New Project – 20 W Stereo Integrated Hybrid Amplifier

This page documents the design and refinement of a 20 W stereo integrated hybrid amplifier. The log below traces development of the amplifier, shown on the right in finished form, from planning through construction and optimization.

20 W stereo integrated amplifier


The objective here is to produce an integrated 20 W stereo amplifier that includes, at minimum, tone controls and a phono preamp. The idea came via a request from my daughter for a tube-based amplifier for dorm room listening to LPs while at college. Much to my surprise, a number of artists are releasing new albums on vinyl as well as CD, and playing records in the "old school" way has gained a certain popularity. My original approach was to combine the 20 W stereo amplifier with elements of the preamplifier that I had already built. This proved to be impractical because of space limitations on the 14-inch by 17-inch chassis. Because of space limitations in the dorm, a single chassis implementation was needed.

My brother had suggested that a solution to this challenge might be found in some products from the 1960s that were hybrid devices, with transistors in the low-level preamplifier stages and tubes in the output stages. This concept had an appeal (it certainly solved the space problem) but only if it was true to the "tube era." Fortunately, copies of 60s vintage transistor manuals from the various manufacturers of the day are readily available. I found some interesting circuits in the 1964 RCA Transistor Manual and the GE Transistor Manual.

Update 1

Good progress has been made on the 20 W integrated hybrid audio amplifier (the "Dorm Amp"), with most of the chassis drill/punch work completed, and the right and left channel circuit boards completed, as shown below.

Amplifier PWB, left channel

The PWBs for the power amplifier stages are essentially identical to the circuit boards for the 20 W stereo amplifier. The right channel board (see right) includes the high voltage power supply for the unit and primary power connections. The left channel (see left) contains only the amplifier circuitry. The same PWB design is used for both amplifier channels as a cost-saving measure.

A bank of two large filter capacitors is mounted external to the right channel PWB.

Amplifier PWB, right channel

The transformers for the Dorm Amp are the same as for the 20 W stereo amplifier, with the exception of the 6.3 V filament transformer used in the 20 W amp. Since the preamplifier is integrated into the same chassis as the power amplifier in the Dorm Amp, there is no need to supply power to an external preamp. The filament transformer (used in the 20 W amplifier to provide filament voltage to the separate preamplifier chassis) is changed in the Dorm Amp to provide 25 V ac to power the transistorized preamplifier power supply.

Update 2

The preamplifier PWB has been completed and preliminary tests passed. The board contains the power supply, right and left channel phono preamp, and right and left channel tone control circuits.

Preamp PWB

The basic design for the phono preamp came from the 1964 GE Transistor Manual and the tone control stage from the 1964 RCA Transistor Manual. Some component substitutions were needed, but otherwise the implementation is true to the original design.

Preamp PWB, foil side

In order to simplify the power supply requirements, the tone control stage was modified to accept a positive supply (V+) referenced to ground. The original circuit required a negative supply.

Update 3

Assembly of the amplifier was completed today. As mentioned previously, the unit closely follows the layout of the 20 W stereo amp described on the Audio Projects Page of this site. The major change is the addition of the transistorized preamplifier board, shown above. Tests and a proof of performance are pending. Because of the similarities with the 20 W amplifier, no surprises are expected with regard to the power amplifier section. The preamplifier board has already passed a basic functional test, but detailed performance tests have not yet been run.

20 W stereo integrated amplifier, front view

As planned, the styling of the 20 W integrated hybrid amplifier is patterned after the 20 W stereo amplifier, and the front panel is similar to the multifunction preamplifier (see the Audio Projects Page). For production economy, commonality with other units was important.

Three inputs are provided on the amplifier—phonograph, line, and a front panel auxiliary input. The "front panel aux" input is a mini jack intended for connection of personal entertainment devices (e.g., iPhone).

20 W stereo integrated amplifier, rear view

Update 4

Preliminary tests on the power amplifier were completed today, with both channels performing as expected. The next step is a series of detailed proof-of-performance measurements of the overall system—from the phonograph and line inputs through to the speaker terminals.

