Aleph 4 Strickly DIY Project Build
After many requests I am now starting a thread to show how easy it is to build a STUNNING PassLabs Aleph 4 on a reasonable budget.
A quick visit to the http://www.kk-pcb.com/index.html web site will demonstrate where it all started.
I looked at the Aleph 2 - a stunning amplifier in its own right.
I looked at the Aleph 5 - which you would imagine to be better than the Aleph 4.
They are all EXCELLENT amplifiers, the choice is yours.
I chose the Aleph 4 as it had rave reviews in the Press and I already had one 35-0-35V 500VA Toroidal transformer going spare.
You have to appreciate that there are three main areas where all High-End amplifier construction can get VERY expensive.
1. HEATSINKS - ALL Class A Amplifiers need MASSIVE heatsinks.
2. TRANSFORMERS - ALL Class A Amplifiers need MASSIVE transformers.
3. PSU CAPACITORS - ALL Class A Ampliers need MASSIVE PSU capacitors.
NEW, these components alone can add up to well over £1000.
SECOND HAND, this is a completely different ball game.
So, the thread is a build project for an Aleph 4, within a shoestring budget.
Where appropriate I will give the source and cost of the new components for those that "Would never use Second Hand Stuff".:cool:
After choosing the amplifier the first concern must be the heatsinks. This beast uses an enormous amount of power and most of it is used to heat the living room.
The Aleph 4 is biassed at 2.4A and uses +/- 48V DC Power Rails.
Therefore we have 24 MosFets each dissipating approximately 40W each - a Grand Total of 960 W of heat to warm the living room.
In the perfect World it would be nice to keep the MosFets below 70 Degrees C in order to ensure a degree of reliability.
In the UK, I am going to use an average ambient temperature of a modest 20 Degrees C.
Therefore we have a maximum temperature rise on the heatsinks of 70 - 20 = 50 Degrees C.
In all my maths I am going to round up the less eloquent figures to make the maths easier, this also allows for inaccuracies in manufacturers data.
The whole amplifier must therefore dissipate 50 / 960 = 0.05 Degrees C / Watt.
I haven't manged to find a supplier that can produce a naturally ventilated heatsink that can dissipate that amount of heat, but don't fret, we also have not got a single 960W source of heat.
What we are going to try to acheive is a heatsink that is equivalent to 0.05 Degrees C/watt which can also accomodate 24 MosFets.
7 Heatsinks at 0.4 Degrees C/Watt each mounted on a disipating Aluminium Chassis.
0.4 / 7 = 0.06 Degrees C/Watt - Close enough to our ideal.
Dont forget that the top and bottom of the case will also be dissipating heat as well as the connector panel.
Ignoring these components, they just help in the equation of heat.
It's important to keep the MosFets in close thermal equilibrium to each other.
In a flat World, it is recommended that 10mm plate should be the minimum to allow each device to share its heat with any attached heatsinks.
In the real World 10mm aluminium plate is EXPENSIVE. However we are only after offcuts and its relatively easy to cut with a JigSaw. After all, we're not after a professional finish at this point.
I was lucky enough to purchase 7 off Heatsinks from E-Bay, each was 197 x 160 x 40mm. They cost me £40.00, an absolute bargain.
What I needed was a substantial 10mm aluminium chassis to attach the heatsinks to so that everything shared the heat load.
I managed to buy several scraps of 10mm aluminium and cut them to
10mm plate 396x160mm
10mm plate 396x130mm
10mm plate 376x130mm
10mm plate 376x130mm
50x50x5mm angle 130mm
50x50x5mm angle 130mm
50x50x5mm angle 130mm
50x50x5mm angle 130mm
10mm plate 416x416mm
3mm plate 416x416mm
The 130mm cut plate is not critical, nor are the angle pieces. The 160mm plate is more critical as it forms the back plate of the amplifier, but hey, it's at the back.
The actual visual chassis is formed by the heatsinks themselves, very much like the original, and heatsinks are generally cut to fairly finite tolerances.
