Vacuum Tube Microphone Preamplifier

Designed by: Douglas Mann

Mentor: Dr. Paris Wiley

Fall 2007

Abstract The microphone preamplifier is considered by most recording engineers to be the most

influential piece of recording equipment in the studio. Preamplifiers designs either strive to

transparently amplify the recorded signal or to add sonic “color” to the signal. Depending on the

recording application, the preamplifier sonic characteristics can vary immensely. In some

applications, such as rock music, distortion can be seen as favorable. In an effort to design a

“colorful” preamplifier, vacuum tubes will be used to implement a design capable for use with

microphones typically found in professional studios. To facilitate compatibility with various

microphones, a switching network is designed in conjunction with commonly seen features as

phantom power, 20dB attenuation, and phase reversal. The design requires the tube operation in

the non-linear region in order to take advantage of the desired distortion characteristics. Three

twin-triode devices will be used to provide gain, requiring the design of a DC power supply

capable of approximately 250 volts and a DC filament heater providing 6.3 volts.

TABLE OF CONTENTS

Introduction....................................................................................................................................................2

Purpose / Background ................................................................................................................................2 Specifications / Requirements....................................................................................................................3 Goals ..........................................................................................................................................................4

Theory ............................................................................................................................................................6

Triode Common Cathode Circuit Components .........................................................................................7 Biasing the Tube ........................................................................................................................................8 Voltage Gain Calculation...........................................................................................................................9 Frequency Response ................................................................................................................................10 Frequency response due to input circuit ..................................................................................................13 Simple Common Cathode PSpice Simulation .........................................................................................14

PSpice Simulation ........................................................................................................................................15

Simulated Results.........................................................................................................................................17

Measured Results .........................................................................................................................................17

Harmonic Distortion ................................................................................................................................19 Frequency Response ................................................................................................................................19

Implementation ............................................................................................................................................20

Prototyping...............................................................................................................................................21 Printed Circuit Board ...............................................................................................................................21

Cost Analysis ...............................................................................................................................................23

Societal/ Environmental Impact...................................................................................................................25

Conclusion and Recommendations..............................................................................................................25

References....................................................................................................................................................27

Appendices...................................................................................................................................................27

Appendix A – Datasheets.........................................................................................................................27

12AXY Datasheet ................................................................................................................................28 12AU7 Datasheet .................................................................................................................................29 12BH7 Datasheet .................................................................................................................................30 Jensen 10K61M Output Transformer Datasheet..................................................................................32 Jensen 115KE Input Transformer Datasheet........................................................................................34 Hammond 269JX Power Transformer Datasheet ................................................................................36

Appendix B – Pspice DC Bias Simulation...............................................................................................37

1

Introduction

In an effort to facilitate the design of a vacuum tube microphone preamplifier, research has been

directed toward the individual components and typical design requirements inherent to a high quality

microphone preamplifier. The summary of this project research will be begin with a discussion of the

reason for the design, followed by the definition of specific requirements, and conclude with a complete

tube amplifier design.

Purpose / Background

Recording music has always been of particular interest to me, and even more so, the equipment

used to do so. Every recording engineer has his favorite “tools” in his recording toolbox, ranging from

filters, compressors, limiters, and delays. Depending on the desired effect, these tools can be used to

construct a masterpiece that is greater than the sum of its parts. Any recording engineer will tell you that

the most important piece of equipment in his recording chain is the preamplifier. Regardless of how good

the musician sounds live, the recording is limited to the sound captured by the microphone. The

microphone signal is the purest representation of what is actually being heard, and the preamp is what

makes this signal large enough for recording. The accuracy of the recording is determined by the

preamplifier. A noisy preamplifier can make even the best musician sound bad on a recording. When

accurately representing the music is essential, as in classical music recording, a transparent preamplifier is

desired. Transparent amplifiers are designed to add as little distortion to the signal as possible. In

contrast, the preamplifier can add desirable distortion characteristics that give the music a certain sonic

signature. For example, the Telefunken V-72 preamplifier used for the classic “Beatles” sound can still

be found in many recording studios today. The V-72 is characterized as having a “warm” sound, which is

desired in many recording applications. Many of the “classic” sounding preamplifiers of yesteryear are

still sought after for the character of the distortion introduced to the sound. In the early days of the

microphone preamplifier, the only devices available to produce gain were vacuum tubes. These tubes are

non-linear and create distortion of the original signal, which can actually prove desirable to many music

listeners. In the age of modern electronics, vacuum tubes have been considered obsolete for almost all

applications, with exception to audio electronics due to this desirable tube distortion. The intent of this

2

specific design project is make use of tube distortion while maintain a predominantly “clean” signal,

capable of withstanding the scrutiny of a trained recording engineer. It becomes apparent that the design

of a microphone preamplifier requires the understanding of both electrical engineering design

requirements as well as artistic recording applications.

Specifications / Requirements To make use of a microphone signal, it must be amplified to a level normalized for the input of the

recording interface. Typically this level is referred to as “line” level, which is approximately one volt

rms. Depending on the sensitivity and output impedance of the microphone used for recording, the signal

can vary quite significantly. All studio microphone preamplifiers account for this range of input signals

in the input stage with switching networks appropriate for the type of microphone to protect the

preamplifier and produce the desired result. Certain condenser microphones have output impedances as

low as 30 ohms, generating preamplifier input signal amplitude of approximately .1mV. With this being

the lowest signal extreme, the design requirement of 60dB gain is necessary to get to the desired “line”

level. At the other extreme, certain high sensitivity microphones can produce signal as large as 2.5 V. To

protect the preamplifier circuitry and condition the signal appropriately, attenuators or pads need to be

added to the input stage of the amplifier. Typically a pad of -20dB is suitable for this application. This is

typically labeled as a “High Z” switch on the front panel of the preamplifier. It is left up to the recording

engineer to understand the level of the microphone used and the appropriate switch setting for that

particular microphone. Many condenser microphones require DC power for operation, known as

“phantom power”, which is supplied as 48V across pins 2 and 3 of a standard XLR microphone cable.