Chassis bottom view

The chassis view of the 20 W stereo integrated hybrid amplifier is shown on the left. The preamplifier PWB mounts along side the left channel power amplifier, as shown on the right. This arrangement provides for good shielding of the low-level audio circuits, since the board is surrounded on three sides by the metal chassis. This position is also as far away as possible from the power supply components, notably the power transformer.

The phonograph input jacks are located only a few inches from the phono preamp stages. The large blue chassis-mounted capacitor on the side panel is for the preamplifier power supply.

Preamp PWB installed

Update 5

The proof of performance was completed today on the phonograph preamplifier stages. Frequency response was within +/–1 dB referenced to the RIAA curve, with the exception of the very low end (below about 70 Hz) where response widened to +/–2 dB down to 20 Hz. Distortion was quite low; typically below 0.4 percent. Hum and noise was 60 dB or better (unweighted).

Proof testing on the tone control stages of the preamp is still underway. The bass circuit performed well. Frequency response was flat from 1 kHz down to 30 Hz, within +/–1 dB. The maximum bass boost (at 50 Hz) was about 15 dB; maximum bass cut was about 15 dB. Additional work on the treble circuit is planned to raise the "influence range" of the treble control. As built, treble boost/cut impacts frequencies (to some degree) below 1 kHz. My design goal is for the treble circuit to be essentially invisible at 1 kHz and instead begin having an impact on frequency response at about 3 kHz. It may be possible to accomplish this by changing certain capacitor component values.

Performance measurements of the power amplifier stages will be conducted later as the final end-to-end test of the system. This particular power amplifier design has been well-characterized through previous work. No surprises are expected.

Update 6

A full proof of performance was run on the power amplifier section today. It performed as expected, closely matching the 20 W Stereo Amplifier described on the Audio Projects Page on this Web site.

Work on the tone control circuit has arrived at a point where the treble function performs as desired. The problem is that in order to achieve the range of control expected it was necessary to make circuit changes that decreased the nominal output level of the stage. As a result, it will be necessary to add a buffer amplifier after the final tone control stage to provide sufficient level going into the power amplifier for full output, plus a reasonable amount of headroom. The buffer amplifier has been designed and will be installed after parts arrive. The amplifier is contained on a small (0.75-inch by 1.5-inch) PWB that mounts (as a daughter board) on the preamplifier PWB.

Update 7

While waiting for parts to arrive for the buffer amplifier, I came up with the idea of an Automatic Protection Option for the Dorm Amp. A first implementation of the Auto-Protect system has now been built and installed. Auto-Protect is an add-on circuit board for the 20 W stereo, 20 W stereo integrated, and 40 stereo amplifiers. The board is intended to add additional layers of protection to what is already provided in the amplifiers. The Auto Protection system adds the following features:
• Automatic Shutoff – While is possible to operate the amplifier continuously for long periods of time, most consumers would likely prefer to not do so. Experience has shown the amplifier may be tuned on to play one or more records and then inadvertently left on after the record ends. Automatic Shutoff provides a fixed maximum run time for the amplifier—adjustable from about three hours to five hours. This time delay can be defeated or reset if desired.
• Over Temperature Shutdown – The amplifier is designed to operate reliably at normal room temperature. If operated at an elevated temperature, component damage or shortened tube life could result. This feature shuts down the amplifier if the temperature inside the chassis exceeds 170°F. This is sized to be below the maximum rated operating temperature of the amplifier components.
• Under-Voltage Shutdown – The power supply is designed to operate reliably under a wide range of conditions. It is possible, however, that because of the failure of a tube or another component, the power supply could be loaded to the point that continued operation could result in damage to one or more devices. This circuit monitors the plate voltage supply and screen voltage supply (if used) and shuts down the amplifier if the voltages are below normal.

These protection mechanisms are in addition to the soft-start circuit and line voltage circuit breaker already provided on all amplifiers.

Auto-protect PWB

The Auto-Protect board measures about 6-inches square. In order to fit in the amplifier, the board is mounted with the foil side facing down toward the bottom plate.

In keeping with the 1960s "tube era" design guidelines, the time-delay and voltage-sensing circuits are adapted from classic GE and RCA designs from the 1964 transistor manuals (which I happened to have on hand). Relay-based logic is used to manage the protection stages.

The Auto-Protect board fits nicely in the 20 W stereo integrated amplifier, although it is rather packed at this point.

