NOTES

• By far the greatest expense in an amplifier is in the cabinet and power supply components – transformer and filtering capacitors. Next in cost are heat sinks and final driver transistors. In this implementation, I built two 3 channel amps because the incremental cost in going from a 500 VA transformer to a 750 VA one (large enough to power three channels) is not that great. Further, there is next to zero extra cost cabinet wise, once the decision has been made to make cabinets from a full sheet of aluminum (see Custom Cabinets). Lastly, with 6 channels on order, I gained significant volume price breaks on heat sinks and transistors. Ultimately, the cost of adding a third channel is very small, all things considered, and a really nice to have to drive a centre sub-woofer. The cost in going to six channels, while not trivial, was still much lower on a per unit basis than building just one tri-amp. In these days of theatre surround and biamped systems, is 6 channels enough?
• I wanted maximum heat dissipation capabilities ruling out locating the heat sinks internally, but also I did not like the idea of heat sinks simply hanging off the back of a cabinet. As a result, I ventured down the route of custom designing the cabinets. The heat sinks are totally exposed in a vertical position, but nicely integrated with the top, bottom and sides of the cabinet (see Pictures). In retrospect, I did get the performance I wanted but custom fabrication of good-looking cabinets is really a labour of love – the cuts must be exceedingly straight, dimensional accuracy in the order of +-.010 inches, and finishing requires many, many steps. If you go this route, add an extra month to the project!

• The cabinets were made from .063" 3003 grade sheet aluminum and are 17" wide (19" faceplate), 13" deep and 5.57" high. This is a common grade which can be anodized even though it is not specifically designed for that. I chose not to anodize due to the large expense and because I wished to match these cabinets to those of an existing set of amps.
 
• Some tips on construction after you have drawn up your cabinet plans:
1. A 4'x8' sheet can be used to make 3, and possibly even 4 cabinets.
2. Most suppliers will also cut the sheet to an accuracy of +-.005", for a modest charge. I recommend use of this to slice the sheet into strips (eg. One strip for all sides and back pieces, one for front pieces, one for bases and tops, etc). This ensures dimensional accuracy and saves on cutting the sheet, and makes transportation a breeze. All that you must do is cut the strips into the appropriate lengths - to make a side only 1 cut is required by you in the workshop. If you plan to anodize or not paint the aluminum, you must instruct the supplier not to mark up the material when they put it in the brake – generally they are probably not too particular about this and can ruin the perfectly smooth mill finish of the 4x8 stock.
3. ¾" x ¾" angles can be used to attach sides to the base and the sides to the top piece – if one wishes to avoid screws in these critical areas, crazy glue or cold weld material (like JB Weld) work well bonding clean aluminum to aluminum. By "clean", I mean it has been sanded with 320 grit to remove any oxide - A must!
4. Two difficult items with custom cabinets are making vent holes and any rectangular holes. In this implementation, I created vent holes by drilling about 100 holes in two circular patterns each approximately 6" in diameter. (see Vent Hole Pattern). I obtained the amount of venting space needed, plus if done carefully, it looks even better than slotted holes. For square holes (for switches, AC plugs or fuse holders) simply cut the hole smaller than what is required and file back to the desired size. .063" aluminum is easy to file and removing 1/8" is not difficult at all.
5. Two extra feet are needed to support the transformer and prevent the base of the cabinet from sagging in the centre. This is a very easy and inexpensive solution when using the relatively thin .063" sheet. Alternatively, one could order .080" aluminum or prepare some means of reinforcing the base.
• I painted my cabinets black to match existing equipment. To get a reasonably professional looking job, I first used an undercoat of flat black, then a partial coat of gloss black on top. This gives a mottled look if one is careful in applying the topcoat sparingly and evenly. The use of toner transfer paper to make colour decals for the front gave a real nice finishing touch.
• An extra polished aluminum front plate was added (smaller in width and in height than the front face) in order to give the allusion of a smaller cabinet - the 5.58" height of the front face plate (5.50" needed to go with vertically mounted Wakefield 423K heat sinks) can look awful big, especially if done all in black. This approach made the amps appear smaller and provided some interesting contrast.

• The instruction's provided by Professor W.M. Leach are excellent and I did not have to deviate from them at all. A big caution! The interface between the drivers to the pcb is the most difficult aspect of this project. Wires must come from behind the pcb (solder side), route to the front, and then the board must flipped to solder. Be sure to leave room in the amp and enough wire to be able to do all of this. If you don't, well, you will have a rather large problem on your hands in connecting the drivers, dc rails, and grounds to the pcb. A revised pcb layout, with all wire connections placed on the edge of the boards and some sort of riser for making a custom cable connection, would be a really nice improvement. As it is, there are wires ducking under pcb mounted heatsinks and care must be taken that wires do not contact the main heat sinks.
• Test every component for its appropriate value and that it is in good working order before installing. One has a choice here, test one component at a time before it is on the board, or possibly hunt for that one component (out of about 100 or so) that is DOA or incorrectly shipped/selected. Hopefully, that failed component didn't take anything else down with it! The choice is yours. My experience - I found a few 7.5 volt diodes mixed up in my 20.0 volt diodes which were picked off the shelf at the supplier incorrectly. If not caught prior to installing, I may have experienced large dc offsets and the source of the problem not an obvious one. On all 6 channels that I did build, because I tested before installing and even after installing (to ensure no heat failures or backward connections), every board worked flawlessly from the get go. A great relief, and ultimately the pre-testing saved much time. The choice is yours...
• Follow instructions on matching input differential amps and matching diodes D13..D16 to give 40 volts. I matched Q1,Q2 pairs and Q3,Q4 pairs to yield dc offset voltages in the order of 5 to 25 mVolts at the output, which is nearly imperceptible in sound.

• Costs were remarkably low - less than 225.00 CDN$/channel or 150.00 US$/channel. Volume purchasing for 6 channels, augmenting the power supply to drive 3 channels, custom cabinets, judicious use of surplus suppliers, and a special sale of Motorola transistors by Newark, all contributed to keeping costs down.
• I can vouch for Professor Leach's comments regarding turn-on thumps - they don't exist with this design and there is no 1 or 2 second wait. Very nice.

MODIFICATIONS

• 20,000 uF per rail (40,000 uF in total) was used for the supply to each 3 channel amp, versus the minimum recommended 8,000 uF. I was fortunate enough to obtain 4x 10,000 uF Mallory caps at 7.50 Cdn$ each after I purchased an initial set of 10,000 uF caps. The bass response may have been helped by the extra energy available to the rails.
• 750 VA, 80 VCT Plitron transformer (model 10A026000). The larger size was used to accommodate the 3rd channel.
• C7: 0.1 uF, 50 volt polypropylene cap (Panasonic ECQP1H104GZ, Digi-Key# P3104-ND) replaced 0.1 uF 50 v metal film. The polyp cap is large but can fit on the board without too much trouble.
• C9: 100 pF (2 connected in series to yield 50 pF), 50 volt polypropylene cap (Panasonic ECQP1H101JZ, Digi-Key# P3101-ND) replaced the 47 pF mica cap. Both C7 and C9 are in the feedback paths (C9 in the high frequency feedback). The polyp caps may have helped in providing an even nicer top end to the amp.
• I biased each channel at 150 mA rather than the minimum of 100 mA. Again, just using my ears, I felt the higher bias resulted in a richer and less tinny sound.