Unfortunately, phantom power can very easily damage ribbon microphones, so it is essential for the user

to make sure this is enabled appropriately. To make use of the preamplifier in a multi-microphone

recording environment, recording engineers often require “phase inversion” to avoid undesired

cancellation that occurs in some microphone configurations. This is easily accomplished by switching the

input terminals to the input transformer, which will be needed to appropriately match differing load

requirements and decoupling the stages. This concludes the “must have” requirements for a serious

preamplifier design. Additional features such as metering, multiple channels, and equalization are also

found in many commercially available preamplifiers. The performance can also be improved

significantly by using input and output transformers specifically designed for audio applications. Benefits

3

include optimal circuit impedance matching, isolation, optimized noise performance, and improved

frequency response. Audio transformers have been researched to great detail, but detailed discussion will

be omitted from this summary due to direct relevancy.

Goals Low Noise

In order to preserve the signal integrity, it is essential to minimize noise. High quality Vishay ™ metal

film resistors are used exclusively for their low thermal noise characteristics. Due to the small

microphone output signal, the contribution of thermal noise can be quite significant. Also, gold plated

switches and jacks should be used to maintain minimal noise value. With the chosen cascode

configuration, any fluctuation in the DC anode voltage will be seen in the output as noise. This requires a

very clean, low ripple DC supply for the anode voltages. In addition, the conventional AC 6.3 V heater

voltage is considered a large source of noise, whereas using a DC heater supply will improve the signal to

noise ratio.

Balanced Input and Output

Balanced interfaces make use of differential signals in order to improve signal to noise ratio. Typical

microphones use a 3 pin XLR connector, which transmits the differential signal across pins 2 and 3, with

pin 1 grounded to the chassis. With a well matched input network, any noise picked up in the cable

across the transmission line will be rejected, since it is common to both sides of the network.

High Gain ( > 60 dB)

Depending on the microphone output impedance and sensitivity, the output signal can vary in the range of

millivolts to greater than to volts. Typically, the output is on the order of 10 mV. This signal must be

amplified to reach a recording “line” level value of 0.775V rms. For a 10mV signal, at least 40 dB of gain

is needed to get to line level. In order to accommodate the complete range of microphone signals, the

typical gain requirement is 60dB. This requirement will be imposed on the design.

Minimize Capacitors In Signal Path

In an effort to preserve signal integrity, minimal capacitors will be used in the signal path. Capacitors will

affect the frequency response of the amplifier and shall be used sparingly.

High Common Mode Rejection Ratio

4

As previously mentioned, a balanced differential signal allows for common mode rejection. In order to

achieve good common mode rejection, it is essential that the network be comprised of resistors that are

well matched. Vishay metal film resistors rated at 1% or better will be used to realize this network.

Clean DC Supply Power For Anode

It is essential to have minimal noise in the power supply to reduce the noise in the output. The power

supply shall make use of large smoothing capacitors in the hundreds of microfarads range in addition to a

large inductor to only allow the DC component to pass.

DC Supply For Filament Heater

The filament heater is powered by either an AC or DC supply to allow the tube to operate correctly.

Typical tube amplifier mains transformers have a 6.3 AC secondary with a current output of several amps.

When AC filament supply is used, the 60 cycle frequency current flow introduces noise that can be seen

in the signal output. DC filament supply will be used to minimize this effect.

ADDITIONAL FEATURES

The following features are to be added to improve the functionality and usability of the preamplifier.

Professional Quality Jensen Input & Output Transformers

Jensen transformers are well known in the audio design industry for their flat frequency responses.

Jensen input and output transformers will be used in this design for that purpose. Specifically, the input

transformer is designed to step up the microphone voltage by a 1 to 5 ratio. The output transformer is

slated for tube output applications.

Signal Polarity Inversion Switch

Depending on a microphones location relative to another microphone, the recording engineer may want to

reverse the phase polarity. A switch will be included to swap the input terminals to the transformer.

+48V Phantom Power Supply Voltage

Some condenser microphones require a 48 volt potential on pins 2 and 3 relative to pin 1. The is known

as phantom power and will be developed by a voltage divider at the power supply.

20 dB Pad

Some high impedance microphones can produce very large output signals. A 20 dB pad will be in place

to allow for attenuation of these “hot” signals.

5

Due to time limitations on the scope of the project, the following features were omitted but would make

the unit more marketable.

VU METER - Analog VU meter to enable viewing of output signal amplitude.

OUTPUT ATTENUATOR – Allows the operator to overdrive the amplifier, but still reduce the output to

the nominal line level.

INSTRUMENT INPUT – Allows proper impedance matching for higher impedance inputs from guitars

and basses.

3 BAND EQUALIZER – Low, mid, high frequency control.

HEATER STANDBY – In order to prolong the life of the tubes, it is important to them to warm and cool

slowly. A switch could be added to pre-warm the tubes before applying the plate voltage.

120 Hz HIGHPASS – When the recording application is not dependant on low frequency content, a high

pass filter could be enabled, preventing any power supply ripple noise to be heard in the output.

Theory In order to implement an amplifier design using vacuum tubes it is necessary to understand the

tube operation of both the DC and AC signals comprising the composite signal to be amplified. DC

values will be notated using capital letters and AC as lowercase. By using coupling capacitor we can

assume that only the AC component of the composite signal will enter the circuit, which is not completely

accurate due to DC components generated by non-linearities not addressed in this discussion. To elucidate

the behavior of a common cathode amplifier, the following circuit in Figure 1 will be used for reference.

6

Rp

Rk

Rg

R1

Ci

Co

Ck

0

Vbb

OUT

IN

Figure 1 -Common Cathode Amplifier

The tube used in the design is a three element device known as a “triode”. The elements are the cathode,

anode (plate), and grid, which can be considered analogous to the source, drain, and gate of a n-type

MOSFET transistor respectively in circuit modeling.

Triode Common Cathode Circuit Components

Input Coupling Capacitor (Ci)- This capacitor isolates the grid circuit from the DC voltage output from

the previous circuit. In conjunction with the grid resistor, the input coupling capacitor has impact on the

frequency response of amplification stage.

Grid Resistor (Rg)- Used to provide a voltage to the grid of the triode. For this particular design the grid

bias voltage is zero. Rg usually set to a high value, 1 Meg Ω for example. The grid resistance contributes

to the input resistance of the amplifier.

Cathode Resistor (Rk)- The cathode resistor is used to develop the bias voltage. As current flows through

the cathode resistor, a positive voltage is developed at the cathode. Because the grid is at zero volts, the

grid to cathode voltage is negative with respect to the cathode, setting the bias operating point. The value

of the resistor controls the headroom, or amount of output before clipping occurs of the stage. In order to

minimize distortion, a suitable operating bias point is essential. As the bias point is shifted, clipping will

7

occur at either the top or bottom of the waveform depending on the direction of the shift. If the cathode

capacitor is omitted the cathode resistor has an effect on the stage gain.