Auto-protect PWB installed

Update 8

The Dorm Amp project was officially completed today with final tests finished on the amplifier. Following previous tests, some changes were made to the tone control stages to improve performance. While the buffer amplifier described previously worked well, I wanted to eliminate it if possible. I decided to trade off treble range for gain. I was was able to increase the output of the tone control stage to greater than 1.5 V rms, which is more than enough to drive the power amplifier to clipping. So, in the final implementation, the buffer amplifier is not used. This simplifies the mechanical layout and results in better noise performance.

The Dorm Amp is now completed and will be installed on Friday. While the project was considerably more complicated than expected, the final result was quite good. The amplifier sounds great.

Update 9

With the Dorm Amp in service, and not content to leave well enough alone, I decided to build up another preamplifier board on the bench and try to improve on the circuit, in particular the tone control stage. The minimum order from PCB Express (the vendor) is two, and so I had a free PWB on hand. I built the board as implemented in the Dorm Amp (above) and then spent a couple of days tweaking the design on the bench.

The initial focus of this work was the phonograph preamplifier. In the first implementation, the phono preamp matched the RIAA equalization curve within about +/–2 dB. I wanted to improve on that and zeroed in on the frequency shaping network components. By using tighter-tolerance capacitors and resistors I was able to bring response to within +/–1 dB of the RIAA curve.

The second part of this project centered on the tone control stage. As detailed previously, in order to achieve the desired performance it was necessary to trade treble range for power output. After some investigation on the bench, I observed that the two-stage circuit did not clip symmetrically. Through adjustment of the bias points I was able to significantly increase output of the stage, which permitted dedicating additional gain to the treble range. The finished circuit provided a maximum output level of 1.5 V into 100 k ohms. The bass range is +/–12 dB and the treble range is +/–6 dB.

During the winter school break, I brought the Dorm Amp back home and replaced the preamplifier board. It passed listening tests with flying colors. Measurements followed and met or exceeded expectations.

The benefit of standardized mounting of major components cannot be over emphasized. The ability to drop a new circuit board into an existing product is a valuable benefit.

Update 10

With some available time on my hands, I decided to take another look at the transistorized tone control stage used in the Dorm Amp. The modified version described in Update #9 has performed well. Still, I was looking for an improved circuit should the need arise to build another transistorized tone control preamp board. As a starting point, I used a schematic diagram of the tone control stage of the Heathkit AA-14 amplifier, which was offered for sale in the mid-1970s. The circuit needed more gain for my particular application, and a different V+ operating voltage. Other notable changes included different (generic NTE substitute) transistors. The original one-stage circuit was modified to add an input buffer amplifier, which provided the needed gain for the Dorm Amp application.

Bench testing tone control stage

The test bed implementation of one channel is shown on the right. This informal layout made it easy to modify the circuit, and to re-use most of the components for something else at a later date.

Measurements on the heavily modified Heathkit circuit provided impressive numbers. Notable among these was the frequency response with "flat" settings of the bass and treble controls. The measured numbers were +/–1 dB from 6 Hz to 285 kHz (see the photo on the left). The high-end –3 dB point was 580 kHz!

Frequency counter readout

Aside from the "wow" factor, of course, response up into the Medium Frequency band has little benefit in an audio amplifier (in fact, it invites parasitic oscillations). But still, it was impressive. As for the numbers that really count, the stage is capable of 2.8 V rms output into 100 k ohm load, with an input of 0.85 V rms. Measured at a typical operating level of 1.5 V rms output (input = 0.4 V rms), THD was less than 0.25 % across the audio band and IMD was less than 0.6 %. Noise was –75 dB or better; this was an impressive number, given the fact that no effort was made to keep lead lengths short or otherwise use good construction practices on the very informal breadboard circuit shown above. At 20 Hz, bass boost/cut was +/–12 dB. At 20 kHz, treble boost/cut was +/–10 dB.

While running performance tests on this circuit, I found a convenient way of setting the "flat" response point for the bass and treble controls. Apply a 100 Hz square wave at an input level that keeps the output well below clipping. Adjust the bass and treble controls for best square wave response as observed on an oscilloscope. The bass control will impact the overall slope of the square wave; the treble control will impact the leading edge of the waveform. Switch the generator to produce a 1 kHz square wave; a very clean square wave should be observed on the oscilloscope.