It's a fairly simple job to build a 4 x square aluminium frame, using 4 cut pices of aluminium and the 4 angle sections. Care must be taken to ensure EXCELLENT thermal contact between all the surfaces. Every hole must be de-burred and all surfaces must be assembled with thermal paste.
At this point it is getting heavy but it is already a very sound structure. If you hold it together with clamps before drilling you don't even need to be too exact where you drill the holes, I used 3 x 4mm countersunk bolts per join. As long as the bolt heads lie under the surface of the aluminium, so that the heatsinks can lie perfectly flush, you don't have to be exact here either.
In the photo you can see the Input and Output sockets.
Here we do have a little engineering problem. The plate is 10mm thick and the sockets just aren't deep enough to accomodate such panel thickness.
The ON/OFF switch - no problem. Choose a round button and mount the switch using appropriate bolts.
The IEC connector - similarly no problem. Carefully mark the cut out and drill holes at the corners of the cutouts. Drill a huge hole in the centre and attack the remaining aluminium with a JigSaw until the tiny remaining bits can be gently filed to a snug fit.
The L/S and Input Phono Connectors needed a different approach.
The Phono Connectors, I've mounted close to the corner so that they are actually mounted on the 50mm corner bracket. The outer 10mm plate only has a 15mm clearance hole drilled in it. The L/S connections are simply counterdrilled from the rear with a 15m drill so that the connctors fit flush with the back panel.
I know its cheating but all the machining so far is inside the cabinet and out of view.
The next job is to attach the heatsinks.
The ones that I bought aleady had 4mm holes for attaching them, but this is fairly simple.
At this point, unless you have access to the experience or tooling required to tap aluminium, I didn't bother. Mine are mounted using through bolts from the outside to the inside.
If you clamp the heatsink to the frame and measure carefully where the exit hole is going to be, simple through bolts will be fine. DE-BURR everything and ensure a cleam METAL to METAL contact with a light coating of theraml grease.
Note that the thermal grease is only there to ensure that the two metal faces join correctly. Too much and you are defeating the obect, you are actuallymaking a thermal barrier which is exactly opposite to what you are trying to acheive.
OK. That's the majority of the case sorted out.
The next task is the electronics.
The first job has to be producing the PCBs themselves.
At first sight, this may seem an utterly impossible task to the true novice. However, it is a relatively straightforward job using many tools that are commonly available.
What you will need:-
1. Pre-coated PhotoSensitive Copper Clad GR-4 Glass Fibre Board.
Commonly available from Farnell, RS and even E-Bay
2. Caustic Soda Crystals. (Approx 12 Teaspoonfuls)
3. Plastic Washing Up Bowl or Food Tray x 2.
4. Ferric Chloride Crystals. (500g)
5. Assortment of 0.8mm, 1mm and 1.5mm Drill Bits.
6. Electric Drill - Nothing Special - I use my trusty Black & Decker.
7. InkJet Clear Acetate Sheet.
Available from Staples or E-Bay
Nothing extravagent or impossibly expensive so far. Be aware that you will need many drill bits as they blunt and snap frequently.
The first step is to produce your PCB artwork.
I downloaded an article from http:\\Mark Finnis' Pass Amplifiers At the top of the page is a tab labelled DOWNLOADS and from there you can download the schematics and the PCB artwork.
At this point I was a bit sceptical about the Main PCB artwork. In my mind there didn't seem to be enough copper for the high current tracks. I was also going to be using non-inductive plate resistors for R42-R52 which have a different lead spacing to those intended on this layout. During my trawl for 220uF / 35V Tantalum Capacitors, I discovered that SMD devices were considerable cheaper than their leaded brethren. Finally, there has been much talk about the input differential pair Q1A1 and Q1A2. Ideally these two should be kept in close thermal equillibrium to each other. To this end I rotated Q1A2 through 180 degrees to that the transistors are now "back to back".