Cathode Capacitor (Ck)- Also known as the bypass capacitor, the cathode capacitor short circuits the

cathode resistance to ground for AC signals. Omitting Ck maintains negative feedback, decreasing the

gain of the stage. Ck should be large relative to Rk in order to minimize shelving, where there is a boost

in the high frequency gain. Ck in conjunction with Rk also effect the breakpoint frequency response

location.

Plate Load Resistor (Rp)- The output signal voltage is developed across the plate load resistor due to the

plate current flowing through it. Therefore it has a direct correlation to the gain of the stage. The plate

load resistor also contributes to the output impedance.

Output Coupling Capacitor (Co)- Similar to the input coupling capacitor, Co isolates the circuit from the

DC voltages of the next stage. The frequency response is also affected by the combination of the output

coupling capacitor and the next stage.

Load Resistor (R1)- The output signal is taken across this resistance. If used in multistage amplifier it is

usually the grid resistance of the next stage. R1 controls the mid-band gain of the amplifier since for AC

analysis, the effective plate resistance is the parallel combination of R1 and Rp. It should be noted that if

R1 is an order of 10 larger it can be generally ignored in gain calculations.

Biasing the Tube

Cathode biasing is preferred over fixed biasing because it is largely self compensating. To determine the

cathode bias point and select the appropriate cathode resistor to get desired plate current, load line

analysis is used. On the plate characteristics curve distributed by the tube manufacturer, a line is drawn to

determine the available operating points. The load line value should be selected to be at least two times as

large as the internal plate resistance, but larger values can allows for larger voltage swings and gain. In

this analysis, the value of 50k ohms is selected for a load line. To draw the line, the value of the Vbb/RL

8

is plotted on the vertical axis and Vbb on the horizontal axis. For a particular grid voltage line we find the

intersection of this line with the load line to determine the plate current. Using the load line drawn on the

attached RCA 12AT7 plate characteristics plot, we can select an operating point of -1.5 volts grid voltage

giving approximately 2.5 milliamperes of plate current at a quiescent plate voltage of 125 volts.

The appropriate cathode resistance can be selected using ohms law, where Vg is the voltage of the grid at

the bias point with respect to the cathode. Ip is the current through the plate, which is also approximately

the cathode current because the elements are in series.

1.5 6002.5

gK

p

V VRI mA

= = = Ω

Voltage Gain Calculation To calculate the voltage gain of the amplifier it is necessary to analyze what components contribute to the

gain. An equivalent small signal model can be used for this analysis.

It should be noted that this analysis is assuming a linear small signal, the model cannot be expanded to

large non-linear signal analysis. The triode dependent current source model is shown below, where gm is

the mutual conductance (transconductance), rp is the internal plate resistance, and μ is open loop gain.

Dependent current source small signal model for triode

The values for gm, rp, and μ are conveniently provided in the manufacturer’s datasheets. For a particular

operating point these values can be determine by the intersection of the plate current lines and the

9

respective gm, rp, and μ lines. This determination is shown in the attached datasheet found at the end of

the document. These values are approximately 2000 μ mhos, 23.5 k ohms, and 45 respectively.

The voltage gain is expressed as the output voltage divided by the input voltage. Using the small signal

model shown below, we can derive an equation for the voltage gain in terms of resistances and

transconductance.

( )

( )

( )

p m gk

out p P p

out m gk P p

in gk

outv m P

in

i g v

v i R r

v g v R r

v v

vA pg R rv

= −

=

= −

=

= = −

Using the values determined above we expect a voltage gain equal to:

.002(50 *23.5 ) 31.97outv

in

vA k kv

= = − Ω Ω = − V/V = 30dB

The negative sign denotes a change in polarity.

Frequency Response

10

As mentioned before the frequency response is largely impacted by the capacitive elements of the

amplifier. In addition these components it is necessary to discuss the effect of the parasitic capacitances

of the triode. These capacitances are shown below.

CGA

CGK

CAK

Anode

Cathode

Grid

Parasitic Capacitances Cga, Cgk, Cak

To further understand the effect of these parasitic capacitances the small signal model is useful. Using the

knowledge of these capacitances integrated into the small signal model, any amplifier stage can be

analyzed.

Small signal model with parasitic capacitances

To analyze an amplifier stage, the following process is performed:

-Short coupling and bypass capacitors

-Short the power supply to ground

-Replace tubes with small signal model

-Reduce series and parallel passive device combinations

11

-Perform analysis and compute desired parameters

It is important to realize that this process is done for analysis, and not to done actually performed in

practice.

After performing the procedure the new common cathode amplifier can be modeled as:

For the common cathode amplifier configuration, the input to output impedance from grid to anode (Cga)

dominates the frequency response and the other parasitic capacitances can be ignored for the most part.

Using Miller’s theorem, a new model is developed by modeling an equivalent capacitor in shunt with the

input signal, which is multiplied by the stage gain. The new equivalent model is shown below. This

model only applies to common cathode configuration.

Frequency dependent small signal model for CC amplifier

12

The value for Cga can be obtained directly from the manufacturer’s datasheet under the title “grid to

plate” capacitance. For the 12AT7, Cga is 1.5 picofarads.

Frequency response due to input circuit

Low-Frequency Response

The low frequency response is controlled by Ci and Rg, acting as a high pass filter. The 3dB low

frequency response is calculated using:

31

2Low dBi G

fC Rπ− =

Using an input capacitor of 0.022uF the lower -3dB point is:

3 6

12 (1*10 )(0.022*10 )Low dBfπ− −= 6

pF

= 7.23 Hz

High Frequency Response The high frequency response is due to the output resistance of the previous stage and the input

capacitance of the current stage. As shown above in the frequency dependent small signal model, the

input capacitance value is determined using the equation:

*( 1)in GK GP VC C C A= + +

2.2 1.5 *(31.97 1) 51.65inC pF pF= + + = The output resistance of the previous stage and the equivalent input capacitance calculated above create a

low pass filter which is located at:

31

2High dBin out

fC Rπ− = ,

where Rout is the equivalent output resistance of the previous stage. Assuming we have a series input

impedance of 100k from the previous stage, the equivalent Ro is (100K || 1 Meg) = 90.91K, giving a

upper 3db frequency of:

31 33.9

2 (51.65 )90.91High dBf kHzpF Kπ− = =

13

This gives a bandwidth of 33.9k – 7.23 = ~33.9kHz.

