Dainos Where to get stuff — an assorted list of some of my suppliers. Tube amp power supply PCB with bias supply. A lifetime electronics geek, I got a ham radio license at the age of 8 and a first-class commercial radio license at This article was published in audioXpress, July The A tube, and a push-pull A amplifier. In this article, Joel Hatch explains how to build a variable frequency synchronous motor controller to control turntable speed. In this project article for audioXpress, George Ntanavaras describes the design and construction of a high-quality RIAA preamplifier for moving coil cartridges based on high-quality Analog Device AD op-amps designed for low-noise audio applications and driven from very low source impedance. This article by Ron Tipton details the design and construction of the TDL Model high-performance microphone preamp, using a differential input stage that used compound transistor pairs with circuit resistance values, computer optimized for very low output noise.
|Published (Last):||25 December 2011|
|PDF File Size:||13.50 Mb|
|ePub File Size:||6.11 Mb|
|Price:||Free* [*Free Regsitration Required]|
The F4 is a unity gain amplifier requiring a preamplifier capable of driving the full-expected output swing. As two channels of the F4 should be able to swing 40 V in Class-A, for W of output power, the preamp must be capable of delivering Enjoy the project! The Pass F4 is a unity gain power buffer, meant to take a high-level output from a preamp and drive speakers with it — an interesting concept.
Figure 1: Balanced mono connection for the F4. A diagram of the balanced mono connection for the F4 is shown in Fig. A preamp with balanced outputs is used for all of the voltage gain and to drive the cables to the power amps, along with the power amp input impedance. The F4s buffer the balanced preamp outputs and provide a balanced signal to the speakers. And, of course, it needed to be called The ImPasse. Figure 2: Classic differential amp. The Road Not Taken I will first examine the design requirements.
In the balanced mono arrangement, your preamp must be able to swing 20V peak on each of its output polarities, and do that driven from a typical line-level source 2V RMS or 2. In the balanced mono connection, the power amplifier input impedances are identical and symmetrical. Thanks to the FET input, their values are also somewhat arbitrary — the intrinsically low gate current of a FET means that voltage offsets due to gate current can be very low even with relatively large gate resistors.
This relieves the preamp from the need to deliver much signal current; it just needs to swing many volts very cleanly. For me at least, the requirement to swing a lot of volts at low current with a simple circuit suggests… tubes. Given the symmetrical loads and the requirements for balance, you have several options. The first that comes to mind is the classic differential amp Fig. With a constant-current sink in the tail and load that is equal on both polarities, the balance is excellent.
Likewise, the output impedance, Zout, can be made relatively low; with symmetrical loads, it is roughly equal to the plate resistance in parallel with the load resistance. Power supply rejection is excellent.
To get to the target gain of 20, you need a tube with a gain somewhat larger than 40; from a single-ended source, gain to each plate of a differential amplifier is half the gain that the same circuit would have in common cathode. Being more exotic, you could check some of the high gm European pentodes, triode connected, in order to achieve the requisite gain with a low plate resistance.
Consider the common tubes. Any cable capacitance will be liable to cause HF losses. The other tubes produce a much worse problem. You can do somewhat better with the exotics. A D3a, triode connected, will have a plate resistance of slightly greater than 2k at 20mA operating current.
This will result in an output impedance of about 4k5, considerably better but still quite high. But with care and attention paid to the cables and amplifier load, this could be a viable option, albeit at some expense — D3a is not cheap, and the current to run two channels of preamp will be 80mA or more.
Figure 3: Simple common-cathode voltage amp feeding a split-load inverter. You can couple the two stages in various ways: direct coupling, RC coupling, or a combination of the two are all simple and effective. Contrary to myth, the split-load inverter has symmetrical source impedance from both the plate and cathode sides as long as the load is also symmetrical. Better yet, the overall gain of the combination of grounded cathode voltage amplifier and split-load inverter will have double the gain of the same voltage amplifier tube in differential mode.
This extends the range of tubes that you can use for the first voltage amplifier stage to include medium-mu triodes such as 6SN7. Medium-mu triodes also have the advantage of potentially lower input capacitance due to reduced Miller effect. If you use a high mu triode such as a D3a as an input tube, the input capacitance for a single-ended source will be nearly pF.
This might upset some driving sources, but these days that would be a minority. Still, it is something to consider, especially when choosing the value of the volume control—assuming a low source impedance driving the volume control, the worst-case source impedance is at the halfway -6dB setting and is about half the volume control total resistance.
