| Home | Audio mag. | Stereo Review mag. | High Fidelity mag. | AE/AA mag. |
PRE-AMPLIFIER TECHNOLOGY has undergone many changes in the past few years. It used to be relatively easy to select and afford what you could be reasonably sure was the most accurate preamp. Doing so was simply a matter of paying your local Audio Research dealer $500 for the latest in carnation of the SP-3 series. Times have certainly changed, though. Several manufacturers, convinced of the marketability of a no-compromise approach, have introduced preamps and power amps in the mega-performance/ dollar category. People who have had the opportunity to audition these units generally agree that the components, circuit topology, regulation, grounding and other techniques used have sonic merit. No doubt this technology carries a high price in the consumer manufacturing arena, but the home constructor need not suffer. My solution to this problem is Daniel (Photo 1), a preamp named in honor of my first child. In discussing this unit, I will refer to my vacuum tube pre-preamp article in TAA 5/84 (p. 7), so you might want to review that text before reading on. Guidelines To help you understand this design better, let's take a look at the rules I followed when constructing the circuit blocks. PHOTO 1: Daniel, a preamp named in honor of the author's son, is a vacuum tube unit that complements his pre-preamp in TAA 5/84. Rule #1: Minimize the amplification stages. Each time the audio signal is processed, some degradation occurs. To alleviate this problem, my design minimizes the active circuitry. Still, the preamp has sufficient gain (and low noise) to accept most moving-coil cartridges directly. In selecting the amplifier topology, I used the following criteria, in order of importance: dynamic linearity, gain and low output impedance. For example, I decided that the cathode follower's lower output impedance did not offset its inherent disadvantages of sonic deterioration and unity gain. Rule #2: Use sonically premium components. Although this would seem to be relatively easy to do, not everyone agrees on which components are best. In the pre-preamp article, I said that each component has its own sonic character. You must, therefore, strike just the right balance among the components to pro duce the desired sonic flavor. This is not to say that you should mask severe deficiencies in certain components, but rather that you should recognize the imperfections in even the finest parts and take care not to compound these weaknesses when making your other selections. Be cause I have chosen the components for this design so carefully, I would suggest that you build the unit as described before making any changes. That way, any experimentation will benefit us all equally. Rule #3: Use regulated power supplies. The active gain stage is really a valve that modulates the current supplied by the power supply in response to the signal at the input of that particular stage. If this supply contains any signals of its own, they will modulate the desired audio in formation. Power-supply signals arise from three sources-the dynamic current demand of the driven audio circuit, the dynamic current demand of other stages and fluctuations in the AC line. A properly designed supply responds minimally to these disturbances, and any reaction should return to normal in the least reaction time and with the least overshoot. Although building a power supply to meet these criteria minimizes its sonic interaction, the supply still has its own flavor. Using individual supplies for each stage also minimizes the interaction among stages. If you wish to experiment, try alternate op amps and series pass devices (e.g., power MOSFETs such as the VN10KM). Again, I chose the devices so as not to accumulate similar sonic flavors. Rule #4: Use optimal signal routing. Because all conductors exhibit impedance, voltage potentials arise during current flow. Conductors used as grounds, or “references, ” in reality are not. We have all experienced the typical ground loop. This usually presents itself at the earliest amplification stage-the phono input-where the gain is sufficient to amplify these small voltages. Although in later stages the required amplification might not be sufficient for you to hear