Typically, large resistive dividers are bypassed with small capacitors to boost the frequency response in

compensation for the high frequency roll off of the previous stage due to input capacitance and output

resistance of previous stage.

Output Impedance (Bypassed cathode, output taken from plate) The output impedance is the parallel combination of the plate load resistance Rp and the internal plate

resistance rp.

Frequency response due to output circuit Low frequency response The low frequency response of the output circuit is dictated by the output capacitance and output

impedance acting as a high pass filter for of the stage. The equation is as follows:

31

12 ( )Low dB

out o

fR R Cπ− =+

In the design example the output resistance is (50K || 23.5K) = 16K ohms. Suppose the grid resistor of

the next stage is 1 Meg ohm. Using a 0.0022 uF ouput capacitor, the -3dB point will be:

31 71.2

2 (1 16 )0.0022Low dBf HzMeg K uFπ− =

+= , indicating a drop at 72.1 Hz of -3dB at a slope of -20dB

until the input low response location (7.23Hz), where the slope would increase to -40dB.

Simple Common Cathode PSpice Simulation By importing the tube models developed by Normam Koren, PSpice can be used to analyze the DC and

AC operation of the amplifier. A reference to this tube library is included in the references section. It

should be mentioned that the pspice simulations are based on models that are imperfect, which can lead to

differences in gain and expected frequency response. For example, in the next page, the average plate

characteristics are shown for both the 12AT7 pspice model and actual tube results at a grid voltage of

14

zero. The pspice tube model uses a linear function to represent these characteristic, whereas it can be seen

that the actual characteristic is non-linear. At small plate voltages, the approximation is fairly accurate,

but as the plate voltage is increased the model becomes increasing more inaccurate. It does act as a

valuable tool to analyze the expected effect of changes to the design and a reasonable approximation of

frequency response. If the reader would like more information on how these tube models are

characterized, reference to a good white paper is included in the references section.

DC Bias Point Simulation

As expected the bias calculations give the expected plate and cathode voltages. The current is exactly

2.5mA as calculated.

Several variations of this circuit have been simulated in PSpice to show the effect on frequency response.

PSpice Simulation To simulate the preamplifier performance, the following schematic was used. It should be noted that the

schematic does not include the input and output transformers used in the actual design. The Jensen 115K-

15

E input transformer has a 1:5 step up turns ratio, allowing more voltage gain. The electronic gain is

measured to be 49 dB from the Pspice simulation, which is a multiplication of 218 times. Coupled with

the input transformer, the gain is five times this value or 1090. This gives a total gain of 61dB, which

meets the design requirement.

0

0

C1

10u

R18

115k

VL3 12AU7

0

U7 12BH7A

R14

330

0

R8

1k

0

VL2 12AX7

0

C2

470uCMAX

C4

220n

R1

220k

R6

1k

R4

150k

C3

470n

V1

250

R16

10k

V6.05Vac0Vdc

R2

1k

R10

1.2k

R19

1k

R15

680

V

R11

1k

VL1 12AX7

0

V

0

R3

470k

VL4 12AU7

U6 12BH7A

0

0

R9

5.6k R13

1Meg

0

R7

1Meg

R5

1.5k

PSpice Simulated Model

By simulating above schematic, the preceding frequency response curve was obtained. The 3dB

bandwidth extends from 1Hz well beyond the audio range up to 100kHz. The actual amplifier design will

fall far short of this performance. The lack of the parasitic capacitances in the PSpice model are most

likely the reason for the unusually large bandwidth. The load is set to the input impedance of the output

transformer which is approximately 10k ohms. The potentiometer equivalent maximum resistance is

represented in this simulation by R18 with a value of 115K ohms.

16

Simulated Results By running the simulation with a logarithmic AC sweep from 1Hz to 100kHz the following frequency

response was obtained.

Frequency

1.0Hz 3.0Hz 10Hz 30Hz 100Hz 300Hz 1.0KHz 3.0KHz 10KHz 30KHz 100KHzDB(V(U7:1)/V(VL2:2))

46.5

47.0

47.5

48.0

48.5

49.0

49.5

(99.431K,46.637)

(1.0000,47.251)

(278.962,49.142)

Frequency Response For PSpice Schematic

Measured Results For the purpose of analysis, a sinusoidal input at 1000Hz with a peak to peak amplitude of 50mV is used

to evaluate gain a distortion. In the figure 1 below, the 50mV input signal is shown with the amplified

output. It should noted that the actual output is inverted due to the operation of the amplifier. It has been

reversed to make comparison to the input. As seen in the plot, at an output near line level (~0.775V

RMS), the signal is undistorted and free from clipping. To make a comparison with the maximum gain,

the potentiometer has to be set at its maximal value of 250k ohms. The plot shown in Figure shows this

maximum output of 30 volts peak to peak. The gain can be calculated as:

3020log( ) 20log( ) 550.05

Output dBInput

= =

17

Which is actually 6 dB lower than expected from the Pspice simulation. The user would most likely not

need this amount of gain to get a line signal. The microphone output signal would have to be

approximately 2mV to require this amount of gain.

Figure 2 - 50mV Input Amplified to Line Output

Figure 3 - 50mV Input Amplified to Line Output

18

The maximum output was measured at 12V RMS. Distortion becomes visible at 3.5V RMS given a 50mV peak to peak input signal.

Harmonic Distortion Using a harmonic distortion measurement device, the following observations were made at various output levels. A 50mV 1000Hz input signal is used and passed through a notch filter to compare the input to the output. It seem like the measurements could actually be better with a cleaner DC power supply. It was observed that the THD percentage improved with the built it 400 Hz high pass filter to block the 120 power supply ripple. Also this measurement improved the instant when the supply was turned off, and a smooth dc output was delivered. After several minutes, the THD measurement increased. This may be an indicator that the power supply is unstable. Due to this drift, measurements were recorded initially and after several minutes.

