PLEASE NOTE: This is a work in progress. I’m doing this build with my daughter to teach her some electronics from the dark ages. It may be a while before it’s completed
I do occasional repairs to old tube-based guitar amps. I grew up fixing tube-based radios and TVs in the ’60s. I thought that it would be interesting to revisit some of the older amplifier designs and create reference implementations to play with.
These days, power and output transformers as not in much demand so the prices are quite high. Because I planned to create several test amplifiers, I thought it might be better to build one power supply and have one output transformer so that the amplifier itself would be simpler and lighter and easier to work with on the bench.
So a while ago I rewound a couple of transformers that had overheated and had winding shorts. One of them was then used to build my bench power supply, and the other was used to create a multi-impedance output transformer. I also made myself a speaker to use in the shop which has a multi-stage power divider so I can run an amp at full power and no go any deafer than I am already. In addition I also have an oil can resistive load that I can use on 100 to 200 Watt power amps.
Before we get started, this is what a Gibson GA-5 looks like. These pictures were from a an amp sale site online.
And I suppose we should take a look at the schematic. These amps were built when radios and TVs were the most common electronics items. The schematic was included with the amp so that a “reliable radio man” could repair the amp. Sadly those days are no longer with us.
We’ll go over the design later.
For the amp build I wanted an easy way to build out the design on the bench and test it easily. I wasn’t interested in putting it in a box or making it pretty. If I like one of the designs well enough at the end I have the option to do a full build and put it in a cabinet.
The basic chassis design is a bit unconventional. I no longer have any kind of machine shop beyond the drill press I have in my garage along with some woodworking tools. When I work on an amp chassis it’s usually upside down supported on two ‘T’ blocks made of 3/4″ thick wood. For this design I thought I’d build the supports into the chassis. Sounds a bit weird I know. I was also horrified at the price of sheet aluminum (OK so the prices were at Home Depot bit none the less …) so I found an alternative. I’m going to say right now that I think this may be the biggest step forward in tube amps in decades. The material I decided to use for the chassis floor and front panel is called Aluminum Composite Material (ACM). This is two thin sheets of aluminum on either side of a very dense plastic core. The outside of the sheet is colored. Mine came with an unprotected grey back side, and a nice shiny white front side with a tear-off plastic cover.
This is how I sketched out the chassis design:
The ACM is easy to cut by scoring both sides with a knife and then bending it t snap it. A little burr removal and it’s pretty spiffy. The big issue was deciding how big to make the floor of the chassis. I wanted it to be reasonable to work on but not too enormous. To figure it out I laid out most of the components on a paper cutting board.
The ACM sheet I bough form Home Depot was 24″ by 12″ so my front panel became 12″ wide and the chassis 10 1/2″ to fit inside the two 3/4″ wide wood end blocks which I cut from New Zealand pine (from Lowes). I made a couple of aluminum angle end pieces, drilled holes for M3 screws and stood it all together to see what it looked like.
The wood ends are tall enough to clear the EL84 tube which is the taller of the two tubes in the design.
I decided what front panel layout I wanted and where I wanted the tag boards under that chassis. The tag boards were set 2″ apart which is a bit tight for the big electrolytic caps but is about as wide as you can go with 1/2 W resistors.
As I mentioned above, this build has no power transformer or output transformer. I have a bench power supply that I’ll use and a bench output transformer that I wound that has multiple primary impedance taps. We” see those later. (And at some point I’ll write them up and add links).
The final touch before sitting my daughter down with a soldering iron was to letter the front panel. In the dark ages when I worked for the Evil Empire in Redmond, WA. I had access to Visio. I love Visio. I could make very nice panel layouts, print them and glue them to the aluminum with 3M phot-mount spray. Now I’m a poor, retired engineer working on a Mac I use a simpler approach to my bench instruments – a Sharpie!
The big grommet is for the wires that go to a big multi-plug on the power supply. I should probably talk about that a bit. I built a power supply that will be adequate for a 50W tube amp. It has 250 V and 450 V HT outputs, -60 V for bias, and of course 6.3 V AC center-tapped to ground for the heaters. I also included +12 V and -12V in case I decide to put some of that modern semiconductor stuff in the preamps. The front of my power supply has a big terminal strip and a bunch of 4 mm banana sockets for the outputs.