A 10k potentiometer combined with a D3a input will give a -3dB point of about k, which is a satisfactory bandwidth, but higher values could begin to be problematic. Gain is also significantly higher than target, so some careful reduction measures would need to be designed in. Fortunately, a very common medium-mu triode, the 6SN7, has demonstrably excellent linearity at these voltage swings — the requisite gain, sufficiently low input capacitance 80pF —and is widely and easily available in an impressive variety of flavors.
Distortion performance is comparable to the exotics, and input capacitance is significantly lower. Among common medium-mu tubes, it is the most linear in voltage amplifier service, and the mu of 20 is spot-on for this application. Figure 4: Outline of the voltage gain stage. And a whole two-channel preamp can be done with two tubes. The fact that I had a single CV a rare British military version of the 6SN7 begging to be used did not influence my choice of tube here, no sir!
The input stage provides all of the voltage amplification and is hence the most critical regarding distortion and available voltage swing. It needs to swing slightly more than the desired 20V peak output because the phase splitter will have slightly under unity gain. It will be lightly loaded—a split-load inverter has the same low input capacitance as a cathode follower, so can have a high input impedance. A 6SN7 in grounded-cathode mode with an 8mA constant-current source plate load will show less than 0.
This distortion is overwhelmingly second harmonic. Not bad for a single device, open loop! Biasing in common-cathode voltage amplifiers typically involves a series resistor in the cathode circuit, with the cathode bypassed to ground.
A high-quality bypass cap is bulky and expensive, so an interesting and very viable alternative is the use of forward-biased diodes for developing a constant voltage at the cathode. LEDs are particularly suitable because of their low source impedance, low noise, and reasonably high forward voltage drop. A cheap surplus red LED will typically show 1. And it has the side benefit of visually indicating that the tube is indeed drawing current.
Looking at the characteristic curves for the 6SN7, you see that for 8mA of plate current and V on the plate, the grid must be about Or the cathode can be 3.
So two red 1. Another bonus of CCS loading of the plate and LED biasing of the cathode is the greatly increased power supply rejection. The outline of the voltage gain stage is shown in Fig. One notable feature is the use of an input transformer. In an earlier article, I justified its use in some detail; basically, it provides outstanding common-mode hum immunity and galvanic isolation, while at the same time facilitating either balanced or single-ended drive.
A unit is appropriate here. All of these have very low distortion and excellent bandwidth. Figure 5: A simple yet high performance CCS. Finding JFETs that will withstand the voltages and currents involved is not trivial. Unfortunately, the biasing arrangements necessary for enhancement mode devices can complicate things. A simple yet high-performance CCS using these wonderful devices is shown in Fig. This two-terminal, self-biasing cascode CCS with a source impedance in the M range can withstand V without breaking a sweat.
To understand the operation of this cascode CCS, first consider Q1. Its current is set by the value of R5. For a fixed current, the voltage drop across R5 is fixed. Considering those flat Id versus Vd FET characteristic curves, that means that the current through the FET is largely independent of the drain-source voltage.
If the current tended to rise, the voltage across R5 would rise, which would tend to drive the FET toward lower current. The opposite argument can be made for the FET tending toward lower current. So, because of that feedback effect, Id is constant. The second FET has a more subtle role. As you modulate signal across the CCS, the drain-source voltage varies.
Though the curves are quite flat i. Additionally, the drain source capacitance is significant, especially at low voltages, limiting the source impedance at high frequencies. Worse yet, at lower drain-source voltages, the capacitance is very nonlinear.
Ideally, you would hold Vds constant. And that is the function of Q2, which is connected as a source follower with the gate driven by the source of Q1. R7 is a gate-stopper, as before, with no significant voltage drop across it. Thus, the drain-source capacitance is not modulated, and the constant Vds across Q1 means that current is held more constant—equivalently, you can say that the output impedance is higher.
The complete schematic is shown in Fig. You can now start assigning component values. Because I had a bag of R carbon resistors, that was my chosen value. If you use R or R or even R, the operation of the CCS will not be affected; it is important to have the body of the resistor as close to the gate as possible. And the CCS sets the operating current of the voltage amplifier tube. FETs vary. A lot. It is not unusual to find two devices with the same part number having Idss or gm values that differ by a factor of three.
I bought several tubes of the DN and kept one. Those parts were all relatively similar, with two or three outliers from the batch of tested. But I cannot guarantee that other lots of these parts will be so closely controlled.
T-reg: A High-Voltage Regulator for Tube Amps
audioXpress August 2009 PDF
audioXpress Supplementary Material