Output Voltage RMS Initial THD % 2 min THD %

0.775 2.4 3.6 1.5 1.9 1.9 2 1.7 1.7

2.5 2.05 2.05 3 2.75 2.75 4 3.8 3.8

4.5 4.7 4.7 5 5.4 5.4

5.5 6.05 6.05 6 6.7 6.7

6.5 7.5 7.5 7 8.2 8.2 8 9.8 9.8 9 16 16 10 25 25

With the high pass filter in place, the THD was measured at 1.6%, which is pretty good considering no feedback is employed.

Frequency Response By measuring the input and output for various frequencies, while keeping the amplifier gain knob set at line output level for a 50mV input, the following frequency response chart was generated. It should be noted that the gain is approximately 27.5 to amplify the input signal to line output level. Due to the inaccuracy of manually recording this data, the plot is far from smooth. The general trend and bandwidth can be determined from this data.

19

As seen in the PSpice simulation, the bandwidth is suitable for the audio range from 20Hz-20kHz. The data measured give the range to be from 3Hz to 31kHz, which is excellent for an audio amplifier.

Implementation Due to the high voltage requirements of this design and the need to have the tube sockets stationary, the

traditional breadboard approach would not suffice. In order to have a rugged yet flexible prototype the

board and turret lug technique was used to implement this design. This process involved a rough

footprint layout on paper to determine component sizes and lead spacing. Once hole placements were

determined, 3/32 inch holes were drilled to allow for the installation of the turret lugs. With the lugs in

place, the bottom of the lug was flanged using a conical tip by applying high pressure via a drill press on

the top of the board. As seen in the photograph below, this creates a clean layout in which terminals can

be accessed from both the upper and lower sides of the board. A rack mountable instrumentation case

was recovered from the scrap pile outside the circuits lab which made a perfect open case enclosure for

the amplifier. Several conventions were followed in deciding on component placing. First, the inductive

components were maximally spaced from the input of the amplifier. The preamplifier board was designed

to have the input signal enter from the right and exit to the left. The connections were made from point to

point to minimize path length. Capacitors were spaced from heat dissipating components such as tubes

20

and power resistors as best as possible. All grounds were connected directly to the chassis using a

terminal strip. Long wire runs were twisted in order to cancel out any electromagnetic interference.

Prototyping

Printed Circuit Board The need for minimizing path lengths and board real estate led to the design of a printed circuit board

model. Using Eagle Layout Editor, a schematic can be constructed in a similar fashion to that of Orcad

PSpice as shown below. The schematic contains information about the actual component sizes and lead

spacing. Several custom components had to be designed, including the Jensen input transformer. The

complete schematic is shown below, including the power supply, heater wiring, and input network.

21

Eagle Layout Schematic for Input Network, Preamplifier, Heater Wiring, and Power

Eagle allows for the above schematic to be translated into a board design. There are trace optimization

features which can help to keep the path lengths minimal. Much of a good PCB design is dependent on

the placement of the components rather the trace locations. Using a similar style of component placement

used for the board and turret lug approach, the layout below was created. Again, attention was paid to

keeping the capacitors as far as possible from the hot tubes. Also, instead of running wires to the switches

and potentiometer, through hole mount components were selected. This significantly reduces signal path

and makes for easier assembly. Further considerations such as ground and power planes could be

employed in this design, but were required more research on proper implementation. The most noticeable

improvement in using a PCB approach is the size of the preamplifier. The case size could be reduced by

half the size using this approach.

22

Cost Analysis The following parts list corresponds to the parts shown on the included schematic.

Part Description Manufacturer Value Unit Price

Bulk Price

R1 1/4 W Resistor Vishay 200 $0.21 $0.06 R2 1/4 W Resistor Vishay 6.81K $0.21 $0.06 R3 1/4 W Resistor Vishay 6.81K $0.21 $0.06 R4 1/4 W Resistor Vishay 68.1 $0.21 $0.06 R5 1/4 W Resistor Vishay 68.1 $0.21 $0.06 R6 1/4 W Resistor Vishay 169 $0.21 $0.06 R7 1/4 W Resistor Vishay 619 $0.21 $0.06 R8 1/4 W Resistor Vishay 619 $0.21 $0.06

23

R9 1/4 W Resistor Vishay 374K $0.21 $0.06 R10 1/4 W Resistor Vishay 250K $0.21 $0.06 R11 1/4 W Resistor Vishay 1K $0.21 $0.06 R12 1/4 W Resistor Vishay 1K $0.21 $0.06 R13 1/4 W Resistor Vishay 470K $0.21 $0.06 R14 1/4 W Resistor Vishay 220K $0.21 $0.06 R15 1/4 W Resistor Vishay 1.5K $0.21 $0.06 R16 1/4 W Resistor Vishay 100K $0.21 $0.06 R17 1/4 W Resistor Vishay 1M $0.21 $0.06 R18 1/4 W Resistor Vishay 1K $0.21 $0.06 R19 1/4 W Resistor Vishay 1.2K $0.21 $0.06 R20 1/4 W Resistor Vishay 1.2K $0.21 $0.06 R21 1/4 W Resistor Vishay 1M $0.21 $0.06 R22 1/4 W Resistor Vishay 680 $0.21 $0.06 R23 1/4 W Resistor Vishay 330 $0.21 $0.06 R24 1/4 W Resistor Vishay 1K $0.21 $0.06 R25 10W Resistor Xicon 1 $0.55 $0.24 R26 10W Resistor Xicon 1 $0.55 $0.24 R27 10W Resistor Xicon 1.1K $0.55 $0.24 R28 10W Resistor Xicon 1.0K $0.55 $0.24 R29 10W Resistor Xicon 2.0K $0.55 $0.24 C1 Electrolytic Capacitor 63V Vishay 220u E $0.59 $0.46 C2 Electrolytic Capacitor 50V Vishay 470u E $0.77 $0.59 C3 Electrolytic Capacitor 150V Vishay 10u E $2.47 $1.43 C4 Capacitor 250V WIMA 470n $1.20 $0.57 C5 Capacitor 400V WIMA 220n $1.78 $0.84 C6 Electrolytic Capacitor 250V Vishay 22u E $0.82 $0.59 C7 Capacitor 630V Sprague .22u $1.60 $0.82 C8 Capacitor 630V Sprague .22u $1.60 $0.82 C9 Electrolytic Capacitor 350V Sprague 350u $0.95 $0.69 C10 Electrolytic Capacitor 350V Sprague 350u $0.95 $0.69 C11 Electrolytic Capacitor 350V Sprague 350u $0.95 $0.69 C12 Electrolytic Capacitor 350V Sprague 350u $0.95 $0.69 C13 Electrolytic Capacitor 50V Xicon 470u E $0.77 $0.53 C14 Electrolytic Capacitor 25V Xicon 10000u E $2.10 $1.24 S1 SPDT Switch Mountain Switch PHANT $4.36 $2.52 S2 DPDT Switch Mountain Switch MIC Z $4.36 $2.52 S3 DPDT Switch Mountain Switch PAD $4.36 $2.52 S4 DPDT Switch Mountain Switch POLAR $4.36 $2.52 J1 XLR Input Jack Neutrik XLR $4.95 $2.15 J2 XLR Output Jack Neutrik XLR $4.95 $2.15 U1 9 Pin Twin-Triode Electro-Harmonix 12AX7 $11.95 $8.95 U2 9 Pin Twin-Triode Electro-Harmonix 12AU7 $9.95 $7.55 U3 9 Pin Twin-Triode Electro-Harmonix 12BH7 $11.95 $8.95

