versa—but because it happens twice, the overall polarity remains the same. Switching the eq in and out will not put your signal “out of phase.” (Thanks to the estimable Walter Jung for the basic designs.)
Once again the amplifiers are the trusty Burr-Brown OPA-604s, and once again they’re helped by current sources on the outputs. Even the worst possible load—midrange set to maximum cut, bass and treble to maximum boost—won’t drive the output sections out of Class-A.
I prefer unbalanced inputs when possible. As I mentioned before, the unbalanced output of this preamp with the eq bypassed places only two active chips between the microphone and the recorder, leading to clear and pristine sound that’s very appealing.†
Still, there are times when you need a high-level balanced output. You may be working in a noisy environment, where the rejection of line crud becomes important. You may have hum problems from ground loops. Or you may own a piece of gear with nothing but balanced inputs. Either way, a professional mic preamp must have at least the option of balance.
Figure 4 shows the balanced output stage. The first amplifier raises the nominal signal level from -10 dBV to -2 dBu, forming the positive half of the balanced output, while the second amplifier flips the signal over to form the negative half. The difference between the two output signals amounts to a +4 dBu nominal output level, the standard professional voltage.
Here I was faced with a dilemma. Using current sources on the outputs, I was able to drive a 5K balanced input tape recorder while remaining in Class-A. This is the lowest impedance most people are likely to see in a project studio, and you’ll only see that if you luck into a used Studer. (Typical digital multitracks have a balanced input impedance of 10K, a breeze to drive with this circuit.)
A few pieces of gear, however, still have input impedances of 600 ohms, the ancient standard inherited from the telephone company, and they draw a lot of current—18 mA on the highest peaks. These chips will provide that—but they won’t be Class-A anymore.
Still, 600 ohm loads are increasingly rare in the world of modern recording gear; Pultec equalizers (and their clones) and phone lines are most likely to show up and cause trouble. Well, if you can afford Pultecs, you can afford to stick an extra high-current board into the mic preamp to drive ’em. And phone lines usually sound terrible anyway, so a little Class-AB isn’t really an issue.
The bottom line is that these balanced outputs will drive anything you’re likely to own in the real world with Class-A quality and no compromises—and when they do drop out of Class-A, they’ll do so a lot more cleanly than most gear.
I should point out that while these outputs are balanced, they aren’t floating; the outputs are still referenced to the ground potential of the preamp. Most of the time this will work perfectly, even in situations that are prone to ground loops, if the gear you’re feeding has a floating (transformer-coupled) input. However, once in a while you may need complete isolation; for those times, good 1:1 transformers are the way to go. I recommend the Jensen JT-11-BMCF ($97.00 each and worth it); you can install a pair permanently inside the case, or connect cables and use them free-standing.
Power from the wall
It’s not time to rush out and build your preamp just yet; we need to talk about power supplies first. Power supplies are often an afterthought in audio design; slap a raw supply together with a wall wart, a diode bridge and a coupla big capacitors, then regulate it with a 40-cent 3-terminal IC, and Bob’s your uncle.
Except that Bob, in this case, is an uncle prone to get into trouble. Power supplies can pass along large quantities of garbage that rides the AC line: spikes and dips from motor switching, hash from dimmers, trash from radio stations, pagers, and taxi dispatchers. The diodes in power supplies can also generate their own radio frequency hash, which can get amplified by resonances in the power supply circuit and passed along into audio chips, causing harsh or “hard” sound.
This power supply is different. Following the ideas of Ben Duncan, a British designer with a great penchant for detail, the raw supply is designed to minimize the generation of RFI and to filter it out when it appears. I use on-card regulators to provide maximum isolation for each channel; to ensure that each regulator sees a stable environment, I use a pre-regulator stage to drive the on-card regulators.
Take a look at Figure 5. This is the raw supply—the one that plugs into the wall. Note the capacitors and MOVs across the AC line; these absorb and deflect RFI, shunting it to the chassis where it will do the least harm, and clamp short-term spikes to prevent their causing pops in the recorded sound.
The transformers seem over-designed, and in fact they are; I specify higher than normal voltages to ensure adequate output when the line voltage is low (as it often is in real working environments). The supplies can take a Ī10% line voltage fluctuation without turning a hair. The transformers are also rated for higher currents than the mic preamp actually draws; they loaf along generating little heat, and I’ve never had one fail on me. (I’ve deliberately used overrated parts throughout the design to assure high reliability and long life.)
