Simple Op Amp Audio Preamps

A Ciphers By Ritter Page

Terry Ritter

2006 April 21

Here are tested designs for simple, low-noise, audio preamps built from modern low-voltage opamps, and designed to satisfy various goals:

• Minimum Components: One opamp package and a few resistors and capacitors.
• Battery Operation: Typically use a single 9V battery down to 4V.
• Reversed Battery Protection: Avoid even transient reversal.
• Low Current: Typically under 2mA.
• High Input Impedence: Opamp input.
• Low Output Impedence: Opamp output with feedback.
• Low Noise: Limited by op amp, cost and availability.
• Audio Bandwidth: Nearly flat from 20Hz to 80kHz.
• Fixed Known Gain and Low Distortion: Here 30dB, typically in 2 op amp gain stages (20dB and 10dB), each set by feedback.
• Cascadable: Connect gain blocks in series; swap to compare noise performance.
• RFI Resistance: Resist radio and shortwave.
• Input Protection: Protect against large input pulses.

Op Amps and Specs

The main issues are voltage, current, noise, and input offset:

• Low Voltage is important if we want to run a 9V battery down to 4V.
• Low current is important if we want the battery to last as long as possible.
• Low noise is important if we want to amplify very low-level signals.
• Offset is less important here.

These are worst-case specs that all marked components should meet; sometimes the actual values will be better or even much better.

             E (V)    I (mA)   nV/SQRT(Hz)   offset (uV)
  TLC272     3..16     3.2        25           10000
  LMC6482    3..15     1.9        37           38000
  LT1884     3..40     2.8         9.5            80
  OP27       8..36    ~6           4.5           100

Feedback Noise

When we use op amps, we add thermal noise from resistors. But the added noise will not matter if we keep it below, say, a third of the op amp noise. Thermal noise varies as the square-root of the product of resistance and bandwidth.

  E(nV/SQRT(Hz)) ~ 0.129 SQRT(R)
  R ~ SQR( E / 0.129 )

     Ohms    nV/SQRT(Hz)   uV RMS (10k BW)   uV RMS (21k BW)
       1M      129.x           12.9              18.9
     300k       70.7            7.07             10.3
     100k       40.8            4.08              5.96
      30k       22.3            2.23              3.27
      10k       12.9            1.29              1.89
       3k        7.07           0.707             1.03
       1k        4.08           0.408             0.596
      300        2.23           0.223             0.327
      100        1.29           0.129             0.189
       30        0.707          0.071             0.103
       10        0.408          0.041             0.060
        3        0.223          0.022             0.033
        1        0.129          0.013             0.019

To get the expected RMS noise, multiply the nV/SQRT(Hz) value by the square root of the measurement bandwidth. Here, "10k" means 10,000Hz (and sqrt(10k) = 100); "21k" means 21,355Hz (and sqrt(21355) = 146).

LMC6482 and TLC272 Preamps

The National LMC6482 and TI TLC272 are dual CMOS low voltage op amps, with somewhat high input noise.

• Battery Protection: 47 ohm R1 protects the battery against accidental short circuits, in the extreme functioning as a fuse.
• Reversed Power Protection: Transistor Q1 is a P-channel power MOSFET, used "in reverse." When the battery is correct, current will flow through the "body diode" D4 inside the MOSFET, even if the transistor is OFF. But when the battery is reversed, D4 is reverse biased, as is the Gate, and no current flows. (See A Modern Breadboarding Technology.)
• Local Bias from LED: The forward drop from red LED D1, running at about 50uA from a 4V battery, makes a low-noise reference of about 1.3V. With a 9V battery this will be closer to 140uA and 1.6V. The board "ground" is the negative supply.