I realize this project is a deviation from the vacuum tube theme of this web site; however, having a hybrid tube/transistor amplifier option may solve certain implementation problems for some builders. The updated circuit will be used in any future build of the 20 W Stereo Integrated Hybrid Amplifier.

Update 11

The improved transistorized preamplifier PWB described in Update #10 was completed today. In a departure from previous circuit boards, this one is a four-layer design, with signal traces on the outer layers, and ground and V+ on the inner layers. This greatly simplifies layout for a complex board and minimizes noise and hum in the circuits. The cost differential to produce the PWB is relatively small—on the order of a 15% premium in small quantities.

Revised preamp PWB

The component side of the new circuit board is show on the left, and the foil side on the right. Performance tests are pending, but are expected to fall closely in line with the breadboard version described above.

Revised PWB, foil side

Update 12

Performance tests were completed on the preamplifier board today, with good results. Performance matched the preliminary figures obtained with the breadboard version described above. The board was then installed in the 20 W integrated chassis and final tests were run. All parameters met the design objectives. Project completed.

Update 13)

The filament circuit of this amplifier (and others described on this site) includes a Hum Balance potentiometer that sits across the 6.3 V filament supply, with the wiper arm tied to a +50 V dc source from a voltage-divider off the B+ power supply. This arrangement is used to reduce hum in the output of the amplifier. The reasoning for this approach, and the the results obtained, are detailed below.

The arrangement of the Hum Balance circuit in the amplifier is based on the recommendations of RCA, as detailed in the RCA Receiving Tube Manual (1974). The basic amplifier circuit, with the 6U8A (substituted for a 7199) input/phase splitter and the 6973 output pair is patterned closely after a circuit described in that publication (and previous RCA tube manuals). In the discussion of minimizing hum in an amplifier, the manual states:

“Heater-type tubes may produce hum as a result of conduction between heater and cathode, or between heater and control grid, or by modulation of the electron stream by the alternating magnetic field surrounding the heater. When a large resistor is used between the heater and cathode (as in series-connected heater strings), or when one side of the heater is grounded, even a minute pulsating leakage current between heater and cathode can develop a small voltage across the cathode-circuit impedance and cause objectionable hum. The use of a large cathode bypass capacitor is recommended to minimize this source of hum.

Much lower hum levels can be achieved when heaters are connected in parallel systems in which the center-tap of the heater supply is grounded or, preferably, connected to a positive bias source of 15 to 80 volts dc to reduce the flow of alternating current. The heater leads of the tubes should be twisted and kept away from high-impedance circuits. The balanced ac supply provides almost complete cancellation of the alternating-current components.”

In the 20 W stereo amplifier, the three methods recommended by the original tube manufacturer (over a period of many years) have been used to reduce hum: 1) parallel filament connection, 2) twisted heater leads dressed against the chassis and (owing to the unique tube mounting method) nearly 1-inch away from any signal-carrying lines, and 3) positive bias on the filament (about 50 V dc in this case).

The RCA Manual, in at least the 1974 edition, included three basic power amplifier circuits in recommended configurations. In all cases, a positive bias is used on the filaments.

In operation of a practical amplifier, the typical impact of the Hum Balance control on the overall noise performance of the amplifier is relatively minor. Adjustment can usually reduce the hum level by a few dB. With the noise level of the tube amplifier circuit well below -80 dB referenced to full power output at 1 kHz (unweighted), the noise component is well under control.

For the tube-based preamplifier described in the book and on this site, I used parallel filaments with a variable resistor to ground. The performance in this case was about the same (a few dB) as the positive bias for the power amplifier. Faced with a greater noise challenge because of the microphone-level inputs, I did not try positive bias on the preamp tubes. Instead, in the second generation of the preamplifier, I used a rectified DC supply to feed the filaments. The dc supply floats and the potentiometer to ground in still used. Taken together, these changes made a significant improvement in noise—well over 10 dB. Long-term performance of the preamp has been very good.