Although minor, I was also concerned that there was a ring of copper all the way round the circumference of the original artwork. EARTH LOOP shouted at me, I've made a small break in the loop.
Using the printers HIGHEST setting, it's now a simple job to print a completely opaque image at the right size on a piece of Inket Ready Acetate.
Pack the wife off for a few hours as we are now going to use the kitchen as a photo processing lab.
To start we need to mix our chemicals.
1. For Developing the exposed Photo Board we need a solution of 1 Teaspoon of Caustic Soda to 1 Litre of Cold Water. DO NOT make it any stronger or you will completely remove the photo image.
2. Ferric Chloride - Follow the Instructions on the packet. It will tell you that this needs to be used at 50 Degrees C but this is not too important.
3. Caustic Soda for cleaning the Photo Board - 2 Tablespoons of Caustic Soda in 0.5L of water - Or just a strong solution.
KEEP the PhotoBoard in subdued lighting. It doesn't need to be BLACK. KEEP the protective coating on the board until the last possible moment. If possible shut the curtains and use a RED light bulb. No need for a special DarkRoom bulb here.
When cutting the board, allow at least 5mm around all the edges, and 10mm where an edge has been supplied already cut. This is because daylight will creep under the protective sheet and cause the edge of the board to become unuseable.
Cutting the board is easilly accomplished with a fine tooth Tenon Saw. A Hacksaw can be used but it's difficult to keep the edges nice and straight.
Exposing the Photo Board.
In the perfect World you will now need an UltrViolet Exposure Unit - EXPENSIVE.
In the DIY World you simply need a source of UltraViolet Light.
I borrowed the next door neighbors sunbed. It would be amusing to take the six boards to the local tanning salon for a quick roasting.
If you are lucky enough to live in the sunnier climates, good ole sunshine will also do the trick but you will have to experiment with exposure timing.
So one last final check.
We have Acetate Image and the ink is thoroughly dry.
We have WEAK Caustic Solution.
We have UV Light Source.
We have Ferric Chloride Solution.
Peel the protective coating off the Photo Board and place the acetate image on the exposed surface. MAKE SURE YOU GET THE IMAGE THE RIGHT WAY UP. It is a good idea to tape the acetate to the board to prevent it moving.
IMMEDIATELY place the Photo Board on the UV light source and expose for approximately 8-10 minutes. It is worth conducting a few test pieces to get the exposure exactly right before starting.
When the 10 minutes is up, remove the acetate and place the Photo Board image side up in the weak Caustic Solution. Like magic you will see your image appear on the board. Gently rock the solution until ALL the inwanted mask is removed and you have a clear image on the board.
Remove the board from the solution and give it a good wash under cold running water.
Place the exposed board, image side up in the tray of Ferric Chloride Solution.
Gently rock the tray continually until all the unwanted copper has been removed. Check that all the copper between closely located pads and tracks has been removed. The time taken for this depends on how warm and fresh your solution is.
When the board has been etched give it a good wash in cold water and inspect it.
If you need to etch it a bit further do so now. If not give it a wash in STRONG Caustic Solution. This will remove all the photo resist.
Give it a good dry and it's ready for drilling.
Most of the holes will be 0.8mm, but your components may differ from this.
Because you are drilling Glass Fibre, the bits will blunt fairly quickly, normally indicated by them snapping. Place the etched board on a piece of scrap wood and carefully drill all the holes.
VOILA - FINISHED
The 4 x MOSFET boards are produced in exactly the same way.
.400A X 48V = 19.2 watts per fet
I don't think your heatsink contact will be sufficient unless you lap the surfaces and use 9 to 12 bolts on each.
All the surfaces are lapped. I take your point about the number of bolts though.
If the 10mm plate is doing its job properly it should be able to dissipate the heat to the centre of each heatsink where the heatsink is most efficient ?
10mm is sufficient for heatspreader , considering height of them ;
be sure to have good contact between Al plates and L profiles , too .
some thermal goo between them is a must , same as between heatsinks and plates
|All times are GMT. The time now is 12:27 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio