Bradley Worley about
My take on the Szekeres Class-A amplifier Subtle tweaks and a really nice enclosure


It had been on my wish list for some time to build a MOSFET amplifier for a pair of decent quality low-impedance (50 ohm or so) headphones, and this circuit finally came along and spurred me into action. I did a quick test-run on a breadboard to gain familiarity with the topology and circuit fundamentals using an IRF630 and some stock resistors I had around. Quite shockingly, it worked rather well, given the paucity of heat sinkage... :P

First design

As a sanity check, I designed and built an essentially 'boilerplate' CCS-loaded Szekeres amplifier, and to my surprise, nothing melted! gEDA gschem and PCB were used to make the circuit schematic and printed circuit board layout. I also used Berkeley SPICE for the first time to estimate the circuit performance. Pictures of the first circuit design follow:

Second design

Because the first design came together so well and actually worked, I was greatly encouraged to revise the design to squeeze as much performance as possible from the beautifully simple Szekeres topology. I decided on a higher rail voltage of +15 Vdc and a 'hotter' biasing voltage, I increased the input AC coupling capacitor from 1.0 uF to 3.3 uF, and I increased the output AC coupling capacitor from 470 uF to 4700 uF. And of course, the pictures:

In order to try and quantify the effects that my modifications had on the performance, I used some bash scripts and gnuplot to compare the SPICE simulation results of the two designs. Needless to say, I was quite pleased to see that gain, headroom, frequency response and phase response were all improved. :) (Trust me, I don't claim to be the first one to make these changes, but the circuit feels more like my baby now, if that makes any sense...) The comparison results:

Amplifier materials

All parts for a single channel of the Lassie amplifier can be purchased from Digi-Key.

Single amp channel bill of materials
Part Description Quantity Price ea. Subtotal
4672K-ND mica TO-220 insulator 2 0.11 0.22
495-4010-ND 3.3uF polyester film cap 1 2.36 2.36
P10263-ND 4700uF FC elecrolytic cap 1 1.97 1.97
P4671-ND 0.47uF poly film cap 2 0.18 0.36
P100KCACT-ND 100k metal film res 2 0.17 0.34
P4.70KCACT-ND 4.7k metal film res 1 0.17 0.17
P68.0KCACT-ND 68k metal film res 1 0.17 0.17
P150CACT-ND 150R metal film res 1 0.17 0.17
P5.1W-1BK-ND 5.1R 1W metal film res 1 0.17 0.17
IRF610PBF-ND IRF610 N-ch MOSFET 1 1.10 1.10
LM317TNS-ND LM317 linear regulator 1 1.86 1.86
Total: 8.89

Power supply design

Since I didn't intend to power the amp from my 13.8 Vdc bench supply forever, I had to design a power supply. A requirement of all class-A amplifiers is a low-noise power supply, as they have zero PSRR. taking cues from gainclone and Morgan Jones' supplies, as well as some scraps of experience, I put together a basic regulated power supply:

Power supply materials

All parts for the regulated power supply of the Lassie amplifier can be purchased from Digi-Key, except the toroidal transformer, which I bought from Parts Express. The bill of materials follows.

Regulated supply bill of materials
Part Description Quantity Price ea. Subtotal
67-1916-ND spst switch with yellow led 1 3.21 3.21
P11247-ND 2700uF 35V FC electrolytic cap 2 3.27 6.54
P10321-ND 47uF 50V FC electrolytic cap 1 0.28 0.28
P10275-ND 470uF 35V FC electrolytic cap 1 0.57 0.57
P4671-ND 0.47uF polypropylene film cap 7 0.18 1.26
CMF1.10KHFCT-ND 1.1k 1% metal film res 1 0.17 0.17
CMF100HFCT-ND 100R 1% metal film res 1 0.17 0.17
445-3741-1-ND 3.3uH 3.8A radial inductor 1 1.80 1.80
1N4001FSCT-ND 50V 1A general use diode 2 0.84 1.68
MUR820GOS-ND 200V 8A ultrafast diode 4 0.11 0.44
497-2982-5-ND LM350K TO-3 3A regulator 1 7.41 7.41
122-605 30VA 15V+15V toroidal trafo 1 25.80 25.80
Total: 49.33

Construction notes

Sticking to what I knew, I developed presensitized boards in sodium hydroxide and etched with ferric chloride. However, given the fact that I had to make the boards in my dorm, I was game for trying new things. First, I successfully incorporated a 25W red safe light. Second, I exposed the circuit boards through two epoxied transparency prints, instead of one, to make the final traces perfect. I also drilled the through-holes on a mill and cut the boards to size using a bandsaw.

