Passlabs Amplifier Projects

The Aleph5 - Construction

The following discussion is aimed at non-EE type people (like myself), who may want to contemplate this sort of project but lack some of the basic electrical theory. The "learned few" will find it almost insulting, so you've been warned ;-)

Note: many of the recommendations are specific to a class A, constant current draw amplifier - class AB may vary.

See larger image ...OK, starting from the beginning we need a power supply (see Nelson's Article). The Aleph5 has 34VDC rails and has a total dissipation of 150W/channel, or 300W total. When estimating transformer requirements there are a number of recommendations:

  1. Always use a toroid if you can.
  2. When choosing the AC secondary voltage you need to allow for voltage drop across the rectifier bridge, plus the internal resistance of the windings, etc. "Rules-of-thumb" vary, but you should not use the ideal value of VDC = 1.414 x VAC. Nelson recommends between 1.2-1.3, as do most other designers / engineers.
  3. The "VA rating" of the transformer needs to be much greater than the constant power dissipation of the amplifier. Here at least 2x the constant power rating is an absolute minimum, with up to 3x being desirable.

So, I settled for a 650VA, 28-0-28VAC toroid from Tortech, who make good quality toroids. This is just adequate on the total power dissipation front, but this company is conservative in their construction and in designing the case the additional diameter of the 800VA unit (which I would class as "ideal" for this amp) didn't fit quite as nicely .... yes, even DIYer's have to make design compromises!

You will note from the picture above that I included a high current (10A) line filter to remove any "trash" from the AC mains. I think if you are building one of these amplifiers this is a justifiable expense.

See larger image ...You will note from the Aleph schematic that Nelson used only 1 bridge rectifier. There have been a number of discussions about this on the diyAudio forum and the concensus seems to be the use of 2 bridges, arranged in the following manner (diag.). You will note left and above that these are mounted on their own heatsink, which I have mounted on a threaded-rod which also secures the toroid.

The next step is filter capacitors. Here you can spend quite a lot of money and at some point you get into the land of diminishing returns in terms of extra $$$ versus power rail ripple/hum. The original Aleph5 used a total of 100,000uF, which is probably adequate. I happened to have a supply of larger caps, so used 4x 80,000uF for a total of 320,000uF. This is actually overkill, and it makes current draw at start-up even more of a problem (uncharged a large cap acts as a short-circuit across the bridge), so the in-rush current limiting thermistor was an essential. The final rail voltage under load is 36VDC.

I considered using Pi-filters, as I had done in my Aleph4, however the design layout of the 5 really didn't suit this, so I thought I would try the amp without filter chokes first and add them if I found the rails contained too much ripple.

I stole the idea of using brass plate for the supply busses from pictures of the X-series amps. The brass off-cut cost little and I simply brushed it with some 180 "wet-n-dry".

Heat dissipation is another major factor in designing/building any Aleph series amplifier. To quote Nelson, "there is no such thing as too much heatsink", or the other memorable quote, "we dont do calculations, we just make 'em really big!".

Back to reality, this can be one of the most expensive parts of the amplifier, both in terms of dictating final size and cost. For we mere mortals calculations really are helpful.

If we start with the power dissipation, we need to account for 150W/channel of heat. As a rough rule we do not want heat sink temperatures greater then 60C, so we need heatsink with a thermal resistance as follows:

  • Assume ambient Temp = 20C
  • For a 60C maximum, we need the temp increase to be <= 40C
  • Therefore, heatsink Rt = (40/150) = 0.266 C/W (max) per channel

The original Aleph5 sinks ran at about 55C and were temperature protected at 70C. We can afford to go slightly higher than this, with the understanding that the higher the FET temperatures, the shorter the life of the devices. I was fortunate to find a heatsink manufacturer in Oz who is happy to sell direct to the public. The unit I used (diagram) was 300mm x 150mm with 40mm fins and dual flanges for device mounting. The heatsink Rt = 0.25 C/W, which was right on the limit, however I would have the rest of the case for additional heat dissipation.

NB: The final amplifier runs a little hotter than I would like (70C) and I think if you are going to build along these lines you would be better-off using 2 of the 200mm heatsinks of the same profile, ie. for 400mm length cf the 300mm sinks shown in these pictures. This should get your amp down into the 55-60C range which is a more appropriate. You can safely run the FETs at 70C, they will simply not last as long. The amp works fine in the current format.

See larger image ...The selection of a heatsink with dual flanges lent itself toward using 1 ledge for both the current source devices and the other for the output devices. For my Aleph4 I had mounted the FETs separate from the main PCB, however I had the idea for this design to use the mounted FETs as the support for the PCB and make the PCB design (stuffed view) encompass the complete amplifier module.

Again I decided to make the PCBs myself (link), and despite 1 foul-up the end result turn-out well. I found a good source (cheaper than Farnell) for positive resist PCB and then used the developer/etching solution left-over from the Farnell kit that I used to make my Aleph4 ..... very bad plan, do not try this :-(

Using the complete kit (boards/developer/etching solution) from Kinsten worked just fine!

If you would like a copy of the artwork, you can download a TIF file here.

Given the modular basic design of the heatsinks, the obvious case structure was to have these as the sides of a standard approx. 19" case. The front panel I decided to make from 8mm aluminium sheet. I used 6mm "socket-head screws" to mount this to the front aspect of the heatsink modules - these have a head diameter of 10mm, which I recessed into the aluminium panel by 5mm. Since I wanted this amp to "look the part" I decided to have the front panel engraved. This can be done by any number of "photo-engraving" businesses; most will be able to take an "EPS" file or something similar. Once engraved I filled the lettering with black enamel, then proceeded to brush the front using "180g wet-n-dry" paper.

Prior to brushing, I mounted the LEDs either side of the logo, recessing a 5mm LED from the back so the front of the LED was just above the surface of the panel, then used a 45deg. countersink bit to make a bevel. Once the surface was brushed I coated it with clear satin enamel spray-pack to protect the aluminium, which would have tarnished with fingerprints etc. if it were left raw.

I was really quite pleased with this .... eventually! During the first run an insect landed in the 2nd spray enamel coat and I had to strip the whole thing and start again :-(

You can download a PDF file outlining the chassis construction, PCB layout + stuffing, plus parts list here.

See larger image ...The back panel was reasonable straight forward. I used gold-plated RCA jacks, Amphenol XLR connectors and gold-plated speaker terminals, all from Alltronics. Since each side was an amplifier module, I kept the AC mains in the centre rear (I always place my AC switch at the rear - aiming to minimize the mains path internally). As for my Aleph4, I combined the switch and over-currentprotection by using a 5A circuit breaker, which then led to a TDK high current noise filter before reaching the transformer.

The top is well ventilated having multiple slots cut prior to powder-coating. You will note that the heatsinks are 25mm above the amplifier base. There were several reasons for doing this, partly to improve ventilation and heat dissipation (note the side of the base beneath the heatsink has ventilation slots also) but mainly because the height of the 80,000uF filter caps mandated additional internal height!

The sheet-metal dimensions for the base, rear and top are also in the above PDF file.

There's more to come ..... please visit again soon ;-).