BoxHead

The long-awaited 2 × 6973 amplifier has been essentially done since April, and it’s called BoxHead. I held off on this post as it was undergoing occasional adjustments since then. Here’s a rundown of the design:

Power supply

The power transformer is a “semi-custom” unit from Edcor. Although it’s not a standard model, it was a custom configuration requested by a previous customer, so Edcor produced one for me without charging the standard custom design fee. Rectification is through a Full-wave bridge of UF4007 diodes with a STPS1150M Schottky diode after the bridge. The series Schottky is included to reduce radiated rectifier hash. The filter has two C/L stages. Both chokes are 9H, but with different current ratings and DC resistances (one is a Weber “general purpose” and the other is a Hammond 154). The first two capacitors are sections of a 100µF/100µF can from JJ. The output transformer/anodes are taken off the first stage; screens come off the second.

Input

The input is a not-so-remarkable 6AU6A pentode stage, biased with an unbypassed cathode resistor. The screen voltage is set through a an unbypassed divider that bleeds through the cathode resistor. The leakage current through the bottom half of the divider was chosen to be about equal to the idle current through the screen. This makes the stage stiffer than it would be with just a screen dropping resistor but still allows some screen sag. Since the cathode bias resistor is reduced in value to accommodate the divider leakage, the amount of cathode degeneration is reduced when compared to a plain unbiased cathode arrangement.

This stage is followed by a gain control with a parallel resistance set so that the low-rolloff frequency increases as gain is increased; this cuts bass as the next stages start to overdrive and helps clean up the distortion tone.

Gain stage

Next is a resistance-coupled 12AV7 gain stage biased rather hot so that it can be directly coupled to the next stage, which has a partially bypassed cathode resistance, allowing some cathode degeneration again and a bit of low-frequency rolloff.The DC coupling between the 12AV7 stages features The Valve Wizard’s diode grid protection circuit.

Tone stack

Next is a Peavey-style T-M-B tone stack and master volume. The Peavey tone is not vastly different from the Fender and Marshall versions. The mid notch frequency seems to shift with the adjustment of treble and bass to a greater degree than happens in the Fender & Marshall.

Recovery, phase inverter 

The pentode recovery and cathodyne phase inverter are an altered version of the DC coupled 6AN8 arrangement used famously in Sunn guitar amps. The pentode section is biased similarly to the 6AU6A pentode at the input stage. The DC-coupling arrangement is improved over the Sunn design by adding a voltage divider between the pentode stage and cathodyne. The divider is designed so that even when the pentode is driven completely into cutoff and its anode reaches the HT supply voltage, the triode's grid will not be driven positive with respect to its cathode, eliminating the possibility of nasty-sounding clipping at the phase inverter.

To select the divider ratio, first draw the load line using the sum of the cathodyne load resistors [22k + 22k = 44k]. Note that the line intercepts the Ec = 0 at about 40V. This means the PI will be reaching clipping when the voltage across the triode is 40V. Subtract 40V from the HT supply and divide by 2 to see where the maximum cathode voltage should end up: [(280V - 40V)/2 = 120V]. So, we want to make sure that when the pentode's anode reaches the 280V HT, the cathodyne grid goes no higher than 120V. With the 470k/330k divider, the maximum grid voltage will be 115V. [280V × (330k / (470k + 330k)) = 115V] The operating point of the pentode plate is set to that the middle of the divider will rest halfway between 0V and 115V. Voltage swing at each output of the cathodyne is more than enough for the 6973s.

Power 

The power stage is push-pull 6973s in shallow class AB1 (nearly class A), cathode biased, with a big 1,000µF cathode bypass. The output transformer is an 8k:8 Weber (W022913) designed as a Fender 6V6 replacement. Output power is about 15W.The power stage was designed so that it can (and did all through testing and development) work with a pair of 6CM6s instead of the pricier 6973s. A 6CM6 (available N.O.S.) is basically a derated 6V6GT stuffed into a noval bottle with a pinout nearly identical to a 6973. The 6973 is current-production from EH.

Construction

The electronics are assembled onto two custom turretboards (pre and power) with the tube sockets drilled through. It's all housed in an aluminum Hammond chassis box attached to the cabinet with 10-32 automotive cage nuts.

Cabinet 

The amp and matching speaker cabinet are solid pine with hand rubbed oil finish and a Marshall "cane" grille fabric. The edges are joined with tongue and groove leaving a distinctive end-grain at the front corners.The speaker cabinet is closed back and includes a 100W 12" Sica Solid ceramic-magnet driver, similar but not identical to Jensen.

Demonstration Video:

See my follow up post with the video!