Hybrid Audio Frequency Amplifier

Transcription

1 Hybrid Audio Frequency Amplifier Kijun Yoo Robert Hasselle Faculty Mentor: Ender Ayanoglu

2 Abstract As technologies have improved, many audio amplifier manufacturers have transitioned from vacuum tubes to solid state circuits for their convenience and affordability. However, in recent years, there has been an upward trend in the demand for vacuum tubes in audio amplifiers for their superiority quality in audio amplification. The vacuum tubes, similar to MOSFET transistors, have multiple methods in which they can be biased which change the circuit topology and the sound that is being amplified. Therefore, we tested multiple circuit topologies including single MOSFETS, cascoded JFETS, and a single current regulating diode as possible biasing methods for the tube in our hybrid audio amplifier. We compared each topology using a program called Right Mark Audio Analyzer in order to see the differences in noise, distortion, dynamic range, and total harmonic distortion. We used the program results to see if there were any audible differences from each circuit topology and also considered breakdown voltage, power supply rejection ratio and current variation in determining the best biasing method. After analyzing the data, we concluded that the single MOSFET configuration had the best overall qualities for biasing the tube in our hybrid audio amplifier. 2

3 Table of Contents Illustrations...4 Introduction....5 Procedure....7 Results 12 Conclusion..20 Work in Progress 21 Bibliography 23 3

4 Illustrations Figure 1 - Schematic of the SOHA amplifier....6 Figure 2 PSPICE model of the amplifier...7 Figure 3 PSPICE model of the amplifier with voltage and current values... 8 Figure 4 Frequency response using PSPICE Figure 5 Completed circuit Figure 6 Block Diagram of our Test Setup.11 Figure 7 Circuit topology of different CCSs...11 Figure 8 Overall results of a MOSFET CCS.12 Figure 9 Frequency Response of a MOSFET CCS.12 Figure 10 Noise of a MOSFET CCS...13 Figure 11 Dynamic Range of a MOSFET CCS.13 Figure 12 Total Harmonic Distortion of a MOSFET CCS 14 Figure 13 Overall Results of JFET CCS 15 Figure 14 Frequency Response of a JFET CCS..15 Figure 15 Noise Level of a JFET CCS 16 Figure 16 Dynamic Range of a JFET CCS 16 Figure 17 Total Harmonic Distortion of a JFET CCS 17 Figure 18 Overall Results of CRD CCS..18 Figure 19 Frequency Response of a CRD CCS...18 Figure 20 Dynamic Range of a CRD CCS..19 Figure 21 Total Harmonic Distortion of a CRD CCS.19 4

5 Introduction Vacuum tubes have been first developed in the early 20 th century by scientists John Fleming who invented the vacuum tube diode, while Lee De Forest discovered the triode tube. These were primitive models that did not use true vacuums. The scientists who really brought the significant modifications were Eric Tigerstedt and Irving Langmuir by creating a stronger vacuum which improved the amplifying characteristics of a tube. These modified tubes became known as the first vacuum tubes. The tubes became refined year after year and were used in various commercial areas including radio broadcasting, television, telephone networks and high quality sound reproduction. In the 1940 s, transistors were first developed and began to replace the vacuum tubes. Transistors had several advantages over the vacuum tube; transistors generate less heat, are smaller, have greater efficiency and better reliability. With the overwhelming advantages over the vacuum tube, transistors started to replace tubes in all electronics devices, including high fidelity audio amplifiers. But after the transition, many people preferred the warm sound of the vacuum tube so they reverted back to using vacuum tubes instead of transistors for audio frequency amplifiers. In our project, we tried to combine the best of both technologies; the amplification is implemented by the tube to provide a better sound while other parts of the circuit use solid state components. Solid state components make the amplifier smaller, consume less power (such that it can run on a battery) and cheaper to build due to the price and affordability issues of tubes. As engineers, we needed to consider beyond the quality of the product which means that we need to consider the budget as well as the sound quality. Since most tubes cost a substantial amount as compared to solid state components, we decided to make a hybrid amplifier in hopes to create an amplifier that is proficient in both sound quality and cost. 5

6 Figure 1 - Schematic of the SOHA amplifier The circuit shown in Figure 1 is going to be the one that we are going to build. The two MOSFETS (LND150) and the resistors R7, R8, R17, R18 are known as the constant current source (CCS) that biases the tubes. A fixed amount of current needs to go through the tube and we are going to measure how changing the circuit topology of the CCS can affect the output sound s noise, distortion, dynamic range and total harmonic distortion. 6

7 Procedure First we had to analyze the circuit and understand how the circuit works in order to perform our design project. So we did some PSPICE analysis with a 12AU7 tube model. The PSPICE schematic for our circuit is shown in Figure 2. We obtained the frequency response of the circuit for the audible range (20Hz to 20 khz) and experimented with the circuit to determine the purpose of various circuit components including the unity gain buffers and the diodes. The resulting frequency response simulation is shown in Figure 3. Figure 2 PSPICE model 7

8 Unity Gain Stage: Because the vacuum tube alone provides sufficient gain for headphones, the unity gain op-amp simply acts as a current buffer. The current buffer is needed because too much current into a headphone can destroy it. In addition, diodes are connected near the input of the op-amp to ensure that the input voltage to the op-amp does not exceed the tolerance which in this case is +0.7V for V+ and -0.7V for V- (spec sheet). When the input voltage swings above or below 0.7V of the supply voltage, the diodes will turn on and guide the excess current away from the op amp and into the heat sink. Figure 3 - Frequency response 20Hz-20kHz 8

