50W Amp project, no more cheapo.

downhere

Member
2001-10-03 10:24 am
I've asked around for information about transformers and boxing, and I've found some sites that have really good ideas. After looking at my old amp chassis, I found that it only accepts transformers 150VA to 300VA (that amp was an old class A design, pretty nice with elna audio caps). It seems a waste to spend 100 or so bux on 20W so I've decided to do the 50W project instead.

1) Interesting case idea. Plastic with foil.
I read somewhere that you could actually shield electromagnetic interference pretty well with a thin layer of aluminium foil stuck with spray adhesive covering a plastic box(these are pretty common) and attaching chassis ground to the aluminium foil. This sounds interesting, and I'll try it with the 20W amp at another time. the result can be sprayed, painted, whatever. So I think it'll save money.

2) I've run into lotsa problems with the layout of the board. An interesting idea was to solder the parts to the breadboard and use conductive pens to draw traces. Any comments on that? I think it'll save alot of time and effort. I probably won't even have to solder, a dab of superglue to hold the parts down and I can draw all the traces I want. Comments? Will it affect the sound?

Other interesting ideas were to point to point the circuit in free air and use casting resin to hold the parts together. It'll be very durable, although I don't know about heat problems. Mounting ICs and heatsinks are still a problem tho. Any ideas on that?

Oh I can pretty much order any transformer I want. Seems that a transformer winding factory is around my area. They only do EI types tho. Should I shield the transformer with heatresistant plastic covered with aluminium foil to prevent electromagnetic interference?
 
Why use plastic coated with some kind of thinlayer of metal (From spraycan).
You will get better result with copperfoil or iron and it is cheaper.
You have to shield your transformer if is near your ampstage. It can be done by building wall's in your box with feedtrough for your cables.
If you want to draw a lot current (more than 100 mA) in the traces on your circuitboard then you have to do a real circuitboard. I would prefer min. 35um copper. but 70um or 105um is better.

Regards

Sonny
 

downhere

Member
2001-10-03 10:24 am
Thanks...
What if I just drew thicker lines?
I'm also contemplating real breadboarding... Buying a 1 inch thick chopping board and driving nails into it to hold down my transformer, using a nail as a star ground point, and nailing down my heatsinks. the circuit can be constructed by supergluing the components down, and connecting lead to lead or lead to wire. Sounds interesting? Probably looks wonderful too if I can glue it properly.
Thanks for the copperfoil idea, I have no idea where to get that tho. I was thinking fine wire mesh instead of aluminium foil, but if aluminium foil does the same thing, why bother shaping wire mesh.
I think the nails into wood idea is rather promising... can imagine binding posts superglued into position with nails for support. Really long and thick nails could also provide some sort of vibration dampening.
Wonder how hard chopping board wood really is tho, wonder if it'll splinter or crack. I'm thinking of getting them from IKEA, cos they're known for using lotsa soft pinewood for stuff.
 
Hi All,

Having been through the stage of thinking 'As long as it makes sound, who cares what it looks like', can I say, please please spend more effort on the construction. So much of the pride in building your own person piece of audio nirvana comes from the build quality, and being proud to show it off to others.

If you want to cut costs, do so where you can, but try to do the following:
Make the effort to layout and etch a PCB. All that is needed is the PCB blank, a permant felt tip marker, and the etchant. The result is physically more sturdy than anything else.
Alternativly, use the 'dead bug' method of construction. Get a peice of copper clad board and use that as your ground and solder components above it. hint: use 10Mohm resistors as stndoffs and supports. Mount IC's upside down (dead bug).

Simple tools for modifying casework can be brought cheap. A nibbler is one piece of equipment I find invaluable (although an aching hand usually results!). Don't glue heatsinks, that's just inviting disaster (most glues get soft when hot). Use screws and bolts.

Cheers, Adrian

PS I almost forgot, saftey must ALWAYS come first.
 
downhere said:

I read somewhere that you could actually shield electromagnetic interference pretty well with a thin layer of aluminium foil stuck with spray adhesive covering a plastic box(these are pretty common) and attaching chassis ground to the aluminium foil. This sounds interesting, and I'll try it with the 20W amp at another time. the result can be sprayed, painted, whatever. So I think it'll save money.

I did this a couple of times, and as long as you're tidy when laying out the aluminium foil, it works great. Just remember, if you're going class-A, you just can't forget the heat disspation. Most modern plastics take heat quite well, but i wouldn't be too comfortable...
 

downhere

Member
2001-10-03 10:24 am
Thanks, yeah I agree good looking stuff definitely makes you feel good. Remember the look of the tube amp kit they were selling at the electronics shop, machined aluminium with matching knobs and all.

Problem is, I don't have a workshop, and there aren't any workshops that'll do metalworking for you at a low cost. Last time I asked, labour was 40, price of materials extra.