The diodes are unusual. They have extremely fast switching times; tests (Audio Amateur, 1/97) have shown that these diodes generate much less garbage than common, slow-switching diodes.†
Instead of using a single huge filter capacitor, I use ten smaller capacitors for each supply. Smaller caps have lower inductances, which translates to much lower impedances at radio frequencies, and therefore to better filtering of RF garbage. To help them out there is a small stacked-film cap wired in parallel with the arrayed electrolytics. A small resistor at the input of the capacitor bank helps damp RFI resonances.
The raw power supply gets its own box; I have battled electromagnetic coupling from power transformers into input stages too many times, and life is too short. Mounting the power transformers four feet away from the audio circuits saves much aggravation. I connect the two boxes using an umbilicus and 5-pin XLR connectors. Use a 5-conductor cable (plus shield), and ground the shield to the chassis connection of the XLR plug at the power supply end only. Leave the other end floating! I use a male XLR connector in the power supply box, a female in the mixer box, and complementary plugs on the umbilical cable.
Note that the main power supplies are always on; the phantom supply can be switched on and off as needed.
[Comments re. System Grounding as seen on Fig. 14:†The cable shield of the power umbilicus is connected to the XLR plug’s shell at the preamp end only. Similarly, the shields on output cables (if you use shielded cables) are connected to chassis ground at one end only. If the cable connecting the mike input to the card uses three internal conductors plus a separate shield, connect the shield to chassis at the jack, leaving the other end unconnected. If it has only two conductors plus shield, connect as shown, using the shield for ground. The 1/4" jacks specified are nylon, and do not ground to the chassis; RCA jacks should be isolated from the chassis with shoulder washers. The “Main System Ground” is a terminal strip, mounted close to the XLR connector where the power umbilicus plugs in.]
If it moves, regulate it
Once the DC voltages arrive at the preamp chassis, they receive a massage designed to filter out as much crud as possible. (See Figs. 6a and 6b.) Ceramic disc capacitors (connected from the umbilical socket directly to the chassis) and ferrite beads filter out RF garbage. A resistor and second capacitor bank filter more junk, including whatever hum made it out of the original bank, and a regulator stabilizes the voltage feeding the preamp cards.
I used adjustable-voltage regulators rather than fixed-voltage chips; they have much lower impedances and noise, and they reject line crud more effectively than their cheaper cousins. Each pre-regulator also drives an LED, giving you a quick check that all three power supplies are working properly.
The on-card regulators themselves (Figs. 7a and 7b) are high-quality, low-noise, and low-impedance. The Ī21.5 regulators that power the audio ICs are themselves based on a high-quality audio IC chip, the well-known NE-5534. Voltage references are provided by high-stability integrated circuits. The phantom power regulator is another adjustable-output IC, this time floating on a 47V zener diode. This provides a stable, quiet phantom supply that has worked well with every microphone I’ve tried.
You’ll need hardware to put the whole thing together. Everything is available from Digi-Key except for flathead screws; you’ll need to get them at your local hardware store. C’mon, I can’t find everything for you!
Wire is a tough issue. You can go utterly nuts about it, as some audiophiles do; the new Parts Connection catalog lists a pure silver hook-up wire at $16.80 per foot (plus shipping).†
I think that’s overkill. The wire you use for carrying signals does make a difference to my ears; the insulating material should be polypropylene or Teflon™ (the latter doesn’t melt when you solder it). The trouble with Teflon wire is that it’s almost always silver-plated, which makes it sound brighter and zingier (no, I don’t know why). Audiophiles often love that sound; I hate it, so I prefer polypropylene.
The conductors inside Belden 8450 are good, but it’s almost impossible to find in small quantities (8451 is more available, but uses stranded conductors, which are more annoying). In the parts list I’ve specified a proletarian PVC wire, but my Generous Offer includes enough 8451 to do the signal wiring. Or you can get as fancy as you like.
Proletarian wire will do fine for the power supply wiring, as it is upstream from the on-card regulators.
Soon it will be time to put it all together and start recording.
You’ll need the usual electronic workbench tools. A 30-35W soldering iron is mandatory (a soldering gun won’t do); if you have a temperature-controlled soldering station, well and good. You’ll also need a damp sponge (a grocery store sponge is fine) and a rack to hold the iron.