• Power Filter Capacitors: The power bypass cap C1 rolls off battery noise. Bias bypass cap C2 rolls off shot noise produced in D1; this would probably not be an issue except in a very low level preamp. Tantalum capacitors are recommended for their good bypass characteristics through RF frequencies. A common electrolytic with a RF bypass in parallel is usually much worse than commonly supposed.
• Input Blocking: If we do not know the voltage of the output bias from the signal source or previous stage, we must assume that level may be either higher or lower than the desired preamp bias, and any single input blocking cap C3 must be bipolar. A 1uF film is appropriate since, with the 100k load of R3, the input path frequency response extends well below 10Hz.
  • If an even lower frequency response is desired, a larger capacitor can be used, with the associated longer charge time.
  • Since cascaded stages need at most one capacitor between them, I generally do not build C3 on the board, but instead depend upon isolation from the output cap in the previous stage.
  • If we can guarantee some reasonable bias voltage from the signal source, we can use DC coupling by shorting C3.
• Interstage Coupling: The interstage coupling cap C4 and load R6 are the same deal as the input path. These probably could be deleted entirely (by shorting C4 and opening R6 if desired). If we also could use DC input coupling, we might delete the local bias subsystem entirely, including R2, D1, C2, C3, R3, C4 and R6.
• Output Blocking: The effect of output blocking cap C5 depends upon load, and 1uF should be satisfactory for full bandwidth response with a 20k load or higher. Lower impedance loads will affect very low frequency response. R9 represents the external load, and generally is not built on the preamp board.
• Multiple Stages: The goal is 30dB (31.6x) for the board, but that is a little much for full audio bandwidth with most low-voltage op amps. Limiting gain to 20dB allows the op amp to cover the audio range and still have a little negative feedback to reduce its distortion. To get 30dB we need 2 stages, ideally a gain of 20dB (10x) and 10dB (3.1x). The largest gain is used first, which boosts tiny signals above board noise.
• Feedback Resistors: The first stage gain is set by (R4/R5)+1 which is (9.09k/1k)+1 for 10.09x or 20.08dB. The second stage gain is set by (R7/R8)+1 which is (20k/9.09k) for 3.20x or 10.10dB. The total gain should be about 32.29x or 30.18dB, although resistor tolerances will affect that somewhat.
• Feedback Capacitors: The impedance of C6 must be much lower than the 1k R5 at the lowest frequency of interest. Simulation shows that 68uF should be satisfactory. Similarly, the impedance of C7 must be much lower than 9k R8, and simulation shows 6.8uF to be large enough. Larger value caps would allow lower frequency operation at the cost of longer charge times. These caps will be operated at the bias level of 1.3V to 1.6V or whatever the particular LED D1 develops, unless DC coupling is used, and then they will follow the input bias level.
• Tantalum Capacitors: Tantalum capacitors are used in the feedback system because they are small and convenient. Although sometimes criticized for audio distortion, much of that comes from misuse:
  • Tantalum capacitors form a semiconductor layer of tantalum oxide under the desired tantalum pentoxide dielectric, and that makes a diode. To avoid distortion, that diode must be biased so that reverse voltage never occurs, even at signal peaks.
  • Tantalums also put more capacitance into less volume, and as a consequence probably do have more dielectric nonlinearity than ordinary electrolytics. But the resulting distortion is related to voltage variation across the capacitor, which is inherently low for very small signals.
  • Some tantalums are especially sensitive to reverse polarity. Even the process of using an ohm-meter to check floating resistors might produce reverse voltages damaging to some types tantalum capacitor. Know the characteristics of the devices and test equipment you use, and take extreme care with both power and test voltage polarity. Getting installation polarity correct is easy here, because each tantalum negative lead connects to the copper surface ground.
  • A big advantage of tantalum capacitors is their ability to bypass high RF frequencies. Good RF bypassing can be important for some op amps, but the ones used here seem fairly docile.
  Or use ordinary electrolytics.
• DC Feedback: Another possibility is to amplify both DC and AC by connecting R5 and/or R8 directly to the bias supply C2. That would eliminate capacitors C6 and/or C7, but probably would require a capacitor of similar size for C2. With DC amplification, the input offset can affect the output bias level, although a single DC stage may be OK.
• Input Protection: Schottky power diodes D5 and D6 conduct input pulses around the op amp inputs. These are 20 PIV (peak-inverse-voltage) 1A units like 1N5817. Surprisingly, the higher PIV models in the same series provide less protection because they conduct less well at low voltages. The data sheet indicates that for an external voltage pulse to reach 0.6V beyond op amp powering levels, one of the 1N5817 diodes already will be conducting 5 amps. The protection diodes probably do confine reasonable input signal levels to under 100mV, which is far above the expected level for these designs.
• Feedback Storage Protection: The feedback bias cap C6 probably will not charge above about 1.6V unless the input is DC coupled. However, when the battery is disconnected, the stored power may conduct through 1k R5 and the op amp input until discharged. To avoid that we use a Schottky as D2. The 9k R8 probably is enough protection for the second stage.
  

