Making a Leslie 147 Valve Amplifier
I had been hoping to buy for a 147 amplifier in distressed condition so that I could renovate it. Two appeared on Ebay and I bid but they went for much more than I was prepared to pay, given the changes that would be needed to fit into my ex-710 Leslie so I made one. Here it is.
Leslie replacement transformers are available in the US at a reasonable price but I have not managed to find any here in the UK. The cost of postage from U.S makes them just too expensive for my taste. I found some suitable transformers on Ebay; not Leslie transformers as such but the right electrical characteristics. I designed out the OC3, used solid state switching for the motors and mounting the h.t. smoothing capacitors internally. The chassis is therefore a little different to the real thing, though the footprint is the same.
I deviate from the original design further in that two printed circuit boards are used. The h.t. circuit board has the rectifier, smoothing capacitors and OC3 replacement circuit. The motor control board performs motor switching, a delay-on h.t. relay and provides a 24v d.c. supply. This second PCB uses a PIC processor for logic and delay timing. I opted for alternative sockets: an IEC connector for the mains input; Bulgin sockets for the motor connections (the motors are 240v and the Leslie design sockets are only 110v rated); a 5-pin XLR socket for the organ connection; a 4-way Speakon socket for the speakers, allowing 8 ohm and 16 ohm outputs. Additional fusing is also provided. Despite these deviations, the amplifier proper remains faithful to the Leslie design.
The H.T. Circuit Board
To keep costs down and ease component sourcing difficulties, I opted to replace the OC3 gas discharge valve with a MOSFET. The OC3 drops the h.t. voltage to the screen grids by around 100v. The volt drop remains more or less constant irrespective of current. Initially, I designed a MOSFET circuit to do the same but quickly came to the conclusion that a straightforward voltage regulator would be preferable; A fixed voltage drop would pass all of the noise on the input to its output, requiring additional smoothing to remove it whereas a voltage regulator would remove much of the input noise, reducing the required smoothing components. A simple source-follower circuit was chosen. The diodes provide some protection.
The sourcing of high voltage electrolytic capacitors is difficult and expensive so I opted to use series capacitors for the first after the rectifier and thereafter 450v devices. In addition, a delay-on circuit was provided so that the h.t. would not turn on until the valve heaters here hot, without which, the voltages would reach around 550v in the warm-up period with the transformer I have.
The rectifiers, smoothing and regulator components were all mounted on a PCB.
Motor Switching, H.T. Delay and 24v Supply
The Leslie is to be used with a Hammond XK3 which requires a 24v signal from the Leslie to detect its presence. Without it, the internal Leslie simulation is turned on. I chose to generate this from the heater supply using a Boost Converter. Triacs are used for the motor switching and a PIC processor provides the logic control for the motors and the delay-on h.t. timing. Here is the circuit. However, additional 10k ohm pull-up (to +24v) resistors are fitted on the input socket.
The h.t. switching uses a relay as the switching element. It switches the supply before it is rectified. It is rated for the ac voltage and probably would break the current if required to do so however, it only closes with the voltage on and opens after the ac is removed. It should then last indefinitely.
Making The Chassis
Having got the transformers and decided upon most of the components, I designed the chassis and drew it up in Qcad for Linux. Ebay provided the steel plate, 1.2mm thick - perhaps a little thinner than ideal but as with all design it is a compromise; in this case between strength and ease of bending. I have a bender, a home made affair constructed of 3 inch square sections of ash. Here it is. A trial bend on a short length produced a reasonably sharp bend but a 500mm wide bend was too much for it. Scoring the inside of the bend produced better results; not perfect but adequate. A Dremel fitted with a round ended dental bit, guided against a metal straight edge performed the scoring to a depth of about 0.5mm.
And here it is in action, bending one of the sides of the chassis.
The steel was cut with an Eclipse sheet saw. It takes a hacksaw blade. The important thing when sawing thin materials is the provide adequate support for the material and to use a fine toothed blade. 24 TPI is OK here.
Here is the main plate ready for bending. Note the score lines.
Here, left is the above plate with the sides bent and right, the finished chassis in the raw. The ends are fastened with pop rivets
The Finished Amplifier
Here it is.The chassis was sprayed with cellulose paint using spray cans. I did try Hammerite but I made a terrible mess of it and burned it off with a blowlamp, wasting a couple of days labour. The output valves are KT88's, second-hand, purchased from Ebay and great value. They are almost identical to 6550's. I love these big valves. It seems a shame that they will be hidden inside a box. The transformers are not so lovely.
Underside view. There are few components in the amplifier proper making point-to-point wiring a simple job. The Leslie design is about as simple as a push-pull amplifier can possibly be.
The two printed circuit boards are shown here. On the right is the motor control board with its relay for h.t. At the top is the h.t. board. The heatsink keeps the MOSFET cool. The smoothing capacitors look minuscule compared to can types but the ratings are more than adequate. The cost of the components is comparable with their size.
A 147 Leslie has a 16 ohm bass speaker whereas a 710 has a 4 ohm bass. My output transformer has 8 ohm and 16 ohm tappings but not 4 ohm, so I changed the bass speaker for an Eminence 15 inch Beta model that I had kicking around. Testing it with the transistor amplifiers, I noticed a reduction in low bass compared to the original, not the mid frequencies handled by the 15 inch speaker but the low end of the pedals. It indicates that the cut-off frequency is higher than previously, though I believe the old and new speakers have similar bass resonance frequencies. I may investigate in the future but for now it will do.
Because the bass and treble speakers are differing impedances, they are driven from different transformer taps and the crossover component values differ from the original; the bass inductor halves and its capacitor doubles in value.
Here is the new amlifier installed in the leslie cabinet. The Leslie 770 transformed to a Leslie 147 or something approximating a 147.
I had tested the circuits in stages as I built the amplifier, so I knew it ought to work. I connected it in the 710, converted to a 770 and by this amplifier to a 147 and tried it out. The slow motors failed to run. A wiring error was put right, then everything worked fine.
Anything I say about the sound is highly subjective, much like talking about ones own children. Even so, I am obliged to say something on its performance..... The sound is a little softer than the transistor amplifier making the high notes a little less shrill. In overdrive it distorts in the manner of a real Leslie amplifier giving what is affectionately called "growl". However, the volume level at which growl kicks in is far too high for use in the home. The neighbours would revolt. The MIG preamplifier attached to my Leslie 760 distorts quite differently, the overdriven sound being poor in comparison but it will do it at low levels.