A Class AB Power Amp

I decided to order a 3U chassis kit for the DYI Audio Store for this build since it matched my class A amp and has a decent amount of heatsink area. These chassis kits come unassembled but have excellent quality metal parts. You can choose various options like the front panel thickness, color, handles, etc. I went with a 10mm aluminum panel.

Here is one channel PCB assembled. Soldering the parts took a couple of hours. The strange-looking device in the center is a high-current choke which is part of the Zobel output network needed for stability (that is, keeping the thing from becoming a giant oscillator). 1/4" Faston tabs are soldered to the PCB for ease of connecting everything together.

There are a few SMD capacitors that I soldered to the board, but you do have to use these. The BOM I referenced has through-hole substitutes which should work just as well. The white Molex connectors are there to plug in the MOSFET output devices using "flying leads" soldered to their terminals. The large red capacitor at the top left is a high-quality 4.7 uF Wima film cap for input decoupling.

Here is a view of the rectifier/capacitor board. The PCB came from DIY member Prasi in India. I thought I would try a different approach with this amp and use an LT-4320 based "perfect rectifier" instead of the traditional diode bridges. The LT-4320 is an IC that controls 4 MOSFETs in a bridge arrangement and causes them to be synchronously switched "on" with the incoming AC from the transformer. This results in lower noise and also has the advantage of nearly zero voltage drop and power dissipation, due to the low "on" resistance of the MOSFETs. (around 0.005 Ohms for the devices I used). The LT-4320 data sheet is here. I am using 4x22,000uF 63 Volt caps for filtering.

Tapping holes in one of the heatsinks in order to mount the power transistors. If you have not done this before, you need to go slowly, use a lubricant and back out the tap when too much resistance is encountered. Clean out the thread, then run the tap back in gently until the threads are deep enough.

Here is one completed heatsink/power transistor assembly. There are two large MOSFETs bolted to the heatsink. Attached are "flying leads" which provide some flexibility in mounting. The MOSFETs are insulated from the heatsink with Keratherm insulators and mounted using fender washers and M3 hardware. The fender washers serve to distribute the pressure from the mounting bolts across the face of the MOSFETs, preventing stress fractures. There is a smaller transistor in the middle which is the VBE bias multiplier. Mounting it to the heatsink provides thermally stabilized bias for the output devices. Not visible are the series gate resistors - these are soldered directly to the gate pins on the MOSFETs and covered with heat shrink. I prefer this to placing the resistors on the PCB because of the inductance of the flying leads. If the MOSFETs were to be soldered directly to the PCB (common in some designs) placing the resistors on the PCB would be fine.

Laying out the major components on the steel baseplate. Not all parts conform to the grid pattern, so some drilling is needed. The gold object is a center punch to mark a new hole for the terminal strip, which will be used for the AC input leads. The baseplate is an option when you order a chassis from the DIY Audio Store, but they are inexpensive compared to the cost of the chassis and add rigidity to the whole assembly. Plus, they give you a lot of flexibility to make changes to the layout. Normally, I would prefer a symmetrical approach with the toroid in the center away from the active circuits, but there was not enough room.

Now the heatsinks get bolted to the baseplate  and the flying leads get plugged into the PCBs. At this stage, a bench power supply can be hooked up to each channel for testing. After the channels are verified, the next step will be to install the toroid and hook up the rectifier/filter board, then test the output voltages from it. Once all that checks out, wiring from the rectifer/filter board will be connected to the amp boards.

Checking the bias and output offset voltage after ramping up the + and - voltages from my bench supply while looking out for the magic smoke and burning odors (none detected, thankfully). This is the advantage of a bench supply. Mine has dual tracking outputs and current limiting, so I can set the current limit very low to make sure there are no shorts or misbehaving semiconductors as I increase the voltage. The meter on the left is measuring the voltage across one of the .22 Ohm degeneration resistors in the output stage. .029 Volts/.22 Ohm = 132mA, which is pretty close to the optimal quiescent current for this amp. The test of the other channel was also good. My bench supply only goes to 32 Volts, but the amp supply will be +/- 50 Volts, so the final bias adjustment will happen then.

Mounting the back panel components: a Neutrik powerCon input for AC, a fuse holder, on/off switch, speaker binding posts, and RCA input jacks. I carelessly scratched some of the paint on the back panel during drilling, so had to spray paint it with some wrinkle finish flat black Rustoleum.

Back panel wiring - ready to integrate with the chassis. The AC ground is bonded to the back panel with a star washer and a lead terminated in a ring terminal for connection to the main chassis.

Here is most of the chassis wiring done, including the back panel. All the power and speaker wires are 14-gauge silicon insulated. The transformer will come last, along with tidying up the wires a bit more. Note the AC terminal block with two surge-reduction thermistors and a 500 Volt safety cap across the line and neutral.

Here is the nearly finished (but working) chassis with the transformer mounted and hooked up. At this stage, it is playing music!

Interesting note for builders: when I hooked the inputs to my preamp I noticed a slight hum in both channels when the preamp attenuator was set to different values. Zero hum at min and max, but some in-between. My first thought was that there was something wrong with the amp since my class A amp did not exhibit the problem. But then I remembered that in the preamp, I had all inputs and outputs floating (not grounded). This did not cause a problem with the class A but with the new amp, a ground loop was rearing its ugly head, The fix was to install a ground loop breaker in the preamp. Ground loops are a plague to builders, worth another article.