This is a small high performance amplifier that does deliver solid high end performance equal to best in class at this power level.
It is a bit complex and the build may not be ideal for the beginner.
The input stage is a single stage fully balanced and of the cascode type. I used a 6n1p because it has good linearity and a fairly high S of about 4.25 paired with a BJT a MPSA42. I used a BJT as it improved performance over a second tube and lowered the required power supply voltages needed. This allowed high gain in a compact cascode stage (more than even the 6n2p) as the cascode stage sets gain almost entirely by S, the voltage to current gain of the tube as the plate voltage is held nearly constant.
The cascode stage also removes the miller effect for plate to grid capacitance for wide bandwidth. This results in high gain, low distortion, wide bandwidth with excellent behavior in the case of transient overload due to a balanced topology.
Next comes the output stage that I used a 6p15p-e.
This is tube was designed for video amplifier use and is not often used for audio output stages. I does have some advantages.
The voltage gain is high for a output pentode so only a small drive voltage is needed similar to the 6BQ5 relaxing the demands on the input stage for improved linearity. The low bias voltage needed also keeps the power loss in the cathode resistor low.
With the suppressor grid grounded the 6p15p proved to be a pretty disappointing output tube with high screen current and limited plate voltage swing.
However the suppressor grid is routed to a separate pin and this allows the suppressor grid to be adjusted on your particular amplifier. In testing I found useful befits. By setting the suppressor grid above the cathode by the correct voltage the plate current delivered increased, screen dissipation dropped, power output increased by about 20% and over all distortion was reduced.
I set G3 to about 28V above the cathode to limit the screen current (and screen power!) at full out power as otherwise the 6P15P screen current rises to high levels when the plate drops below about 100 volts.
With this 6p15pe changed from a lackluster to stellar performer. You will note the design provides a zener regulated DC voltage on the suppressor grid.
NOS 6BQ5 Note plate and screen current levels
6P15P with G3 tied to cathode. Note the lower plate current and higher screen current
6P15P with G3 set 28V above cathode.
Note the plate and screen current now almost perfectly match a NOS 6BQ5
For this amplifier I used a toroidal transformer from AnTek part number MP-10W80.
It provides wide bandwidth and is conservatively rated delivering over it's 10 watt rating at low distortion.
Toroidal transformers tend to be less tolerant of DC imbalance in the output stage so care setting the bias current and good output tube matching is helpful.
Many amplifiers with non-balanced designs look great on steady sine waves but when the transient behavior is investigated, especially during a transient overload the non-balanced design almost always undergo a DC level shift often made far worse by power supply sag.
This can result in long recovery periods where the amplifier performance can be pretty nasty. This is of course is hard to measure in the real world but spice simulations allow this to be seen.
I took a number of measures to reduce this issue. First the amplifier is fully balanced front to back. Second the power supply is fully regulated for all low level circuits. The circuit uses partial cathode bias with some of the bias being supplied from a negative voltage to keep the required cathode resistor values low and to help stabilize the operation point during over load.
There are adjustments provided to trim both the DC and AC balance for tube tolerance. The cathode resistor is bypassed by a very large capacitor so the bias voltage takes a long time to drift off during transient overload. There is only two gain stages the input and output stage with a single coupling capacitor between keeping phase shift down. Finally the input stage is of the cascode type providing wide bandwidth and low phase shift.
Is all this bother needed?
Based on listening and measuring it seems worthwhile.
At 1 watt the THD is about 0.015% and at 10.5 watts about 0.2% THD.
Further I see the issue of performance loss during transient overload as a overlooked factor of significance in the difference in "sound" of different hifi amplifier designs. Of interest negative feedback often makes this issue worse with a longer recovery time.
Guitar players know the transient overload effect well as extreme over load is normal and desired in their amplifiers and considered a key part of the "sound".
By the way for a excellent source of ideas and understanding check out Tube cad .
Spice sim of the ptoro type
Proto type PCB. Note the use of surface mount resistors where possible. This allows a reduction in size and so stray capacitance for improved performance.
Comments