  

• Simulated Frequency Response: The green curve is from the output of U1B before C5, and the yellow curve is the signal across the 20k load R9 after C5. The flag is positioned at 20Hz. The basic response is down about 0.2dB at 20Hz and at about 80kHz.

  The external low frequency response could be improved by increasing the output capacitor C5, or increasing the load impedance, represented by R9.
  

  

  The LMC6482 board is about 1.5" x 2.1" in size. It takes about 1.1mA with 9.5V power. The bias LED is barely lit at 160uA, but clearly visible in the dark, and the bias is about 1.56V.
  

  

  The TLC272 board is about 1.5" x 3" in size. It takes about 1.3mA with 9.5V power. The bias LED is barely lit at 160uA, but clearly visible in the dark, and the bias is about 1.57V. The 100 ohm resistor added in the output path may help reduce ringing with a large capacitive load.

LT1884 Preamp

The LT1884 is a dual bipolar low voltage op amp with lower noise than many CMOS op amps.

• Resistor Values: The stage gain ratios are the same as in the CMOS versions, with much lower values. The lower values produce less noise than the op amp inputs, and so avoid affecting the overall noise value.
• Storage Protection: The lower feedback resistance values make the stored value protection Schottky D5, and now also D6, even more important.
• Input Capacitor: The input DC blocking cap C6 does double duty: First, it conducts signal to the op amp input, but even 1uF is more than enough for that. The second purpose is to conduct bipolar transistor base noise from inside the op amp to be shunted by the signal source. That of course presumes that the source has a low output impedance, since, if not, there is no advantage. The larger the cap the better, but with that comes greater pulse conduction, turn-on delay and power storage. Simulation indicates that 3uF is satisfactory to shunt frequencies above 1kHz. Those represent about 95 percent of the bandwidth (and white noise energy) from 20 to 20kHz. That should provide more than a 20dB (10x) reduction for transistor base noise, again assuming any reduction is to be had.
  

  

• Simulated Frequency Response: The green curve is from the output of U1B before C5, and the yellow curve is the signal across the 20k load R9 after C5. The flag is positioned at 20Hz. The basic response is down about 0.6dB at 20Hz and at about 80kHz.

  The basic low frequency response could be improved by increasing the value of the feedback bias caps C3 and C7. The external low frequency response could be improved by increasing the output capacitor C5, or increasing the load impedance, represented by R9.
  

  

  The LT1884 board is about 1.5" x 2.1" in size. It takes about 1.7mA with 9.4V power. The bias LED is barely lit at 160uA, but clearly visible in the dark, and the bias is about 1.56V.

OP27 Preamp

The OP27 is a bipolar op amp with even lower noise. Unfortunately, it also requires more supply voltage, more supply current, and a higher input bias voltage. Fortunately it has higher gain than the other op amps, so that a single stage in the package can produce 30dB of amplification. Unfortunately, it is also obsolete.

• Feedback Resistors: The gain is set by (R7/R6)+1 which is (3010/100)+1 for 31.10x or 29.86dB.
  

  

• Simulated Frequency Response: The green curve is from the output of U1 before C4, and the yellow curve is the signal across the 20k load R8 after C4. The flag is positioned at 20Hz. The basic response is down about 0.5dB at 20Hz and about 80kHz.

  The basic low frequency response could be improved by increasing the value of the feedback bias cap C3. The external low frequency response could be improved by increasing the output capacitor C4, or increasing the load impedance, represented by R8.
  

  

  The OP27 board is about 1.5" x 2.1" in size. It takes about 3.3mA with 18.1V power. The dual bias LED's are dim but visible at 300uA, and the bias is about 3.19V.

Terry Ritter, his current address, and his top page and electronics home.