Casing for a 41Hz AMP6

I built a 41Hz AMP6. This was because I needed a new amplifier, my brother (who cares about such things) suggested a Tripath-based one, I couldn't find one built that was quite what I wanted, but I could find the kit from 41Hz. It uses the chip I wanted, it was (nearly) all through-hole components, it looked reasonably straightforward.

So I bought it (painless and straightforward) and built it (fairly painless, but I did need a new soldering iron - it really does need a 50W temperature-controlled one). (On the other hand, any justification to buy new tools is a good thing, in my opinion.)

Then I needed to put it in a case. It's just a power amp (no controls), so it could be a plain grey die-cast box down the back of some furniture. However, a nice bit of copper pipe caught my eye on ebay...

The Amplifier and Modifications

It's a small compact board. This is built in accordance with the instructions (very nearly - you can play spot the difference if you like).

But one of the things I wanted to do was have a brighter fault light and also have an illuminated switch to indicate status. The switch I wanted to use needed a 12V supply. The board is designed with a couple of MOSFETs that switch status LEDs, but the circuit is such that it only puts about 3V across the LEDs on board, even though the amplifier has a 14.6V power supply.

So I took the transistors, current limiting resistors and LEDs off the board and instead took the 14.6V supply and the gate signals from the board on hookup wire and routed them off to a daughter board that would give me 12V switched by the signals from the chip.

This shows the mods schematically. The circuit snippet to the left is as on the amplifier board, and on the right is what I've put on the daughter board.

This is actually reasonably easy to understand, even if you don't know what you're looking at, at a simplified level:

The transistor is the thing with three connections labelled D, G and S. On the circuit board photo above, the pads are laid out in order D G S from top to bottom. It works like a switch - for this particular transistor, when the voltage at G is more than about 2V above the voltage at S, the switch is on and electricity can flow from D to S. The wire 'X' goes off to the amplifier chip, which either applies 5V or 0V.

If X is at 0V, the transistor is off in both cases, so no current flows and the LED (triangle with two little arrows) is off. D is at 12V, S is at 0V.

If X goes to 5V, the transistor switches on. In the existing case though, as current flows, a voltage occurs across the LED and S rises from 0V. When it gets to about 3V, G is no longer more than 2V above S, the transistor switches off, current stops flowing and the LED goes off, the voltage at S drops, which means G is now more than 2V above S, so the transistor switches on again. Actually, it doesn't oscillate, because the transistor is not a perfect on-off switch, it has a smooth (but quick) turn-on. In reality, it settles with about 3V at S, limiting what you can get across the LED (or whatever else you substitute) to about 3V.

In the modified case, when X goes to 5V, the transistor switches on, but since S is fixed at 0V, the transistor stays fully on regardless of what happens, because G stays 5V above S. With the transistor on, D gets pulled down to 0V, and the full 12V is available across the LED.

Casing Build

This is a picture-heavy build sequence. As noted above, I spotted a nice looking copper pipe on ebay, and thought it could be big enough (105mm diameter, 210mm long) to fit the amplifier in, so I thought I'd give it a go.

Amplifier. Some aluminium angle (50x25mm, 6mm thick) that I bought (ebay) several metres of years ago specifically to make heatsinks, initially for some high-power LEDs I was playing with, but it's also handy for power transistors and so on. Some 30x10mm copper bar bought (ebay) specifically for this project.

Shaping the copper bar so it fits...

nicely in the copper pipe.

(Incidentally, using an angle grinder on pure copper is a delicate exercise - rather akin to trying to cut cheese with a circular saw. I roughed it out with the grinder, but did quite a bit with a coarse file.)

Drill the aluminium to take the heat sources. Counter-sunk holes for the fixing bolts, for reasons that will become apparent.

Flatted the copper face and the aluminium just by rubbing on fine wet-and-dry paper stuck down on a pane of glass. (This is overkill, but may as well do a job properly).

Solder a couple of bolts through the copper bar (with their heads recessed) then solder the shaped copper bar and a couple of 6mm brass rods into the pipe. This is just plumbing solder, though it takes quite a hefty blow-torch to get the copper pipe up to temperature - it's a very effective heatsink, since there's about 1.3kg of basically pure copper here, with quite a large surface area.

I like the colours.

Pretty, pretty.

All this gets cleaned off, of course - it's the residues of the flux.

I used a template to make end-caps. That's made from MDF by the simple expedient of taking a piece, screwing it down to a board on a router table and rotating it past a straight cutter. That gives me a circle with a hole accurately in the centre.

The template is a neat fit in the tube.

Double-sided sticky tape attaches pieces of 5mm acrylic to the template.

Then a bearing guided cutter on the router table takes off the excess.

The resulting acrylic disks are a snug fit in the tube, resting against the soldered-in bits. This looks a bit scratched and battered partly because the flash catches all the surface marks, and partly because it is. For the visible end-cap I did almost the same process, but started by sticking thin plywood to each face of the acrylic to protect it during all the machining operations.