these voltages clearly, they can obscure the audio signal. In that case, you must pay close attention to the individual stages, their bias currents, and the audio and power return paths. You must also be concerned about the filament cur rents, which by comparison dwarf the bias currents in each stage. Circuit Description The first amplifier stage (Fig. 1) has a 6DJ8 in a slightly unconventional cascode arrangement. The upper section operates in a fixed-bias mode in stead of the typical self-bias arrangement. This dramatically increases the perceived sense of dynamics, which was perhaps the cascode's major sonic limitation. It is very important that you bypass this grid with a high-quality capacitor. In the SP-10, Audio Research has actually “fixed biased ” this grid with a regulated power supply. I tried this approach, but found no sonic advantage, probably because of the grid's extremely high input impedance. ------------ 8 NOTE: THE VALUES IN THE SCHEMATIC ARE NOMINAL. SEE THE PARTS LIST FOR IDEAL VALUES. -------------- PARTS LIST Capacitors 1,7 0.014F, S00V disk ceramic (§539-GP110) C2-4 22yF, 350V electrolytic (§19AG022) cs 100 uF, 25V electrolytic (#20NK100) C6 20,0004F, 25V electrolytic (#539-CGS203U025V3C) c8,9 1uF, 450V electrolytic (§19AF001) C10, 101, 103, 104, 106, 107, 109, 201, 203, 204, 206, 207, 209 0.14F, 400V Mylar (§23DK410) C102, 105, 108, 202, 205, 208 0.47uF, S0V ceramic (Sprague #2CZ5U474X00S0C4) C110, 210 0.1 uF, 425V WonderCap (IAR #UK-343/104420) cm, 211 0.0033,F, 630V polypropylene (§23PP233) C112, 113, 212, 213 0.0015 uF, 630V polypropylene (§23PP215) C114, 214 (select) polystyrene (=220pF) C115, 215 0.014F, 630V wWonderCap (IAR #UK-363/103620) C116, 117, 216, 217 0.47yF, 425V WonderCap (IAR #UK-343/474420) C118-120, 218-220 5 uF, 25V electrolytic (#20NK0O05) Resistorstt R1, 114, 127, 128, 135, 136, 214, 227, 228, 235, 236 33Q carbon R2 470k, 2W carbon R3, 130, 138, 230, 238 R4, 102, 106, 110, 202, 206, 210 RS, 6, 10, 103, 107, 111, 203, 207 211 1002 carbon ud 3k, 2W cabon RS ~75k, 2W carbon R9 2200 carbon od presi. p R101, 105, 109, 119, 131, 132, 139, 201 : : : ! 205, 209, 219, 231, 232, 239 3.3M carbon EE vie a 208 212 > ' ; og EE SE a aan ig R117, 118, 217, 218 (dow gain, MM) *R117, 118, 217, 218 4700 carbon 10k carbon 100k, 1W carbon 47k, 1% (§29MF500) 20K, 1%, 1W (§28MX200) 3900, 1% (#29MF500) (high gain, MC) R120, 220 R121, 221 (low gain, MM) *R121, 221 (high gain, MC) R122, 222 R123, 223 R124, 224 R126, 129, 134, 137, 226, 229, 234, 237 R140, 240 3900 carbon R141, 241 910 carbon TTAIl resistors 0.5W metal film unless otherwise noted. 1002, 1% (#29MF500) 475k, 1% (#29MF500) 150k, 1% (#29MF500) 100k, 1% (#29MF500) 715k, 1% (#29MF500) 95.3k, 1% (#29MF500) 10M carbon 3900, 1% (§29MF500) Diodes, Fuses, ICs D1, 2, 4,101, 201 1,000V, 1A rectifier (#333-1N4007) D3 50V, 25A bridge (#33BR250) DZ1 15V, 1W zener (§597-1N4744A) DZ2, 23 15V, SW zener (§597-1N53XX) F1 3A, 125V slow-blow (#504-MDL-3) F2 1.5A, 125V slow-blow (#504-MDL-1.5) Ic opto-coupler, triac out (Motorola #MOC4030) IC2 1A adj. regulator (National Semi. LM317T) IC101-103, 201-203 LF351 op amp (National Semi. LF351N) IC104, 204 3A adj. regulator (National Semi. LF317K) aiec. = LED 1 100mMW, 200mA LED (§35BL501) 1 MJE340 Q101-103, 201-203 NPN high-speed transistor (#333-KN4400) T™ 120V ct, 0.25A transformer (Signal Transformer #2416120 T2 20V ct, 5A transformer (Signal Transformer #241820 V101-103, 201-203 dual triode valve, 6DJ8/6922 (1) 2 ” diameter capacitor holder (§539-VR8); (4) PC-mount fuse clips (#504-1A1120-10); (2) 2x2x1.25 ” TO-3 heatsinks (§33HS-306); (1) T0-220 heatsink (#33HS-222); (1) screw-type relay socket (Midland Ross/Midtex §670-0125); (6) 9-pin min. tube sockets (TRW #9PC-B1); (11) 7-pin/pole terminal strips (WIBA #25-104-0753); (1) 4-pin/pole ter minal strip (WIBA §#25-104-0453) 3C, 110V AC relay (Midland Ross/Midtex #157-237200) T AIll parts sources are Mouser unless otherwise noted. --------------------------- The first stage is primarily responsible for the preamp's signal-to-noise ratio and input-overload margin. I have shown it in two bias/gain schemes. The un-starred component values provide 31dB of gain and should be used with high-output 2mV cartridges, while the starred bias provides 38dB of gain and is