For Vp=100V, biased at 10mA, the circuit has the following characteristics:- Gain ~2
- Zin >1M
- Zout ~3.5k
- CMRR 50dB @20kHz (ideal resistors)
- CMRR 37dB @100kHz
- CMRR <20dB @1MHz
This circuit will drive the unmodified 24k input impedance of the P3A amp without any difficulty, although a following impedance of at least 35k would be preferable.
Current source
Replacing the typical cathode resistor with a current source greatly improves CMRR and also has beneficial effects in terms of THD. Unfortunately, it necessitates either a low voltage supply or transistors capable of high voltage operation. It is possible to use low-noise transistors (BC549 etc) which will deliver high performance.
Components
For low noise and maximum CMRR, all resistors should be 1% metal film. The plate resistors must be rated at least 1W (2W preferable). The other resistors may be 0.25W. The output coupling cap must be rated for the full supply voltage, and preferably over-rated by a respectable margin. If this capacitor fails, you will have the full plate voltage at the output!
Power Supply
Proper power supply design is critical for two reasons: one, this is intended to be the input to a power amp, so any hum or noise will be greatly amplified; two, the PSRR of this circuit is almost nil.
The supply above has negligible ripple at the 50mA load of five tube circuits (simulated to <1uV/-117dBV, but in reality this is probably below the noise floor of the supply). For a stereo pair less filtering would achieve the same result. Duncan's PSU Designer is very valuable for tweaking supplies. If desired, 1-10nF bypass capacitors could be added around the diodes in the bridge.
A disadvantage to this supply is that it requires a very long time (20s) to stabilize at the rated voltage. At the output this appears as a slow rise to nearly 4V before the output bias returns to 0V. In the P3A amp the power-on sequence is filament supply, B+ supply, then solid state supply once B+ stabilizes.
Modifications
This very useful circuit is capable of being used as a differential buffer, differential to single ended receiver, or single ended to differential phase splitter/differential transmitter. As shown it is differential input and SE output with a gain of about two. If the inverting input were tied to ground, it could be used for SE input (unity gain). In this configuration, the inverting input could be used for a NFB port. Differential output would be achieved in any configuration by adding a coupling cap to the plate of the tube on the left (gain increases by 6dB).
Gain may be increased (up to about 7 SE or 14 differential) by decreasing the value of the 500 ohm degeneration resistors. Maximum gain will be achieved in a common cathode configuration (no resistors). For higher gain, a higher mu tube such as 12AU7 or 12AX7 must be used. Higher mu tubes tend to have higher Rp, though, making them less able to drive a difficult load.
Information and a spice model for the 5687 may be found here.
Thanks
Thank you to the folks at DIYaudio, especially Harry, Bernhard, and hagtech, without whose assistance this would not have been possible.
Additional testing
The table below lists simulated harmonics at various plate voltages and bias currents. The measurements were taken for a 1.5Vpp 1kHz input signal, output into a 20k resistive load bypassed by a 100pF capacitor.
|
|
|
|
|
|
|
|
|
|
|
| 30V 10mA |
50V 10mA | 80V 10mA | 100V 5mA | 100V 7mA | 100V 10mA | 120V 10mA | 150V 10mA | |||
| 2nd | -23dB | -64dB | -76dB | -73dB | -73dB | -76dB | -77dB | -78dB | ||
| 3rd | -67dB | -68dB | -82dB | -69dB | -74dB | -82dB | -83dB | -82dB | ||
| 4th | -42dB | -92dB | -93dB | -92dB | -91dB | -92dB | -92dB | -93dB | ||
| 5th | -100dB | -121dB | <-120dB | <-120dB | <-120dB | <-120dB | <-120dB | <-120dB | ||
| 6th | -53dB | -92dB | ||||||||
| 7th | -84dB | |||||||||
| 8th | -62dB | |||||||||
|
|
|
|
|
|
|
|
|
Measurements taken with the balanced circuit showed considerably less distortion, particularly even order. (as would be expected)
Contact Info Disclaimer