Date: Wed, 15 Jul 1992 17:01 EST
From: CLEWER@LANDO.HNS.COM
Subject: Homebrew D76 clone - an amateur's experiences (Thanks ARC)

This last winter I completed my latest in a series of homebrew audio
projects.  I built, from scratch, a replica of the Audio Research D76
vacuum tube power amplifier.  Or at least as close as I could come
given certain constraints (engineering and economic).  I thought I'd
like to share my experience with the rest of you.

I found the schematics to the D76 in an issue of the Audio Amateur.
Since it stated in the caption that ARC does not object to home experimenters
building a unit for personal use, I decided to go for it.  I had designed
and built other tube amps before, both for hifi and instrument applications,
but never one with such a complicated topology.

The main obstacle to building this amplifier was to locate a suitable
output transformer.  I used a pair of Hammond 1650R transformers, which
are rated at 100W output power.  The circuitry of the D76 uses feedback
from the OT secondary to the cathodes of the 6550 output tubes.  The 
secondary arrangement of the 1650R is such that simultaneous 4-8-16 ohm
taps are not available, but I found an arrangement where two 4 ohm 
windings could be used in a balanced configuration.  Since I am using
the amp to drive 4 ohm speakers, I use one of the secondary windings
as an unloaded feedback winding, and the other to drive the load.

To simplify the circuit, I did not use ARC's constant current source 
loading of the driver stage, nor did I use their 4 triode phase inverter.
I substituted 50k plate resistors from a 500V B+ supply for the
current source, and used a "paraphase" arrangement for the phase 
inverter.  I included potentiometers to balance both those stages,
but I found that the cross coupler stages and the local negative
feedback used in the design made these adjustments unnecessary.

I had to be creative in getting components for the power supply.
I used a multiple transformer arrangement for the B+ transformers,
and stacked two 350 volt (raw) dc supplies to get a +700 volt raw
B+ for the KT88 (not 6550) output tubes.  I designed MOSFET power
supply regulators for the +600V (plate), +300V (screen) and +315V 
(driver) B+ supplies.  These regulators used LM10s as the reference
and feedback amplifier, and provided 600V from 20 mA to well over 500 mA.

The C- (bias) supply was also regulated.  Absence of the bias output voltage 
disables the main power relay to prevent output stage burnout in the case of 
bias failure.  Heaters likewise were dc regulated, with a slow turn-on 
circuit to gently heat up the filaments.  It takes about 30 seconds for the 
heater supply to ramp up to full rated voltage, at which point the main 
power relay is tripped so as not to apply B+ across cold cathodes.

As a final touch, I installed four differential amplifier / voltage window
comparator circuits across each of the output tube cathode resistors (used
for cathode current sensing) which drive 4 tri-color LEDs.  Red indicates
current 2% or more over nominal, green indicates 2% or less below nominal,
and yellow is nominal plus or minus 2%.  Bias adjust pots are located 
on the front panel beneath the corresponding LED.  This gives me an constant 
indication of the quiescent cathode current through each of the output tubes, 
and allows me to monitor and adjust the grid bias without the use of 
external voltmeters.

The whole amp was built on Bud chassis, with one for the amplifier, and
another mounted at right angles for the power supply.  Perforated panels
were used for the top, bottom, sides and rear.  The front is a 19 inch
rack panel.  Every piece of metal was hand formed and cut, which by far
took most of the time to construct this beast.  The circuits were laid
out on copper clad perforated board, with tracks cut away with a Dremel
Moto-tool--this also took a lot of time.  Next time I do this, I'll make
a PC board.

I had to play around a bit with the feedback compensation capacitors.
I needed to raise the value of this capacitor by about a factor of 3
so that the amp was stable at high powers.  At low power, no oscillatory
behavior was evident.  Seems that the gain/phase relationship changes 
with output power...hmmm.  Changing the compensation is to be expected
however when one substitutes a different output transformer for the
specified model.

I love the way these amplifiers sound.  Very smooth and silky, with a
slightly rolled off top end (about 1.5 dB down at 20 kHz) and flat 
to down below 5 Hz!!!!  The output power is just over 80 WPC into 4 ohms
at onset of visible clipping.  No problems with capacitive or inductive
loads too.  Unfortunately, I don't own a distortion analyzer so THD and 
IM numbers aren't available.  But I am happy with how it sounds when music 
is run through it.  It provides good detail on acoustic and classical pieces, 
and has enough guts to handle Nirvana and the Pixies.

Now, on to the next project...I hear that Hammond has a 250 watt output
transformer....

Let's see, 6 KT88's in push-pull parallel should just about do it...

--------------------------------------------------------------------------
Richard Clewer | 
clewer@hns.com |