T1 Microphone Input

Transformer Jensen JT-115K-E $37.91 $31.45

T2 Microphone Output

Transformer Jensen JT-10K61-

1M $68.51 $58.54 T3 Power Transformer Hammond 269JX $35.65 $30.78

24

L1 Choke Hammond 8H $30.40 $24.50 L2 Choke Hammond 6mH $9.60 $7.70

B1 Bridge Rectifier International

Rectifier 600V $0.45 $0.24

Total $269.00 $206.33 The unit price is based on the cost of components for one unit. The bulk cost is based on pricing for

construction of 1000 units. A typical single channel tube preamplifier is priced in the range of $700-$900

dollars, making this unit economically viable if the performance meets the customer’s requirements. This

component price does not include the cost of fabrication, which would add approximately $25.00 dollars

to the price for a double layered printed circuit board design at 1000 units. Also the enclosure, depending

on the type selected could add another $40-60 dollars per unit. This would bring the overall bulk

production cost to approximately $300.00 dollars, which is still reasonable in comparison with the market

price for a similar unit.

Societal/ Environmental Impact The impact of this project on society is minimal. It is geared to meet the requirements of a very specific

group (recording engineers) for a very specific application. It would be nice if it could slow global

warming, but due to the inefficiencies of class A operation, it may have the opposite effect. When

possible, the components selected were ROHS compliant. This means that they use no more than the

agreed levels of lead, cadmium, mercury, hexavalent chromium, polybrominated biphenyl (PBB) and

polybrominated diphenyl ether (PBDE) flame retardants. The high DC voltage needed for operation is

dangerous and could potentially cause serious injury or even death. If the product was actually designed

for manufacture it would be very important to consider proper grounding techniques and make sure the

risk of shorting out to the chassis is eliminated. When replacing the tubes, it is important the user follow

proper disposal guidelines as the materials contained in the tubes can be detrimental to the eniviroment.

Conclusion and Recommendations Conclusion

The design of this tube driven microphone preamplifier has been a valuable learning experience. Many

aspects of the process of bringing a design from conception to market have been exercised. The historical

significance of this design played a large part in the need for developing a design to emulate this sonic

aesthetic. Modern design validation techniques such as PSpice were used to determine circuit sensitivities

25

and verify performance. Component selection and pricing was thoroughly exhausted, to attempt to find

the best component for the application, while minimizing cost. Project management had to always be at

the forefront, making sure that the project would meet the micro deadlines in order to finish before the

end of the semester. The difference between design and implementation became quite apparent when

taking a schematic to a two-dimensional drawing and finally to a mess of solder and wires. I was forced

to consider the wire lengths and grounding points that have been ignored in electronics projects up to this

point. It drove me to explore the world of PCB design, to have some control over all the rats nests and

general layout chaos. Many mistakes were made under the soldering iron, and later corrected under the

oscilloscope. A 3 hour forecast easily became an all night endeavor. Troubleshooting skills were

sharpened, and theoretical concepts such as KCL and KVL were confirmed and put to use. Personally, I

developed a much better implementation philosophy throughout the process; assume nothing, test

everything. In theory, it should be easy to hook up 10 components and turn the power on and go home

with a perfect circuit, but this is hardly the reality. I found that unless I measure the voltages and currents

at each node while sequential connecting components, I was destined for failure. The slow and

methodical approach is one I will employ in future projects and hopefully in industry.

Recommendations

After completing this process, I have arrived at many things that I would have done differently in

the next design. The main problem with my design is the quality of the power supply. I underestimated

the importance of this component in the performance of my amplifier. The ripple was directly responsible

for the noise realized at the output. Regardless of large smoothing capacitors, the ripple was too

significant for clean output. In addition, the mains transformer selected was oversized. At the output of

the AC to DC conversion stage, 360 VDC was available, where only 250 V could be utilized. This meant

dissipation of approximately 3.3 Watts as heat, which is wasteful, but also shortens the life of the other

components in the amplifier. In addition, the power supply proved to be ineffective due its vulnerability

to AC input line fluctuations. Depending on the power outlet, the voltage supplied might be the designed

250V all the way down to 220V. This drift had a direct on the symmetry of the distortion, which is

unacceptable as shown in the previous plot of the distorted output signal. In addition, the heater supply

was far from optimal. The ideal heater supply would slowly warm up using a gradual ramp up in DC

current and turn off in a similar manner. The filaments should be warm before applying the high voltage

anode supply. With the current configuration, both are applied simultaneously. In the next design, I

26

would have a heater pre-warm function, where the filament has its own transformer and switch, which is

enabled before turning on the high voltage supply. This feature is seen is many guitar amplifiers to

prolong the life of the tubes.

In addition, I would like to explore a more efficient design that is more cost effective, and takes up

less space. I would like to try a hybrid design utilizing a similar first stage, but with a low anode voltage

and bias point, with subsequent stages using mosfet transistors for the remaining gain. I would like to

prototype the design completely on a large breadboard, a have the final product made on a printed circuit

board.

My advice to anyone wanting to develop a project similar to this would be to keep the scope small.

Start with one component, measure the resistance, write it down. Add the next component, check the

continuity, measure as much as you can, write it down. Otherwise, you may be are bypassing the

learning process in addition to the cathode resistor.

