Amplifier chassis prototype

To complete the Vacuum tube amplifier project, I am going to need a custom chassis built as there is nothing available to purchase that will meet the requirements.

I designed and rendered the required parts in SolidWorks, and then got some prototypes manufactured.

amp render with tubes

This view shows some tubes installed.

amp render no front

The front and top panels removed showing the circuit boards

amp renderpreview

This rendering has no tubes installed, which shows the holes that need to be cut for the transformers and tube sockets, as well as the LCD and volume knob.

The parts arrived from the prototype manufacturers. I must say I am impressed with the precision of the parts. The surface finish is not so great, however I’m not really concerned about this at the moment, as I sacrificed the surface finish to reduce the cost of the prototypes significantly.

The front panel was CNC machined from an 8 mm aluminium bar stock, while the top and rear panels were laser cut from 2 mm aluminium sheet.

The screws which hold the case together do not quite sit perfectly, as I needed to countersink the holes for the case screws by hand because the laser cutter cannot cut countersunk holes. This should not be an issue with the final parts as they will be cut by machines more precise than myself.

Front panel with LCD
Close up of LCD
Rear panel showing unbalanced inputs, 1 speaker terminal, serial port and power socket
All tubes and transformers installed

Once I am happy with all the dimensions of the cuts, I can proceed to order new parts with a nice surface finish, perhaps anodised aluminium.

Amplifier chassis prototype

Prototype SMD assembly

SMD assembly is not as hard as it looks, and you can get pretty good results with hobbyist grade equipment in your home workshop. A microscope is nice to have, but not a necessity.


I got some of my boards made over at OSH park. They are very nice quality, and pretty affordable. The only downside to the OSH park PCBs is the left over tab on the edge. Can’t complain though as it is a prototype PCB and if I was going to do a batch of more boards you can get the boards routed out properly, with their medium run service.


Nothing a quick file can’t fix.


A quick wipe with IPA and the board is ready for applying solder paste. I really like the ENIG finish too.


I applied the solder paste by hand with a syringe. It doesn’t have to be perfect.


I highly recommend using a nice pair of SMD tweezers, as they will make your life so much easier. They are also non-magnetic so the components don’t get stuck to the tweezers.

All the components are placed. Again this doesn’t have to be perfect, as when the solder re-flows, the surface tension will pull the components into place, most of the time. The board is ready to go in the re-flow oven.


I place the board on some scrap boards to keep it off the grills, which can act as a heat sink and components directly above the metal grill might not get hot enough. I also tape the thermocouple with some Kapton tape to the scrap board, to avoid disturbing the components on the board you just placed components on.


The re-flowed board. A few things still need to be fixed up though, because I applied the solder paste by hand, some pads had too much paste, others not enough.


The re-flow is complete, and everything looks pretty good. Could have used a tiny bit less solder paste on R12, R13, R9. It’s pretty easy to wick a bit up though.


There are also a couple of bridges on the QFP100.


Flux is your friend.


A quick touch with the iron, and you can drag them to one edge where it is easier to wick away the excess.


I like to use a curved tip.


All bridges are removed.


Also there might be a few solder balls hanging around.


Which are easily removed with a nice pair of tweezers.


The reworked board, with all excess solder removed. It’s covered in flux. Time for a clean.


I placed the board in an ultrasonic cleaner, with Electrolube Safewash Super SWAS I can’t recommend this stuff enough, it works great, and after 15-20 minutes in the ultrasonic cleaner.

I rinsed the board off with filtered water (to reduce water spots if you have hard water) and placed the board into the re-flow oven again. This time with a temperature of 80-90 degrees C for about 10 minutes, which is enough to evapourate the water, but not enough to melt the plastic connectors. You have to be careful though, not all plastic connectors are created equal.

Take the assembled PCB out of the oven and let it cool. The board is now immaculately clean.

Prototype SMD assembly

Preamplifier and volume controller

Quite a few years ago I had a plan to build a Vacuum tube amplifier, but I also wanted to add some modern features not usually included in the old school valve amps, such as digital volume control via a rotary encoder, input selection via a serial RS-232 connection, as well as an LCD display, or possibly even a VFD (vacuum fluorescent display).

I finally got around to designing and manufacturing a printed circuit board. My first design from scratch actually.

I chose to base the design from a PGA2310 IC and started from there, I ordered a DIP package from Digikey. I also had an Arduino mega2560 laying around, so I was able to connect them up with a breadboard, and do some basic testing.

I love the look of SMD boards, and a lot of the new chips these days are only being manufactured in SMD packages, so I didn’t want to waste time making a through hole board. Also soldering SMD components is not a problem for me at all, as I actually prefer them over through hole.

Although, in hindsight I should have made the electrolytic capacitors through hole, as the SMD versions are just too fragile, and can quite easily rip the pads straight off without much force at all (which I had to learn that the hard way.)

altium 3d top

The final design, rendered in 3D with component bodies. I couldn’t figure out why the QFP100 was purple with green pins though.

altium 2d top

2D view showing the top layer and ground planes.

altium 2d power plane

2D view showing the +5 V power plane, and +15/-15 V polygons


The rendered image from OSH Park


Printed a 1:1 paper model to double check footprints


Finally received the purple package


Test fitting the through hole components. They all fit perfectly!



Inspecting the tracks and solder mask through the microscope. The solder mask is slightly offset, but nothing to worry about.


Yes, it actually works. Doing some initial tests with Arduino to test the hardware is working, uploading the program using ICSP header. I plan to write the final code using Atmel studio.

I also tested the audio section, and it’s working perfectly so far. I also plan to do some audio measurements at a later stage, and see how it really performs.

Although this process has taken me quite a long time to do, I have learnt so many things, as well as now when I start on a new design, I will do many things differently. For example, I would use smaller connectors, and populate the board more densely to save on space (and therefore money.)  I would also try to use a smaller microcontroller if I was going to make more than a few boards.

Preamplifier and volume controller


After looking into some articles about embedded linux, I came across a blog by Henrik Forsten, which I found really interesting and I thought it would be a good way to get into building, assembling and one day, designing some boards by myself.

I grabbed the design files from Henrick’s github, uploaded them to OSH Park, and ordered a set of 3. I also ordered a solder paste stencil from OSH stencils for the top layer (as I thought it would have been pretty tricky to apply paste via a stencil after the top layer had been populated.)


Setting up the stencil, by taping it to the desk


Line up the stencil, tape only one side, and add some paste


Spread the paste into the holes. Be careful not to lift up the stencil while spreading, otherwise the paste can spread over too much.


Carefully remove the stencil, and place the components with tweezers


Place the board into the re-flow oven. I taped down a thermal couple onto another PCB I used to prop up the board I want to re-flow


Try to get the board as level as you can, so the components stay in the middle when the solder re-flows.


I used Henrik’s re-flow board, running a python script to handle the PID control loop. A AU$30 oven from Kmart works pretty well.


The top layer, with all the SMD components in place. I decided to do the USB-A connector later by hand, to make adding the components to the bottom layer easier.


Applying the paste to the bottom layer by hand.