I'll consider the dead bug method, but won't it look as bad as the gluing it to the board method? Sounds interesting tho, I can imagine using thick enameled copper wire to hold the board. Does the copper keep its shine? If it does then I'll definitely use it as ground.

I'm considering supergluing white plastic feet to the wood as a sort of mounting. Probably looks bad I noe, but it'll allow me to remove the board easily. Are there any other ways of mounting PCBs painlessly?

I was thinking that brass nails would match the colour of the chopping board, as well as the orangy brown of the transformer... Then I'll fine wire mesh the transformer, and then the entire chassis to block EMI. Or would a more cagey sort of barbecue mesh look nicer. (they all probably suck.) If you were me, and you'd be mounting it on a piece of wood, how would you shield the parts from EMI without spoiling the looks?
 
downhere said:
How many layers of foil did you put?
adhesive used?
pictures?
: ^)

Dunno if you asked me, but whatahell.. this is MY method, crude, slow, but it works :) Basically you lay a large piece of common, kitchen aluminium foil over the box and press it over the sides so it takes the internal shape, more or less. Now, you remove the foil, cut the excess material and (here's the tricky part), slowly and carefully, you glue it to the box, pusing it against the walls so it lays as smooth as possible. I use common superglue and this works very good, specially if you use the side of the foil feels like plastified (at least the one i use is this way). It might take more than one piece, but as long as all the pieces touch each other, the shielding works.

I did this on a couple of boxes and a guitar cavity and works wonders on almost any surface. Bear in mind that the foil is very easily scratched, so be gentle the whole process. AND, you can't solder over Al, so you'll have to come with some way of attaching a ground wire to it. Anyway, it's easy, cheap, and it works.
 

downhere

Member
2001-10-03 10:24 am
Paulb, i've looked at the transformer prices,found that the cost of building a 20W amp and a 50W amp is roughly the same, so I'm going for the 50W. It'll mean scrapping the 1 channel I completed, but it'll also give me a chance to try and think about the chassis before I construct the circuit.

Gonna cut excess LM3876 pins off with a lead cutter, and just tape, stick, wind wires to the leads. I saw them selling "Jumper wires" at the electronic shop, looks like 22/24 awg wire pre stripped and tinned , and it sells really cheap too, 2.50 for a rather large bag. Is that any good? the leads are about 4 inches long, 1 inch stripped at one side and half a cm on the other. I'm thinking of nailing the board with say half inch brass nails, and just winding the leads of the components together on the nails. any extra connections I'll just use the jumper wire + solder or conductive pens to connect it. I think I can make it look pretty much like the schematic this way. Trying to avoid soldering altogether... its bad for health. :)

I think aluminium foil would look cool, as well as copper foil. Aluminium foil is definitely easy to use, there's also a shiny side and a dull side. I think I'll use expanded polystyrene glue (Por by UHU).

For filter caps, any reccomendations? I'm using 2200uF x 4 and I can find Elna caps, Rubycon caps, Jamicon caps in my area. Which brand/type is the best? Should I get low inductance, long life, minature.... They probably be connected by enameled wire to keep the look... infact, won't clear enameled copper wire look the best for wiring my board? ... Any opinions on that?
 
Hi downhear,

About shielding a transformer against emitting EMI. This is not as easy as just wrapping foil around it. One of the reasons we like toroidal transformers so much is they don't have near as much of a problem in this area as EI transformers do.

Lay your board out so all your sensitive, small signal components (and circuit traces) are as far from the transformer as possible. If you can, put the transformer in a back corner and put all of the power supply caps, etc, between the transformer and the amplifier board. Make sure you don't rount any of the signal cabling near the transformer either.

Be carefull to not make harnesses with just signal wires, always include return wires (grounds) in the same harness. If the return wire is run along side a signal carrying wire then the area of the loop created by the signal wire and it's return is very small and this cable (or trace) will not be as vunerable to emmisions. If you depend on another harness (or the chassis) for the return connection, then you will have created a much larger loop area which will act as an antenna and cause you grief. And twist (3 turns per inch) all of your harnesses inside the amp, both input and output wiring. This will help reduce the amount of noise the input wiring picks up and how much the output wiring radiates.

When you mount the transformer, leave the wiring long enough so you can rotate it 90 degrees if necessary. Many times you can fix an interference problem by just rotating the transformer.

Phil Ouellette
 

downhere

Member
2001-10-03 10:24 am
Just received my LM3876's today, need to complete my amp in a week. Need to populate my stripboard before I send the dimensions for the metal case. Decided to get a contractor to make it after all.

The EI transformer I got is huge, mounting it on a chopping board for now, with 20,000uF per channel. The power supply would be the easy bit, and I plan to make it as flawless as possible as I would like to test different amp circuits.