References [1] F. Langford-Smith, The Radiotron Designer’s Handbook, 4th ed., Sydney: Wireless Press, 1953. [2] G. M. Ballou, Handbook for Sound Engineers, 3rd ed., Oxford: Focal Press, 2005. [3] H. J. Reich, Principles of Electron Tubes, Peterborough: Audio Amateur Press, 1995. [4] Sedra & Smith, Microelectronic Circuits, 5th ed., New York: Oxford, 2004. [5] P. Horowitz, W. Hill, The Art of Electronics 2nd ed., Cambridge: Cambridge University Press, 1995. [6] M. McCorquodale, “Vacuum Tube Preamplifier Analysis and SPICE Simulation”, http://www.eecs.umich.edu/~mmccorq/diversions/simulation/index.html, 2005. [7] W. Marshall Leach, Jr., "SPICE models for vacuum-tube amplifiers," J. Audio Eng. Soc. Vol 43, No. 3, March 1995, p. 117

Appendices

Appendix A – Datasheets

27

28

29

"Ei-RC" - Electronic tubes factory Data Sheet

12BH7 Page 3 of 4

Telephone: +381 18 550 741

Typical characteristics and operating conditions

Class A1 amplifier

Vertical deflection amplifier

Anode voltage Va 250 350 (V) Grid voltage Vg -10,5 (V) Anode current Ia 11,5 16 (mA) Transconductance S 3,1 3,1 (mA/V) Amplification µ 16,5 16,5 Anode resistance (approx.) Ra 5,3 (kΩ) Cathode bias resistor Rk 560 (kΩ)

Peak to peak sawtooth component

Vpp 25 Grid input voltage

Negative peak component Vp 32

(V)

Peak positive pulse component Vp 670

Anode output voltage

Peak to peak sawtooth component

Vpp 230 (V)

Maximum ratings

Class A1 amplifier

Vertical deflection amplifier

Anode voltage Va 300 450 (V) Peak positive plate voltage 1500 (V) Anode dissipation (each section) Wa 3,5 3,5 (W)

Peak negative pulse grid voltage 250 (V)

FAX: +381 18 550 806 Postal address: Bul. Sv. Cara Konstantina 80-86, 18000 Nis, Yugoslavia Electronic mail: eirc@eierc.com Web site: http://www.eierc.com/rc

30

"Ei-RC" - Electronic tubes factory Data Sheet

12BH7 Page 4 of 4

Telephone: +381 18 550 741 FAX: +381 18 550 806 Postal address: Bul. Sv. Cara Konstantina 80-86, 18000 Nis, Yugoslavia Electronic mail: eirc@eierc.com Web site: http://www.eierc.com/rc

Average cathode current (each section) 20 20 (mA)

Peak cathode current 70 (mA) fixed bias 0,25 Grid circuit

resistance cathode bias 1 2,2 (MΩ)

Average plate characteristics 12BH7 (each section)

Plate voltage (V)

0 100 200 300 400 500 600

Pla

te c

urre

nt (m

A)

0

10

20

30

40

50Vg=0V

-5

-10

-15

-20

-25

-30

-35-40

Vf=12,6V

31

LINE OUTPUT TRANSFORMER4:1 CT or 8:1 with FARADAY SHIELD

M Distortion 0.007% typ at 20 Hz and +4 dBu output levelM Wide bandwidth: !!!!3 dB at 0.04 Hz and 60 kHzM Drives 600 SSSS loads to levels up to +23 dBu at 20 HzM Excellent time domain performance: DLP 0.2EEEE typ 20 Hz to 20kHzM Appears as 11 kSSSS load to vacuum tube driver circuitry

This transformer is designed for very high performance vacuum tube line outputstages. Driving signals should be free of DC and source impedance under 1 kS.The split secondaries may be series connected for 4:1 with center-tap, orparalleled for 8:1 operation. A fully enclosed channel frame is standard.

TYPICAL APPLICATION

JT-10K61-1M4:1 CONFIGURATION

32

PARAMETER CONDITIONS MINIMUM TYPICAL MAXIMUM

Input impedance, Zi 1 kHz, 0 dBu, test circuit 3 10.0 kS 11.5 kS 13.0 kS

Voltage gain 1 kHz, 0 dBu, test circuit 1 !13.8 dB !13.4 dB !13.0 dB

Magnitude response,ref 1 kHz

20 Hz, 0 dBu, test circuit 1, Rs=50 S !0.1 dB !0.01 dB 0.0 dB

20 kHz, 0 dBu, test circuit 1, Rs=50 S !0.5 dB !0.37 dB 0.0 dB

Deviation from linear phase (DLP) 20 Hz to 20 kHz, 0 dBu, test circuit 1, Rs=50 S +0.2/!0.1E ±2.0E

Distortion (THD)

1 kHz, +4 dBu, test circuit 1, Rs=50 S <0.001%

20 Hz, +4 dBu, test circuit 1, Rs=50 S 0.008% 0.05%

1 kHz, +4 dBu, test circuit 1, Rs=600 S <0.001%

20 Hz, +4 dBu, test circuit 1, Rs=600 S 0.020%

Maximum output level 20 Hz, 1% THD, test circuit 1, Rs=50 S +21 dBu +23 dBu

Common-mode rejection ratio (CMRR)60 Hz, test circuit 2 114 dB

3 kHz, test circuit 2 70 dB 80 dB

Output impedance, Zo 1 kHz, test circuit 4, Rs=50 S 133 S

DC resistanceprimary (GRN to BLU) 740 S

secondaries in series (ORG to RED) 42 S

Capacitanceprimary to shield and case, 1 kHz 300 pF

both secondaries to shield and case, 1 kHz 300 pF

Turns ratio secondaries in series 4.070:1 4.082:1 4.095:1

Temperature range operation or storage 0E C 70E C

Breakdown voltage(see IMPORTANT NOTE below)

primary or secondary to shield and case, 60 Hz,1 minute test duration

250 V RMS

JT-10K61-1M SPECIFICATIONS (4:1 series secondaries configuration, all levels are output unless noted)

All minimum and maximum specifications are guaranteed. Unless noted otherwise, all specifications apply at 25EC. Specifications subject to changewithout notice. All information herein is believed to be accurate and reliable, however no responsibility is assumed for its use nor for any infringements ofpatents which may result from its use. No license is granted by implication or otherwise under any patent or patent rights of Jensen Transformers, Inc.IMPORTANT NOTE: This device is NOT intended for use in life support systems or any application where its failure could cause injury or death. Thebreakdown voltage specification is intended to insure integrity of internal insulation systems; continuous operation at these voltages is NOT recommended.Consult our applications engineering department if you have special requirements.