9 Figure 4 PSPICE model with voltage and current values Constant Current Source (CCS) Alternatives: A constant current source is needed in this circuit to bias the vacuum tube. Although the CCS can be implemented using a variety of methods, several things must first be taken into consideration. First of all the CCS has to be reliable which means the breakdown voltage must not be too low or else some voltage variation can cause failure. Secondly, it must be able to proficiently reject the ripples from the power source because the power source supplies a significant amount of noise. Current variation can also have a negative impact on distortion of the sound because the higher the variation in the current that enters the tube, there will be more distortion and noise. Therefore, the ideal CCS would be to have very low current variation and very high PSRR. In the original circuit a single MOSFET is used as the CCS. Other alternatives that we considered are cascoded BJTs, cascoded JFETs, a current regulating diode (CRD), and cascoded BJTs with a CRD. To be able to compare the effectiveness of the different CCSs, three criteria were taken into consideration PSRR, breakdown voltage, and current variation. Breakdown voltage is a manufacturing parameter while PSRR and current variation were measured via PSPICE simulation. 9

10 Topology Breakdown Voltage (V) Power Supply Rejection Ratio Current Variation (ua) Single MOSFET (LND150) Cascoded JFETs (J113) Cascoded BJTS (PN2907A) Cascoded BJTs with CRD Current Regulating Diode (1N5297) 500V -68 db V -73 db V -34 db V -55 db V -53 db 6.6 After we did all the PSPICE analysis, we built the circuit in which we could change constant current sources. The completed and fully functional circuit is shown in Figure 5. From there, we hooked it up in a feedback loop shown in Figure 6. And then we used a program called Right Mark Audio Analyzer to get the graphs for noise, distortion, frequency response, dynamic range, etc for each corresponding constant current source. Figure 5 Completed circuit Figure 6 Block Diagram of our Test Setup SOHA Amplifier Computer Output Input Left Right Left Right 10 Speaker Output

11 Figure 7 Circuit topology of different CCSs Results RMAA Test results for a LND150 MOSFET CCS: Figure 8 Overall results of a MOSFET CCS 11

12 Figure 9 Frequency Response of a MOSFET CCS 12

13 Figure 10 Noise of a MOSFET CCS Figure 11 Dynamic Range of a MOSFET CCS 13

14 Figure 12 Total Harmonic Distortion of a MOSFET CCS 14

15 RMAA Test results for a cascoded JFET CCS: Figure 13 Overall Results of JFET CCS Figure 14 Frequency Response of a JFET CCS 15

16 Figure 15 Noise Level of a JFET CCS Figure 16 Dynamic Range of a JFET CCS 16

17 Figure 17 Total Harmonic Distortion of a JFET CCS 17

18 RMAA Test results for a CRD CCS: Figure 18 Overall Results of CRD CCS Figure 19 Frequency Response of a CRD CCS 18

19 Figure 20 Dynamic Range of a CRD CCS Figure 21 Total Harmonic Distortion of a CRD CCS 19

20 Conclusion From our results, we found out that the single MOSFET CCS had the best overall numbers in our Right Mark Audio tests. The MOSFET CCS had the best noise and dynamic range numbers out of the three CCSs we tested. Although it had the worst total harmonic distortion (THD) numbers of the three CCSs, the high breakdown voltage, excellent PSRR and low current variation more than compensates for the high THD. Coming in second was the cascoded JFETS because this CCS had similar numbers to the MOSFET CCS in terms of PSRR and current variation. However, the noise level, dynamic range and breakdown voltage were all considerably lower than the MOSFET CCS. The only bright spot is that the THD was the lowest of the three but one category cannot compensate for the multiple areas in which this CCS falls short in. The CRD CCS was in the middle of all 4 categories in the RMAA test. Looking at those numbers alone, the CRD would be second of the three in CCSs. But one major issue with this CCS is that the current variation is seven times bigger than the MOSFET and 660 times bigger than the JFETS. The PSRR also drops off considerably compared to the other two CCSs so the CRD comes in last among the three. Another important consideration is size and cost of all three CCSs. The CRD CCS uses the fewest amount of components (just one) while compared to the other two which both used three components. In terms of cost however, the MOSFET CCS was similarly priced as the JFET while the CRD was the most expensive. The CRD was more than 2.5 times expensive than the MOSFET and almost four times as expensive as the JFET. For many consumers who want a discrete component audio amplifier, size wouldn t make as big a difference as cost. Considering size, cost, and sound quality, the best overall CCS is the MOSFET. 20

21 Work in Progress 21

22 22

23 Bibliography Cavalli, Alex, Mark Lovell, and Bill Pasculle. "Headwize Projects - a Simple Tube/ OpAmp Hybrid Amplifier by Alex Cavalli." Headwize. June Jan < headwize.com/projects/cavalli2_prj.php>. "Mb3k.Com the Place to Be." Mb3k. 8 Mar Mar < soha.php>. Rossel, Jeff. "Glass Jar Audio." Glass Jar Audio. Jan.-Feb Mar < Smith, F. Langford, ed. The Radiotron Designer s Handbook. United States of America: The Wireless Press, York, H. Lewis. Amplifiers. Great Britain: Focal Press Limited,

24 24