Exams are over, and i have free time to go speaker building. Any really cheap but good designs out there? I don't mind low efficiency, but I need good midrange/treble and decent bass response, -3 dB at 50Hz? My budget's $200 or so, including cabinets, but I'll get contractors to make the cabinets too, budgeting $100 on speaker drivers.

My question is, to avoid clipping, how can I connect more than one LM3876 to reduce the load on each device? I'm using ESP's design, with supply at +- 40V, and I'm considering including a switch to run the same signal through 2 seperate amplifier circuits. However, this would take up some space, and use up more heatsink than I would like to. Also, when the signal from the preamp is lowered in volume, and consequently voltage, could the sound be distorted?
 
Regarding Copperfoil:
The "transformer winding factory" got it. If you ask them nicely, they will give u some.
They use it for shielding, and they use the same width for the copperfoil as the height innside the EI. They wind it along (parallell) with the windings. But NOT inside the EI, must be winded outside of the iron. Otherwise this will cause a 1 turn that is shorted, and will get VERY hot!

Regards former trafowinder.
 

downhere

Member
2001-10-03 10:24 am
Hey thanks, will call them up and ask for some. Also, I've been reviewing the circuit over at ESP. Can someone explain the reason why volume pots should be placed at the input instead of controlling gain itself. Secondly, the National docs reccomend 100K on the signal input, in addition to a volume pot in serial, while Elliot went with a 1K resistor. If I were to add a volume pot on the input as suggested by National Semi, should the value still be 10K?

Controlling gain would be more beneficial to audio reproduction IMHO, as lower gain = lower distortion, which is what i've read. (or understood so far at least) lower gain also produces less heat at lower volumes than would a circuit set at high gain. Also, would reducing gain in all the circuits i come across improve the sound?

Trying to streamline Elliot's design as I'll be building 4 copies of the same circuit. Will post diagrams(hand drawn) when I'm finished with it
 
If you adjust the volume by changing the gain, you'll either end up:
1) Doing it by changing the negative feedback ratio, which will give you different characteristics at different volumes. You could get into stability issues if you're not careful.
2) Changing the operating point of the circuit. Again, you'll end up with different sonic characteristics at different volumes. In extreme cases, you could conceivably burn up a gain stage by running too much current through it.
It's best to diddle with the signal levels.
Yes, as a general rule, lower gain means lower distortion. But this is one of those things that you design in in the beginning. Tubes tend to have lower gain than solid state--something gleefully pointed out by solid state folks. They miss the point. With the lower gain comes lower distortion. It's perfectly reasonable to run a tube circuit with 10dB negative feedback, whereas a solid state circuit will require something on the order of 30-40dB feedback in order to perform as well. Since negative feedback has unfortunate sonic consequences, it's not surprising that many solid state circuits sound dry and lifeless. Yes, it's possible to design a low gain solid state circuit. Yesterday afternoon I was playing with a JFET (2N5457) circuit that only had 15 dB of gain. No, I don't know what the distortion was, I'm not equipped to measure such, but with no feedback at all it was -3dB at 150kHz--good enough for audio.

Grey
 
Downhere

Grey has covered most of the points I was going to make, and I'm not going to get into a discussion with him on the merits of tubes v BJT v MOSFET (wrong thread!). However, I would just like to add one point.

As Grey has said, if you make the feedback ratio variable you can run into stability problems. Also, if the amplifier is non-inverting, the gain cannot be reduced below unity so there will be a limit to how much the volume can be reduced. Gains below unity can be achieved with an inverting amplifier, but then you will need an additional inverting stage if you wish to retain absolute phase.

Geoff
 
downhere said:

My question is, to avoid clipping, how can I connect more than one LM3876 to reduce the load on each device? I'm using ESP's design, with supply at +- 40V, and I'm considering including a switch to run the same signal through 2 seperate amplifier circuits. However, this would take up some space, and use up more heatsink than I would like to. Also, when the signal from the preamp is lowered in volume, and consequently voltage, could the sound be distorted?
You can bridge the LM3876s - this effectively the same as two amplifiers, each driving a speaker of half the impedance. The 3876 is not recommended to drive a load of 4 ohms at +/- 40V supply voltage, so this won't work unless you use 16 ohm speakers.
You can also parallel two of them, so each contributes half the output current (effectively, each driving twice the impedance). This is difficult to do, because the different offset voltages will fight one another. Please read the app note referred to in this thread carefully before deciding:
http://www.diyaudio.com/forums/showthread.php?threadid=508
 
I thought of another approach: you could bi-amp, using a separate amp for each driver (woofer, tweeter). It requires an active crossover, but is probably the best approach for sharing the load among power amplifiers. Not an even split, but still a split.
See the article at the ESP site for more details.