It’s useful to have an aloe plant to treat the inevitable burns. First chill the burned part in ice water, then rub aloe leaf juice on it. This works a lot better than either measure alone.
Good electronic solder is important; Radio Shack’s standard solder is okay, Ersin MultiCore™ is better, and the fancy silver doped stuff from WBT is the best I’ve used. (Audiophiles claim to hear differences from one solder to another. I don’t go that far, but will testify that I have fewer cold joints with WBT. It costs a bundle, but it’s worth it.)
You’ll also need an electric drill, or preferably a drill press. I finally bought a Delta bench unit, and my home-built projects are a lot cleaner-looking as a result. Along with the drill (press) you’ll need high speed bits: 1/8", 5/32", 3/16", 1/4", 3/8", and 1/2". You’ll also need a center punch (or you can get by with a large nail) and a hammer to hit it with.
To make clean holes for the input connectors, you’ll need chassis punches: 3/4" (for male XLR jacks) and 15/16" (for female ditto). They’ll set you back about $60 for the pair, but they’ll last well into the next century. If you’re feeling flush, you can spring for a 1/2" punch too; it’ll make much cleaner holes than any drill. Don’t forget a wrench to crank the punches.
Finally, the usual stuff: needle-nosed pliers, screwdrivers (flat and Phillips), diagonal cutters, and wire strippers.
I am board
Time to look at the circuit boards. If you’re ambitious and have done it before, you can make your own; send a SASE to the magazine (see sidebar) for the board patterns and layouts. These don’t appear in this article because they’d have to be printed at precisely their actual size, which would use up six pages of the magazine! Remember to make two audio boards for a stereo preamp. (By the way, if you’re keeping track of Figure numbers from Parts 1 and 2, these would be Figures 8a/b, 9a/b, and 10a/b.)
You can do it—but you may not want to. You’ll need to make photographic copies of the patterns, print them onto sensitized boards, develop the images, and etch them in extremely corrosive and toxic chemicals. Then you must stand over a drill press, tediously drilling out each tiny hole (and replacing drill bits as they break off). Feh. If you’d rather not go through this, check out the Generous Offer at the end of the article.†
In any case, you now have a set of circuit boards; time to load them up. Sort out your parts (a muffin tin is helpful), bend the leads carefully with your needle-nosed pliers, and insert them into the board one by one, soldering as you go.
First do the jumpers (there are twelve on each audio board), then resistors and trimpots, then capacitors, then diodes and transistors, then (finally) integrated circuits. Before you insert the power supply transistors and pre-regulator ICs, smear them with silicone grease (available from Radio Shack) and attach their heat sinks.
Electrolytic caps have their terminals marked; be sure the plus terminal on the capacitor matches the one on the circuit board, or the cap may explode. Similarly, make sure the ICs are inserted correctly; there’s a “1” on the circuit board next to pin 1, and a corresponding dimple or notch on the IC package. The ICs in this design aren’t as sensitive to static electricity as some I’ve tried, but it’s still a good idea to wear a loop of wire around your wrist, with the other end connected to a real ground (perhaps a water pipe—and not a PVC one, please). Finally, make sure you mount diodes correctly; the banded end is shown on the layout.
You may notice a few components on the circuit board layouts that don’t appear on the parts lists. Don’t panic—these boards were originally designed as part of a modular mixing system, and some vestigial designations remain. Rather like your appendix—there, but useless.
When all the parts are inserted and soldered, check the boards with a magnifying glass. Look for solder bridges between conductors (easy to do, especially on IC packages) and bad solder joints (they look like a pile of crystals rather than a smooth surface of metal). Attach the input transformers (there are two holes for screws, included with the transformers, and two to thread wires through.) Twist the output wires (yellow, orange, black and white) together and solder them to their respective terminals (yellow goes to T, the rest connect to U). Similarly, twist the input wires (red and brown) together; red goes to MIKE IN +, brown to MIKE IN -.†
We’ll attach wires to the boards later.
Build the external power supply box first. Figure 11 shows a practical layout; drill out 1/8" holes to mount the raw supply board, 5/32" holes to mount the transformers, and 1/8" starter holes for everything else. Enlarge the starter holes to 1/4", then to 3/8"; use a 3/4" chassis punch to make the hole for the 5-pin XLR connector. Insert the connector into the large hole, then mark and drill the 1/8" holes for the connector’s mounting screws.
