Quite awhile back I got an email from John Swenson asking if the active circuitry used in the self bias could be used as an amplifier. I mentioned that I thought it would work and in fact had thought of the ideas several times in the past but had never pursued the idea. John went on to do experiments and came up with several interesting tranconductance amplifier ideas.

After hearing of his successes I decided to try some ideas of my own. One of the early experiments was to compare a single IRF820b MOSFET to a modified self bias CCS amplifier. I started calling the basic building block the Solid State Pentode based on the way the circuit works.

While the characteristic curves of single MOSFETs looks very Pentode like they do not have the low input capacitance of a Pentode. The cascode circuit below acts very much like an ideal Pentode as the gate to source capacitance of the upper MOSFET is shunted to the source of the lower MOSFET via C5.

Here is the schematic of the basic SSP building block.

About the only changes to the self bias CCS to convert it to an amplifier is to connect the lower end of C5/R14 to the source of Q1 and drive the gate of Q1 with the input signal.

One of the first experiments was to compare the performance of this circuit to a single IRF820b MOSFET as a single stage amplifier. Due to my crummy note keeping I do not know what the load resistance, operating current, or gain was in the following images. As a guess, as I was working on driver stages at the time, the circuit would have been setup for a gain of around 20 and should have been running 20ma current range.

The source impedance was lower than the images show as there was a 27K resistor to ground on the gate of the MOSFET. This is in parallel with the source impedance shown in the screen shots. The corrected values would be 34.9, 964, 7.3K, 13.5K and 17K.

First up is the frequency response of the single IRF820b MOSFET driven from several source impedances:

Note how much the high frequencies are rolled off by the high input capacitance of the MOSFET.

Next is the frequency response of the SSP

With the cascode circuits low input capacitance their is no issue with driving the stage from higher impedances. It looks like there is a very small amount of peaking when the source is around 10K.

Here is the distortion of the single MOSFET at 1Khz and 44Vrms output. Note how the high order harmonics increase with increasing source impedance. The same corrected source impedance values from above should be used.

Distortion for the SSP at 1Khz and 44Vrms. There is more second harmonic but the higher order distortion is way down. This is a case where I would glady take an increase in second harmonic distortion in trade for virtually getting rid of high order harmonics. 

The next experiment was a CCS loaded differential stage. This stage is a tranconductance amplifier, meaning that an input voltage is converted to an output current. The goal was building a solid state version of the Tabor amplifier with an interstage transformer between the input and output stage. I was using 6.6K as the load for the input stage as this is what the estimated input impedance of the Tabor output stage. Later I substituted a 6.6K output transformer and found that by increasing the operating current and B+ voltage the single stage could deliver 10W into 8 ohms at quite low distortion. Not bad for a single stage amplifier. The output impedance is very high. With the amplifier delivering 10V rms into 8 ohms I clipped a second 8 ohm load resistor in parallel dropping the load impedance to 4 ohms. The output voltage dropped to 5V rms.

Here is the schematic of the input tranconductance stage with values for both input stage and tranconductance output stage amplifier uses.

After the success of the input stage I wanted to mate it to the output stage and see the results. Here is the schematic of the first prototype of the "Solid State Tabor".

A photo of the prototype. Ugly but functional

The VR tube regulator was replaced by the simple resistor divider shown in the schematic above. The power supply voltage divider is not obvious. From B+ the path is the both sides of primary of the output transformer through the 2 feedback resistors then through the secondary's of the interstage transformer to R8 to B-. The ratio of the feedback resistors paralleled + the DCR of the primary of the output transformer + the DCR of the secondary's of the interstage transformer to the value of R8 sets the ratio of B+ to B-

Here are the distortion measurements into 8 ohms. The interesting thing to note is the difference between 7 watts and 8 watts up. 7 watts is the power limit for cascode operation. At 8 watts the "plate" voltage on the output MOSFETs reaches down to the gate voltage of the upper MOSFET. With the drive current coming from the lower MOSFET the upper MOSFET saturates. The output voltage will pull down to just a few volts across both output devices. You can see the high order distortion increase when the cascode circuit operation is lost. Very much like the comparison of single MOSFET to cascode shown at the beginning of this page.

This is a very nice sounding amplifier. Totally quiet on my 95dB speakers. Sounds very much like the tube version of the Tabor but cleaner. No hint of "solid state sound".

I suspect the cleaner sound is the lack of microphonics of the 6AU6 input tubes and the directly heated output Pentode.

Since this project I have built the "Big Brother" to this amp. The larger version runs B+ at 350 volts and the output stages are idling at 180ma. The output transistors have been upgraded to pairs of FQAF5N90 900V 90W MOSFETs running in parallel. The "big" amp puts out 25 watts per channel and measures very similar to the 10 watt version. It was used to power the Bastanis Apollo speakers at RMAF a few years ago.

Here is a photo of  the "big amp". Down side to class A solid state is the heat sinking needed. This amp dissapates 250W idling and stabilizes at about 60 deg C. Need more heat sinking...