There are a number of approaches to minimizing hum, other than the one used on this amplifier. Testing the various approaches, I put a power amplifier on the bench and measured the following results:

• With +50 V dc through a potentiometer (the Hum Balance control) to the floating filaments, as recommended by RCA: noise = –92 dB relative to 12.5 W maximum output at 1 kHz. This is the as-built configuration.
• With the +50 V supply removed and the arm of the Hum Balance pot tied to ground: noise = -89 dB
• With the arm of the Hum Balance pot connected to a variable voltage source of 0 V to +50 V: at 0V = –89 dB, between +10 and +50 V = –92 dB (relatively constant over this range)

So, given the RCA recommendation of +15 to +80 V, and the ample margin on heater-to-cathode voltage (50% of the recommended average value), I will stay with the Hum Balance circuit as-designed. It is always good to reflect on circuit choices, even if they end up at the same place.

Update 14

The cost of building any project is an important consideration for the audio enthusiast and hobbyist. The focus of the first-generation Dorm Amp was to optimize the various elements that went into the project. Cost was a secondary consideration. Still, cost is an important design element.

The total BOM (including shipping expenses) for the 20 W stereo integrated amplifier with the auto protect option was $2,785. The major cost centers included:
• Electronic components = $839
• Transformers = $486
• Printed wiring boards (PWB) = $816
• Plexiglas cover = $150
• Tubes = $95
• Decals = $75
• Front panel = $140

Builders could eliminate the PWB cost by using hand-wired terminal strip construction. Note that four PWB designs are used in this amplifier—one for each channel, one for each speaker output circuit, the preamplifier, and the auto-protect circuit—for a total of six boards. These costs apply to minimum quantity orders. With higher quantities, the cost per board drops dramatically.

Other potential areas of cost reductions include the back and top-side decals, and the Plexiglas cover.

The total time required to build the amplifier—from ordering parts to completing performance measurements—was approximately 30 hours. This estimation does not include circuit rework, which was a one-time effort.

As with any electronic product, costs fall rapidly as the volume (number of units produced) increases. For a one-off project, however, there are are not many ways to reduce the cost without reducing the feature set

Update 15 (January 2017)

Continued work to improve the Dorm Amp has resulted in a second generation unit. Detailed step-by-step documentation is provided in a 196 page User and Assembly Manual that is available for download. If you are familiar with the Heathkit assembly manuals of the past, the approach taken with this new manual should seem comfortable. For those who want a printed version, it is available for purchase on Lulu—a web site for specialized, print-on-demand documents. (There is charge of $14.95 for the printed manual; see "20 W Stereo Integrated Hybrid Amplifier: User and Assembly Manual." on the Lulu web site.)

In addition, a ZIP file is available for download that will assist in building this amplifier. The ZIP file contains the following individual files:

• Schematic diagrams of the preamplifier, amplifier, and auto-protect circuits as ".pdf" (Acrobat) files
• Bill of materials for the amplifier and auto-protect circuits as an Excel file
• Printed wiring board layout files for the preamplifier, amplifier, speaker terminal, and auto-protect boards as ".pcb" ( ExpressPCB) files
• Front panel layout as a ".fdp" ( Front Panel Express) file
• Chassis layout, bottom view, as an Acrobat " file
• Chassis layout, top view, as an Acrobat file
• Drill pattern for the chassis as an Acrobat file
• Layout for the acrylic trim piece as an Acrobat file

Note that the files above are provided as-is. Every effort has been made to make sure they are complete and accurate, but no warranties are expressed or implied. Builders are encouraged to double-check the information contained in the above files prior to proceeding. For the front panel layout, users can customize the text as desired; e.g., "Built by John for Mary, January 2016."

The current version of the Dorm Amp builds upon the previous versions, adding some new features and simplifying construction. A description of the latest design can be found in the User and Assembly Manual, and for the sake of simplicity will not be repeated here. The estimated cost to build the amplifier is $4,500 for parts, and about 28 hours assembly time. As noted previously, cost reductions are possible, depending on the preferences of the builder. For example, the BOM cost can be reduced by eliminating the add-on auto-protect board.

Considering the price tag, this project may be outside the range of some builders. However, you may find that elements of the available downloads will be useful in other projects, or wish to adapt some of the circuits for your own projects. The available documents are, thus, offered as a starting point for future efforts.