Here is something I designed a couple of months ago just for fun and as a light design challenge, I might as well post it here. The aim was to design a tube-style passively equalized phono amplifier, but entirely with discrete high-voltage transistors, and to better the performance of the former. Simplicity was also a goal and the circuit is as simple as I can make it without significantly sacrificing performance.
The circuit was inspired by, and is topologically identical to the “His Masters Noise” design published over at the other place. But instead of being comprised of two vacuum tube common-cathode amplifier stages and a cathode follower output buffer, it is comprised of two bipolar transistor common-emitter amplifier stages and an emitter-follower output buffer. As in the tube design, both gain stages are “loaded” with current sources to ensure a healthy PSRR.
The gain distributions are similar. The first stage has a gain of approximately 100 and overloads at 1Vpeak input. Total gain at 1kHz is 47dB, giving an output of just under 1.2V with an input of 5mV.
The first stage is specifically designed to be driven by the secondary winding of a phono cartridge transformer connected in the place of V1 shown in the schematic. I based and conducted the simulations on a secondary / source impedance of 1k. A 1:1 transformer would suit MM, a 10:1 or greater for MC.
The second gain stage (which contributes practically nothing to the overall noise performance) overloads at approximately 4.5V.
Here is the circuit:
[ Guests cannot view attachments ]
Making common-emitter amplifier stages with high voltage transistors perform better than equivalent common-cathode tube stages requires a few tricks. Emitter degeneration is mandatory, for both DC stability and linearity. I have also used DC feedback in the form of voltage divider bias from the collectors to stabilize the DC operating point of each gain stage (at half the supply voltage). This is mandatory here as, unlike common-cathode tube stages, emitter degenerated common-emitter amplifier stages have much too high a collector (as opposed to anode) impedance to be properly biased with a current source alone. I’ve also buffered each stage with an emitter follower. This is an extra level of complication but it is necessary to mitigate the effects of Miller capacitance and to provide a decent input impedance to each stage.
Here is a plot of the total output noise with the source impedance set to zero. The noise vs frequency contour is of course shaped by the passive EQ. The noise of 3.6uV at 20Hz equates to an input referred noise of 1.8nVsqrtHz. This is approximately equivalent to a 150 ohm resistor, and is much smaller than the intended 1k source / transformer / cartridge resistance / impedance.
[ Guests cannot view attachments ]
THD at 1kHz with an 18mV input signal to produce an output voltage of 3Vrms is 0.0088%. The distortion spectrum is very pretty. Basically entirely 2nd with just a little 3rd poking through. Here is the FFT:
[ Guests cannot view attachments ]
Here is the frequency response / gain plot. Note that I have NOT bothered to optimize the passive EQ filter values yet. The values given in the schematic are just a ballpark calculation for proof of concept.
[ Guests cannot view attachments ]
While the circuits PSRR is good, a regulated 200V supply would be preferable to an unregulated one. In fact, separate R-C supply rail decoupling between the two gain stages and the emitter-follower output buffer would probably be a good idea too.
This is a pretty basic circuit that I wouldn’t mind soldering up and measuring, just for the hell of it (and to stick it to the tube-heads
) but I just have too many other projects to solder up first. If anyone out there is excited enough by the design to want to build and measure it (probably not) I could find the time to layout a PCB.