To make it easy to connect and disconnect several preamps and power amps I added a couple of tails to the power supply that use 9-pin Mate-N-Lok connectors. These are a pin of a pain to assemble but make it easy to have everything in one plug. The pin numbers on the moldings are hard to read so I color coded them with pain before assembly as it’s quite a pain to remove pis that are put in the wrong holes.
Getting back to business here, this is what the underside of the chassis looked like after fitting the front panel components, the tube sockets, and the tag strips.
The first task for my daughter was to assemble the power cord with its plug (as seen above and then attach the ends of the power cable to the tag strip on the chassis. At this point she’s been transported back to the 50s when every TV and radio was built this way.
Next up was adding the heater wires. We used green 18 GA stranded wire for these which is pretty standard. The high voltage wiring in the power cable was silcone wire rated at 1 kV which is very flexible. Inside the amp it will all be 22 GA solid core with the usual 300V PVC covering.
To avoid putting two wires into those little tags on the 9-pin tube bases we ran two twisted pairs from the large tag strip where the power cable comes in.
The great thing about having the heater wires in place is that we could plug in the tubes, plug the power cord into the bench supply and fire it up. A few cautionary remarks about the fact that we’d now have 250 V DC on the chassis and it was time to turn it on.
It’s hard to see in the pictures but both tubes lit up. My daughter was impressed. A few hours of instruction and work and we had something to see working.
Before we go any further. Let’s look at the schematic for what we’re building. I put this together in Eagle. I use Eagle for lots of electronics designs. It’s a little clunky but gets the job done. I have the free ‘home use’ version which has a few limitations but is otherwise very good.
Please note that this is the schematic for my ‘bench’ version which uses an external power supply and output transformer. This also has a couple of extra jacks labeled send and receive. I added these so I could test some tone control circuits. NOTE: The output form the send socket is way too much for an effect pedal. I’ll publish a more complete version that includes the power supply components and output transformer later.
The first part of the real amp assembly was to add the power supply smoothing caps and resistors. I spent some time earlier laying out all the parts on the tag strips to be sure we had enough room to space things out. I wanted to be able to get at every component for measurements/alterations/whatever. The two rows of 20-way tag strips provided just enough connections without having to double anything up or mount parts on the tube bases or front panel components.
The next two pictures show the first couple of capacitors and a power resistor added to the tag boards along with some HT and ground wiring. The final picture shows the rest of the components laid out to see how it’s going to fit.
After deciding where all the components were going to go on the tag boards, the next step was to solder all of them in place. After that, we started adding the ground wires to everything that required a ground. The following pictures show the results of part of this work. Not all the grounds were in place for the last picture.
The rest of the ground wires were added followed by the power distribution from the smoothing capacitors. Just before we tested the power wiring for real, I realized that we had no high voltage bleed resistor so I added a 470 kohm across the second smoothing capacitor just so that the amp would discharge rather than sitting there waiting to zap one of us after a test run.
By this point my daughter has gotten her soldering skills back. Her previous experience was when she was about 10 years old. So it had been a while. Progress isn’t very rapid because we’re only working for an hour or so each day but it’s satisfying to see the work progress. We are now at the point where we can start adding the signal wiring. The plan is to bring the amp up one tube stage at a time. We’ll start with the input triode and end with the EL84 power tube.
After all the components and power connections were in place, the next task was to wire up the input stage to the first half of the ECC83.
We put in the ECC83 and fired up the amp on the variac for a quick smoke test, Then up to full voltage for some DC measurements. With no EL84 in place there is very little drop across the power chain resistors so the anode resistor of the first stage was at 220 V and the anode at 108 V. The cathode voltage as about 1.2 V giving us a current of around 0.54 mA through the tube which is right where we want it.
Happy with the DC values we put a 600 mVp-p sinewave into the input and observed about 45 Vp-p at the anode giving us a gain of 75 for the stage. As a final test we reduced the input voltage to just a few millivolts and hooked the effect send output to my test amp. A little U2 played through the input sounded good.
Next up is to finish the second stage (the other half of the ECC83).
MORE LATER …