Deburr the small holes with a 1/2" drill bit. Wash the chassis with soap and water to remove oil (you do oil your chassis punches, I hope?) and finger marks. Label the switch and fuses with rub-on lettering, available at electronics or graphic arts stores. I use Datak™ or Chartpak™ letters; every other brand I’ve tried dissolves in the protective coating. Spray with transparent lacquer (in a well ventilated place!) to protect the lettering, or gently brush lacquer over the labels.
Mount the terminal strip in the supply box (you can find terminal strips at Radio Shack or any electronics supply house). Mount the raw supply board temporarily, using hexagonal spacers and 4-40 screws; use lock washers between the spacers and the chassis. Attach the XLR connector using 4-40 flat head screws, lock washers, and nuts.
Measure out lengths of wire from the circuit board outputs to the XLR connector, and from the “ch” terminal to the chassis ground lug on the terminal strip. Solder the wires to their appropriate terminals. (It’s a good idea to label each wire with a piece of masking tape.) Mount the transformers in the box and connect their secondary wires to the circuit board. Re-mount the circuit board on its spacers, this time using lock washers on the mounting screws. Solder the wires to the XLR terminals as shown in Figure 11.
Attach the fuse holders and phantom supply switch to the chassis, as well as the terminal strip. Check the power switch with an ohmmeter to be sure the On position corresponds to “up.”
Connect the transformer primaries as shown, but don’t solder anything yet. Insert a grommet in the hole for the power cord, thread the power cord through it, and tie a knot in the cord. (This is an old radio trick from the 1930s, known as a “poor man’s strain relief.”) Be sure there’s enough wire protruding after the knot to connect to the switch and fuse holders; strip and connect the power cord; note that the ground wire (green) from the power cord connects to the grounded lug of the terminal strip. The chassis ground wire from the raw supply board (terminal “ch”) goes to the same lug. Solder all connections on the fuse holders, terminal strip, and switch.
The power box should now be complete. Do a “smoke test”: plug it in and, standing well back, turn it on. If there’s no smoke, measure the voltage between the chassis and the ground (round) terminal of your wall outlet. This should be zero. If it isn’t, unplug the unit and wait five minutes for the caps to discharge, then check all connections.
Assuming the smoke test comes out okay, check the output voltages at terminals 1-5 on the board, with the ground terminal of your voltmeter connected to the chassis. The voltages should be:
These voltages are approximate, and will depend on the voltage coming from your wall outlet and the quality of your transformers. Don’t worry if they’re high, but if they’re more than 10% low, unplug, wait for discharge, and check all your connections again.
Wire up the umbilicus. Use your voltmeter to test it, making sure each pin on the source end connects to the matching pin on the other end, and no other.
The Thing itself
Time to wire the preamp. You can lay it out in any way that works for you; figure 12 shows one possible configuration, while figure 13 shows the equivalent front panel layout. Drill holes as before, starting with small ones and enlarging them as needed. (Don’t forget to drill the small alignment holes for the pots.)
Remember that you’ll need a 15/16" punch for the female XLR connectors; again, it’s much easier to punch and drill the small screw holes on the XLRs after you’ve made the large holes. Placing a piece of wood under the metal panel while you’re drilling makes for cleaner holes. When you’ve finished drilling and deburring, wash, dry, and label the panel as before. Leave room for the knobs when you’re labeling the eq pots, and don’t label the main fader yet.
You’ll note that although the audio boards allow you to mount eq pots directly onto the board, using the panel mounted pots as the main support for the board, I haven’t done this. In the first place, I don’t really trust this type of mounting; a hard jolt causes the board to flex, possibly damaging the pot-to-board connection. Second, this configuration makes replacing bad pots a real chore—and pots are the most likely components to go bad over the years. Finally, if you drill mounting holes a fraction of an inch out of line, the board won’t mount in the box at all. No thanks.
Figure 14 shows the ground wiring of the system. Mount the back panel parts and the pre-regulator board (the latter temporarily); measure the wires from the pre-regulator board to the umbilical connector, then remove the board and solder on the wires. Measure and solder the wires that will run to the circuit cards and LEDs; tape them out of the way so they won’t short to the chassis during the next smoke test. When the ground wiring is complete, attach (and tape) the power wires.
Remount the pre-regulator board (this time using lock washers), plug the umbilicus into the power box and the chassis, and plug in the power box. Measure the outputs of the pre-regulator board, using terminals GGG as a reference; they should be within 10% of the following:
If any of these is seriously wrong, unplug the power pack from the wall and wait five minutes for the whole thing to discharge, then check your connections again.
Important: DO NOT plug or unplug the umbilicus when the power box is plugged into the wall; you can blow things up that way. Instead, be sure the umbilicus is connected before you plug in the power box. (Bitter experience.)
The Real Stuff at last
Mount parts on the front panel, including the knob on the gain control but not the LEDs or knobs on the eq controls, and temporarily mount the audio boards. (Remember that the swinger of a miniature switch points up when the handle points down.) Measure out wires from the audio boards to the back and front panel connections, including the eq pots; cut and strip these wires, remove the boards, and solder the wires to their respective terminals. (Again, labeling the wires with bits of masking tape and a Sharpie™ really helps.) Solder the ground and power wires to the audio boards and remount the boards using lock washers. Now solder the other ends of the wires to the jacks and controls.
You’re done! Well, almost. Plug the box in. Plug a low-level signal source (a Shure mic-level oscillator works nicely) into each input, connect the output to your system, and verify that everything works.†
It’s time for the most tedious task: calibrating the gain controls. Be sure the eq is turned off and the bass rolloff is set to “Flat.” Decide which output you’re going to use most (-10 dBV or +4 dBu) and connect it to your recorder. You should have a high quality audio voltmeter (most digital multimeters won’t do) and a high quality signal generator (I’ll tell you in a moment how to make do without these things).†
Connect the voltmeter in parallel with your preamp’s output, and the signal generator (set to an output of -50 dBu) to its input. (In the following discussion, I’ll assume you’re using the -10 dBV output; scale appropriately for the +4 dBu output.) Making sure the signal source level remains constant, measure the gain (in dB) from the source to the output, with the fader wide open; it should be a bit over 50 dB. Lower the fader until the gain is exactly 50 dB. Mark this setting, then lower the gain control until the gain is -45 dB; mark this setting too. Continue lowering the gain control in 5 dB increments (or finer, if you want to take the trouble) until you get to 0 dB of gain. You’re not likely to use any lower settings, so turn the control all the way down, mark that spot, and repeat on the other channel.
Using rub-on transfer letters, mark the various gain settings. Remove the gain control knob and carefully unscrew the parts from the front panel, then remove it. Place the panel on a sheet of newspaper and spray with transparent Krylon or brush lacquer over the labels. When it’s dry, replace the panel and remount the various parts. Repeat for the back panel.
Insert the LEDs into the front panel. Connect and solder the wires from terminals EEE-FFF and LLL-OOO. Thread an inch or so of heatshrink tubing (Radio Shack again) onto each wire and twist the wires onto the LEDs’ leads, but don’t solder them yet. The cathode end of each LED has a longer wire; it corresponds to the end with a bar on the circuit diagram.
Plug the power box in and check to see that all three LEDs light up. If one doesn’t, untwist the connecting wires and connect them the other way around; that should fix the problem. When all the LEDs are working, unplug the box, let it discharge, and solder the wires to the LED leads; use a butane lighter to shrink the heatshrink tubing around the solder joint.
Turn the gain back to full on and set the shafts on the eq pots to their approximate midpoints. Set the signal generator to 1800 Hz; turn the eq switch on. The output should change slightly; adjust the midrange eq pot until the output remains the same with the switch on and off. Repeat this test with the signal generator set to 100 Hz, adjusting the bass eq pot, and with the generator set to 10 kHz, adjusting the treble eq pot. Go back and recheck all three (the controls interact a bit); when all three are correct, mount the knobs on the shafts with the pointers straight up.
In a pinch you can get away without the signal generator and fancy voltmeter. There are several test CDs on the market with sine waves at various frequencies; use a DI to connect your CD player to the preamp input. (Setting the CD player on repeat avoids aggravation.) Use 440 Hz or so for the gain control calibration. Your eq settings may not be quite as exact using this method (your CD player or DI may not have perfectly flat frequency response) but you should get pretty close—and you’ll tweak eq by ear anyway. Another possible tone source is a synth or an electric organ; put a rock on the appropriate key.
For measuring, I find the headroom indicator on my DAT machine is actually a good deal more accurate than my fancy voltmeter, although I need to do a bit of arithmetic to figure out the preamp’s gain!
The preamp, as described, has a midrange eq that’s limited to a single frequency: 1.8 kHz. As I said last month, when more flexible eq is needed I’d be inclined to patch in a fully parametric unit such as the UREI. But if you’d like increased flexibility built into the preamp, Figure 15 shows one possibility. C403a-404a mount into the circuit board where C403-404 went before, but they’re now much smaller; if left alone, the peak/dip frequency will be 14.4 kHz. The switches, mounted on the front panel, connect additional capacitance in parallel with C403a-404a, lowering the operating frequency. (Parts List 1100, which ran last month, lists the extra pieces needed for this variation.)
When I designed this circuit, the Neumann U87 was the hottest mic typically found in a studio (except for a few oddballs that put out line-level signals). The input stage was designed to be overload-proof when fed a loud signal (drumbeat or yell). Times change, and recent microphones from several manufacturers are hotter than the U87. This poses the possibility of input amp overload and input transformer saturation (possible even with these excellent trannies).
I’ve designed an input pad that will prevent both, two versions of which are shown in Figure 16. Let me reiterate that I hate switching mic-level signals; even with the specified gold contact switches, there’s too much potential for disaster. So I don’t really recommend the switched version unless you’re doing live recording and habitually use hot condenser mics on horns, guitar amps and vocal screamers. Instead, I suggest using the doodad shown in the right half of Figure 16—the same circuit, but shorn of switching—and building it into a cylindrical XLR-F/XLR-M connector. These pluggable pads are always in my mic suitcase…just in case.
While you’re buying those little metal cylinders, get a couple of extras and wire them to reverse polarity—pin 2 to pin 3, and vice versa. I seldom need to invert a mic signal, but I’d much rather pop one of these in line when needed than include a “phase reverse” switch in the preamp—one less thing to screw up the signal.†
In choosing parts for this design, I’ve specified good quality passive components—polypropylene caps, metal film resistors, conductive plastic pots—without going hog wild about it. Believe me, there are much more expensive parts out there, and you’re welcome to use them: MIT capacitors, Vishay foil resistors, and stepped attenuators instead of audio taper pots. By carefully browsing the Parts Connection catalog, you can easily double the cost of building this preamp.
Is that worth it? Up to you; the better components will make a difference in the sound, but a very slight one—the law of diminishing returns sets in pretty quickly. Personally, I’d rather spend the extra bucks on more microphones, but you may feel otherwise—and there is certainly a feeling of satisfaction in wringing every possible ounce of performance out of a circuit.
The Generous Offer
Most of the parts are easily available (at least for North American readers) from the sources mentioned last month. A few are not; mail order houses won’t sell you 6' of 5-conductor shielded cable for the umbilicus, or phantom power resistors matched to 0.1%. And making circuit boards, as mentioned above, is a real pain. So I’m making this generous offer: I will gladly sell you a set of Difficult Parts, as follows:
2 audio circuit boards
1 pre-regulator circuit board
1 raw supply circuit board
6’ 5-conductor shielded cable
4 matched resistors for the phantom supply circuit
The price is still up in the air (I’m negotiating with the circuit board maker), but I expect the whole thing to go for between $50 and $100. If you’re interested, send me a SASE (c/o Generous Offer, Recording, 5412 Idylwild Trail, Suite 100, Boulder, CO 80301) or email firstname.lastname@example.org with “Generous Offer” in the Subject line; by the time this article hits the stands, I’ll have the numbers.†
This design is open-ended; you could incorporate various sections into a modular preamp/mixing system and otherwise make the design sit up and do tricks. The power supply design is broadly applicable; you can use it, with voltage changes, for many audio construction projects (including digital/analog combinations, changing the phantom power supply to a 5V supply for digital chips).
Most importantly, though, you can use this preamp to make very, very good recordings. Go out and build one—get some gut-level experience with gear, get more than your money’s worth, and get a great deal of satisfaction in the bargain.
Oh, and don’t forget the aloe plant.
Paul J. Stamler specializes in traditional folk music, classical music, and songs about chickens. He may be reached via email@example.com