The case construction was started once I had laid final plans for the circuit board layout. I first used a band saw to cut a rough block (approx 4.5"x3"x1.75") down to approximate size and then I used a shell cutter to precision mill the block to 3.5"x2.0"x1.5". I drilled through-holes for the phone jacks (3/8" dia.) and DC power jack (7/16" dia.) and then made through-holes for the side-panel mounting screws using a #25 bit. The mounting holes were then tapped for #10-24 screws. On the top and bottom of the case, I drilled out TO-220 mounting holes using a #25 bit and then, using a 1/4" endmill, milled out room for the screw heads to fit flush with the case walls. Finally, I hollowed out the interior of the block using a 3/8" dia. end-mill.

Mounting the circuit boards inside the case was tricky, since electrical isolation was required. The boards were secured by two stainless steel #6-32 hex head machine screws through the two TO-220 packaged components, with mica insulators underneath them and a nylon nut above them, ensuring isolation. A section of teflon tubing was also cut to thread onto the screws between the screw threads and the TO-220 tabs, and the mounting holes of each TO-220 component were widened with #19 drill bits in a vise. The conductive traces on the bottoms of the circuit boards were isolated from the case with spray-on conformal coating.

The mirrored finish on the aluminum was achieved by first using 600, 1500, and 2000 grit wet sanding paper under a faucet, with a sandal for a sanding block, and then applying a white polishing compound until the part was shiny. :) I was able to pick up the above supplies and terry cloths for polishing from a local AutoZone, and the finish, as is visible in the photographs, is quite spectacular.

Finally, the wooden sides were cut from a stock 0.5" thickness red oak board using a table saw, and then sanded down to the precise dimensions of the aluminum housing. Before finishing, through-holes for mounting were drilled into the wood using a #3 bit and countersinked. I milled a 0.2"-deep recess into the inside of each wood panel with a 3/16" endmill. I then applied a first coat of stain to the wooden panels before filling the pores of the oak in with minwax filler and applying the final coats of (minwax cherry) stain. Green adhesive felt feet were added to the undersides of the wood panels to keep the aluminum chassis section off the table surface for better cooling.

The power supply enclosure was a bit trickier, as more parts were used in the final case. Three 1/4"-thick aluminum blocks were cut to rough size and endmilled to 1.15"x3" (1) and 5.25"x3" (2). Eight mounting holes in each top and bottom face were made using a #3 drill and a 5/16" endmill for the #10-24 screw heads which would be used. The same tools were used to mill a pocket for an M5 bolt to mount the toroidal trafo to the bottom panel. Vent holes were milled using a 1/8" endmill in the top panel. The small aluminum block was drilled to hold the TO-3 voltage regulator such that it is isolated from the other two aluminum panels. Wood side panels were cut and mitered to dimensions required to form the outer walls of the case. Both materials were finished using the same techniques used on the amplifier case.

Listening notes

My first impressions of the finalized amp were made with 32 ohm Philips phones, which I had thought until now were better than average. To be safe while burning in the output electrolytics and heating up the two active elements per channel for the first time, I stuck with the Philips. Once I trusted the reliability of the amplifier and power supply, I plugged in my new Sennheiser HD595 phones for the real tests. Honestly, the combined amp and phones sounded beyond amazing. It felt as though someone had taken hold of the left side of the frequency response curve and ripped it off. Bass was stronger and more tightly controlled, mids were warm and clean, and highs were defined without being too bright. I know the topology isnt for everybody, and it isnt the lowest distortion out there, but the sound, in my humble opinion, is wonderful.

But then again, I'm biased. ;)

Amplifier pictures

Some pictures to prove the amplifier exists and works! :)

Enclosure pictures

Some pictures of the enclosure I built from scratch for the amplifier, thanks to the patient instruction of Scott in the Texas Tech Chemistry department machine shop.

Power supply pictures

Some pictures of the power supply and its enclosure I built for the Lassie amplifier. I had to match the beauty of the amplifier itself, so yet another aluminum-and-oak enclosure had to be fashioned for the supply too.