JENSEN TRANSFORMERS, INC., 7135 Hayvenhurst Avenue, Van Nuys, CA 91406-3807, USA9/01 (818) 374-5857 FAX (818) 374-5856 www.jensen-transformers.com

33

MICROPHONE INPUT TRANSFORMER1:10 STEP-UP FOR HIGH IMPEDANCE AMPLIFIERS

M Ideal for FET or vacuum tube input amplifiersM Wide bandwidth: !!!!3 dB at 2.5 Hz and 90 kHzM 20 dB of voltage gain with Noise Figure of only 1.5 dBM Input impedance of 1.4 kSSSS for loading loss of 0.9 dBM High common-mode rejection: 110 dB at 60 Hz

This transformer is designed for highest practical step-up ratio. Its secondarysource impedance makes it ideal for use with low noise FET or vacuum tubeinput amplifiers. The primary is fully balanced and its leads may be reversed toinvert polarity, if required. A 30 dB magnetic shield package is standard.

TYPICAL APPLICATION

JT-115K-E

34

PARAMETER CONDITIONS MINIMUM TYPICAL MAXIMUM

Input impedance, Zi 1 kHz, !20 dBu, test circuit 1 1.33 kS 1.40 kS 1.47 kS

Voltage gain 1 kHz, !20 dBu, test circuit 1 19.65 dB 19.75 dB 19.85 dB

Magnitude response,ref 1 kHz

20 Hz, !20 dBu, test circuit 1 !0.50 dB !0.26 dB ±0.0 dB

20 kHz, !20 dBu, test circuit 1 !0.25 dB !0.13 dB +0.1 dB

Deviation from linear phase (DLP) 20 Hz to 20 kHz, !20 dBu, test circuit 1 +3.5/!0E ±5.0E

Distortion (THD)1 kHz, !20 dBu, test circuit 1 0.001%

20 Hz, !20 dBu, test circuit 1 0.065% 0.15%

Maximum 20 Hz input level 1% THD, test circuit 1 !4 dBu !2.5 dBu

Common-mode rejection ratio (CMRR)150 S balanced source

60 Hz, test circuit 2 110 dB

3 kHz, test circuit 2 70 dB 78 dB

Output impedance, Zo 1 kHz, test circuit 1 17.0 kS

DC resistancesprimary (RED to BRN) 19.7 S

secondary (YEL to ORG) 2465 S

Capacitances @ 1 kHzprimary to shield and case 475 pF

secondary to shield and case 205 pF

Turns ratio 1:9.95 1:10.00 1:10.05

Temperature range operation or storage 0E C 70E C

Breakdown voltage(see IMPORTANT NOTE below)

primary or secondary to shield and case, 60 Hz,1 minute test duration

250 V RMS

JT-115K-E SPECIFICATIONS (all levels are input unless noted)

All minimum and maximum specifications are guaranteed. Unless noted otherwise, all specifications apply at 25EC. Specifications subject to changewithout notice. All information herein is believed to be accurate and reliable, however no responsibility is assumed for its use nor for any infringements ofpatents which may result from its use. No license is granted by implication or otherwise under any patent or patent rights of Jensen Transformers, Inc.IMPORTANT NOTE: This device is NOT intended for use in life support systems or any application where its failure could cause injury or death. Thebreakdown voltage specification is intended to insure integrity of internal insulation systems; continuous operation at these voltages is NOT recommended.Consult our applications engineering department if you have special requirements.

JENSEN TRANSFORMERS, INC., 7135 Hayvenhurst Avenue, Van Nuys, CA 91406-3807, USA9/01 (818) 374-5857 FAX (818) 374-5856 www.jensen-transformers.com

35

Hammond 269JX Mains Transformer Datasheet

269JX 50 115 60 250-0-250 60 6.3 V@2.5 A - X5

Mechanical & Schematic Data:

Vertical Mount ("X" mounting)

* NOTE: Filament #1 and #2 windings may not exist or do not have C.T. connections - refer to table

above.

Dimensions (Inches) Mtg. Style A B C D E

G-Mtg. Slot

X2 1.88 2.44 2.5 1.5 1.56 .19 x .31 X4 2.19 2.5 2.63 1.75 1.44 .19 x .25 X5 2.19 2.63 2.63 1.75 1.56 .19 x .25 X6 2.5 2.75 3.06 2 1.69 .203 x .38 X7 2.5 3 3.06 2 1.94 .203 x .38 X7A 2.5 3 3.13 2 2.06 .203 x .38 X7B 2.5 3 3.13 2 2.19 .203 x .38 X8 2.5 3.25 3.06 2 2.19 .203 x .38 X9 2.5 3.75 3.06 2 2.69 .203 x .38 X9A 2.88 3.5 3.5 2.25 2.44 .203 x .38 X9B 3.13 3.5 3.5 2.5 2.38 .203 x .38 X10 3.13 3.5 3.81 2.5 2.19 .203 x .38 X10A 3.13 3.5 3.88 2.5 2.38 .203 x .38 X11 3.13 3.75 3.81 2.5 2.44 .203 x .38 X13 3.75 4 4.56 3 2.81 .203 x .38 X14 3.75 4.5 4.56 3 3.31 .203 x .38 X15 3.75 5 4.56 3 3.81 .203 x .38

36

Appendix B – Pspice DC Bias Simulation

R7

1Meg

79.71VR1

220k

197.3V

R9

5.6k

0

R15

680

U7 12BH7A

0

162.3V

V

0V162.3V

VL2 12AX7

U6 12BH7A

232.2V

0V

4.549uV

VL4 12AU7

1.009mV

0V

R8

1k

1.008mV

0

8.103V

R13

1Meg

R5

1.5k

250.0V

VL1 12AX7

R4

150k

0

R18

115k

R19

1k

C1

10u

C2

470uCMAX

R11

1k

0

162.3V

C3

470n

R6

1k

R3

470k

R14

330

80.63V

4.549V

V610Vdc

0

0

0V

R10

1.2k

00

164.1V

1.992uV

526.9mV

R16

10k

C4

220n

R2

1k

V

0

0

V1

250

79.71V

991.9nV

VL3 12AU7

37