Some shaping of the disks (by saw and file), and mount fused terminal blocks on one of the disks gives me this transformer sandwich,

and that fits in the bottom of the tube.

At 3'o'clock on this you can see I added an earth connection to the case. Given how conductive the case is, I decided a reliable earth connection would be a good idea, so soldered on a short bolt.

View from the bottom.

The top cap (the visible one) is in black acrylic. This is treated more carefully than the other disks. I stuck thin plywood on both faces to do all the machining and drilling, then stripped it off as late as possible.

The red bit at centre is a 5mm flat-top LED carefully glued (cyanoacrylate) flush in a snug hole. This is the overload / fault light controlled by one of the daughter board transistors.

The back is not quite so tidy. There's a doubler that has a tapped hole in it plus a nut glued on (cyanoacrylate). I think either the nut or the tapped hole would give enough strength, but I've put both in. The doubler is solvent-bonded to the disk, and the speaker connection posts also bolt through it. The LED is obvious.

The amplifier is mounted to the aluminium angle, and the hookup wires noted above loop round to the daughter board. This is the same spec transistors as supplied in the kit, but wired up differently (as discussed above). There's a 3V zener lurking under the red heatshrink (mostly hidden) dumping some of the 14.6V supply to get it closer to a nominal 12V. The black heatshrink covers the current-limiting resistor for the face-plate fault LED. The jumper at the top edge of the board duplicates the jumper you could put on J2 pins 7 & 8 on the amplifier board, which I used while testing before a switch was wired in.

The daughter board mounts on the end of one of the screws holding the amplifier to the aluminium angle.

The connections are fitted to the front plate, and then wired up to the amplifier. Since the wires cannot be guided as the face-plate is installed, I put heat-shrink on everything to avoid any shorting out.

Testing that it works all wired up before installation.

Now for assembly. Heatsink compound on the flat copper face.

From the top, I slot the amplifier into the case and it drops over the bolts fixed through the copper bar. The photo is out of focus where it matters, but you can see the bolt projecting through the aluminium angle. Fitting a washer and nut over that 'blind' (working just by touch, stretching through or around the wires in the way) was a bit fiddly, but not too bad (worse is to come).

Then the face plate is pushed into place.

I happen to own a rigid endoscope (ebay - where else?). It's surprisingly handy surprisingly often. In this case, since all the wires between the face plate and amplifier just get scrunched up, I checked that there wasn't any visible straining or potential shorting-out.

There wasn't. This is what I expected would be the tightest fit - it's the signal pin of the inputs just clearing the top of the aluminium angle. They clear, but this is why I put heatshrink on.

So the endoscope just told me that everything was fine - I didn't make any changes as a result of what I found with the endoscope.

Wire it up again and check everything still works with the faceplate in place (before trying to get the transformer in the other end).

Nylon studding screws into the nut and tapped doubler on the back of the top plate.

At the bottom, the power supply wires also just scrunch up as the disk is slid into place. In this case, however, there's no access once the disk is in. This is checking that they appear to curl up tidily and let the disk seat fully. (A benefit of using clear acrylic for the lower disks, though actually it's just what I had most of in my scrap heap).

This photo shows just how close everything is inside - note how tight the amp circuit board is to the disk here, how close everything is at the top (see the endoscope picture above), and how the power terminal blocks have to be staggered around the other innards.

This is more overkill. I decided to shield (as much as I could reasonably easily) the transformer from the amplifier. This is probably redundant because toroidal transformers don't radiate badly anyway, but I'm not entirely convinced putting it all in a copper tube won't focus something somehow. So I cut a lobed disk of thin copper sheet that is incorporated in the transformer sandwich, between transformer and acrylic. It just relies on being a tight fit between the flaps and the copper tube.

So then I can fit the transformer to the disk (this time including a mains in lead), and connect up the power wires. The mains lead has strain relief just by a snug-fitting grommet with a tight cable-tie around it.

Slide the transformer home.

This was actually by far the fiddliest step in the whole assembly, and I had to take it out and put it back three or four times before I think the disk has properly seated inside, without trapping wires. Even now I'm not 100% certain, because it's all completely hidden from view.

The transformer is eccentric, but I wanted a gap round the edge of the base disk for rubber feet. To accommodate this the nylon studding is just forced to bend.

A word about the studding - even if it could follow a straight path, I'd use nylon. A potential hazard with toroidal transformer is that if you use a metal bolt through the middle, and it shorts to a metal case round the outside, it acts as a one-turn secondary. You get a very low voltage, but at nearly infinite current capacity, and shorted. Bang.

With a copper case (lowest electrical resistance there is, short of solid silver) and a relatively narrow gap between mounting disk and case, I was worried that a conductive path right round the outside of the transformer would be a real possibility. Hence, I wanted to ensure that there's no conductive path through the middle, so needed a non-conductive bolt, hence the nylon studding.

Fit base plate (came with the transformer) and tighten nut.

Done

Finished. I'm very pleased with this.