suitable for most moving coils. Should you require additional gain, refer to my pre-preamp article. Note that the upper grid's fixed bias point must also shift with the chosen gain arrangement. The first stage feeds the RIAA net work. Although people still disagree about the sonic performance of passive equalization, I believe the debate is meaningless outside the context of the supporting circuitry. For example, correctly designed active equalization that is driven by a circuit with proper output impedance will sonically outperform a passive design with little regard for correct source/sink parameters. My circuit is directly coupled into the RIAA equalizer. The coupling capacitor is located at the highest impedance point in that path-the next stage's control grid. The values I have chosen will no doubt spark some controversy, and I invite your input. FIGURE 2: Schematic diagram of the power supply 's preregulator section. FIGURE 3 ---------Fig 4; Fig 5 The second phono stage (Fig. 2) is identical to the high-level section. Both are comprised of a dual triode operating with the sections in parallel. This configuration has demonstrated stability and reliability and, most important, a neutral sonic character. Although the output impedance does not equal that of the cathode-follower configuration, it does not suffer from the sonic deterioration of single-supply cathode followers. This circuit provides 26dB of gain and can easily drive an input impedance of 20k or greater. Note the resistor in series with the output capacitor. As part of the total design, it compensates for the high-frequency edge I find objectionable in the 6DJ8 and the ability of the Wonder Caps to pass information. Another option would have been to use a capacitor with a high-frequency balance that could exactly compensate for the 6DJ8. I have tried this, but have not been satisfied with the long term results. Each gain stage is supported by identical high-voltage regulators, which operate at the same output voltage and share a common bias and reference. They are supplied by a floating LM317 circuit. Although the 317 has some limitations (see “Measuring Power Supply Output Impedance,” TAA 1/83, p. 13; 2/83, p. 20, and 'A High-Voltage, Low-Z Regulator,' TAA 2/84, p. 25), this mother/slave arrangement takes ad vantage of its virtues, while isolating its limitations from the audio circuitry via the slaves, which are optimized to meet the regulator criteria. In addition, to minimize the impedance of the interconnecting conductors, these slaves are close to their respective amplifier stages. This consideration is often neglected in commercial designs. The two filament regulators (one per channel) are high-current LM317Ks in TO-3 packages. I have heard conflicting views of the application of filament supplies, but no one method is consistently favored. Choices range from one regulator per stage (the auto-battery approach) to a high-impedance current source. The configuration I selected provides distributed heat dissipation, channel-to channel symmetry and minimum ripple. Any other sonic advantages are welcome but purely accidental. PHOTO 2 PHOTO 3: The author's preamp prototype is housed in an SP-6. Fig. 6 The master power supply (Fig. 3) consists of high-voltage/low-current and low-voltage/high-current sup plies. Both supplies and the associated AC line switching are contained in a separate chassis. This ensures that any magnetic fields and mechanical vibration resulting from the power transformers' proximity to the audio circuitry are of no concern. The high-voltage master supply is a voltage doubler, which I chose primarily because the transformer is available from the same vendor. The raw filament supply is a full-wave, center-tap type. A bridge rectifier is specified to allow a negative voltage to be developed. This supply is always “on, ” but the filament current is interrupted by a relay contact. This relay, which also switches the high-voltage trans former primary and AC outlets, is energized via a zero-crossing opto coupler manufactured by Motorola. Construction One circuit board (Fig. 4 and Photo 2) contains all the audio circuitry, mother/slave high-voltage regulators and filament regulators. Building the circuit board is straightforward if you follow the stuffing guide (Fig. 5). Double-check the regulators (both high and low voltage), especially for component polarity. You may house this in the container of your choice, with your preferred switching setup. My prototype is housed in an SP-6 shell (Photo 3), which serves my auditioning needs. I have shown one possible switching configuration in Fig. 6. You can cut construction time for the master power supply by using the circuit board containing the fuses, opto-coupler and high-voltage doubler (Figs. 7 and 8). I housed my prototype in a Bud AC-421 measuring 5 by 3 by 9.5 inches (Photo 4). The layout (Fig. 9) is somewhat tight, so configuration. FIGURE 7: Master power supply circuit board pattern. FIGURE 8: Stuffing guide for the master power supply board. you might wish to use something slightly larger. Again, pay close attention to component polarity during board assembly and hand wiring. Initial Power Application Verify master power supply operation first. Access to a minimal 3A Variac (variable AC) supply is helpful but not mandatory. To check the low-voltage dual supply, disconnect the T1 primary at the relay contact (K) and apply line voltage to the master. If you are using a Variac, apply the voltage slowly, noting line current draw. Should this exceed 1A, turn off the power and recheck your wiring. A successful test will result in approximately + 18V at the bridge plus/minus terminals relative to F-. Next test the relay energize circuit by shorting the power (PWR) terminal to the -20 terminal. This should energize the relay coil. Then power down and connect the output of the master to the filament regulators on the main amplifier/regulator board. Again, power up slowly and check for 6.3V at the output of each regulator (across C120 and C220). Power down. Next test the high-voltage master. Photo 4: As you can see, the master power supply layout is a bit tight in this Bud 4.421 shell, even with the circuit board removed. FIGURE 9: Layout of the author's master power supply in a Bud AC-421 5-by-3-by-9.5-inch enclosure. Connect the T1 primary to terminal T2 of the circuit board. Then connect a 4k, 25W power resistor across the high-voltage (HV) and ground (GND) terminals and place a DC voltmeter across this resistor. If you are using a Variac, bring the voltage up slowly, monitoring the AC line current. Any reading of more than 1A indicates a problem. If you encounter this, double-check your wiring. If you are not using a Variac, the fuse will be your indicator. If the unit passes these tests, check for approximately 300V across the 4k resistor and power down. Disconnect the 4k resistor from the master and connect HV and GND to their respective terminals on the amplifier/regulator board. Power up (slowly with the Variac, looking for an AC current less than 1A) and monitor the voltage at the output of the mother high-voltage regulator. This should be 265V + 10V. If the voltage is outside these limits, bring it into range by adjusting the value of R8. This has been specified in the schematic as 75k, 2W. To lower (or raise) the output voltage, lower (or raise) this value. You should not have to deviate more than 50k in either direction. Remember to power down when changing this resistor. Once you have attained 265V, make sure that each slave's output is 15V less than the mother's output. If there is a malfunction in any slave, chances are none will operate correctly. If you encounter a malfunction, remove all op amps and series pass transistors (use a solder wick) and insert the slaves one by one until you find the culprit. Reconnect the T1 primary to the original relay con tact. This completes the master sup ply and mother/slave regulator verification. Unless you want to trim the RIAA equalization, your work is finished. You may alter C114 and C214 slightly to compensate for component variables. Select this value to attain an attenuation of -13.7dB at 10kHz. I have found that most failures and noise problems occur within the first ten hours of operation, so you might want to let the unit burn in on the bench. If you are like me, however, curiosity will prevail, and you will want to try out your new preamp right away. ------------------- Also see: A DIGITAL NOISE GENERATOR, By Benjamin L. Poehland
|
Prev. | Next |