advertisement


Discrete DIY chip amp on copper heat sink

Yup, fairly standard gainclone. Chipamp.com sell the boards/kits.. Simple build and sounds very good :)

Sam
 
a few years ago I built the point to point chipamp from the link, its still working perfectly well today driving the speakers on my parents TV. I never really subjected it to any really critical listening but from what I heard it sounds fine...

The only point I would make is that using the component values in the link I found that the gain (set at 33) was way too high. If you download the LM3886 spec sheet you will find the calculation to work out the resistor values for lower gain. In the end I think that I found a gain of about 15 to 18 ish worked for my application.
 
Added directly to that - this linked schematic has no coupling cap and gain is maintained down to DC. It will probably exhibit a bit more DC offset than you might expect, which will vary with the volume pot setting (since the input bias current x vol pot setting is multiplied-up by the gain figure; though the low 10K pot value will greatly help minimise this effect). Should be OK, but do check your source *does* have a coupling cap or you could end up with a bit too much DC on the output easily - say over 100mV.

ETA: checked the LM3886 datasheet; input bias current is given as 1uA max. So, for example, 1uA flowing through a 10K pot at the mid-way mark (5kohms assuming an AC-coupled source) is 5mV; multiplied by the gain shown of (22k/680 +1) = about 165mV. That's probably an unlikely -in-practice upper bound - and still doesn't look too bad. Normal input bias is given as 0.2uA average, or 30mV offset under the same condition = just fine.
 
I've built three of those, just following the instructions like "painting by numbers", plus some very useful advice from various PFM members, in particular Martin Clark.
They sound very, very good, to my ears at least.
 
Thanks Paul and Martin. Is the copper and direct link wiring of benefit?

How would they compare to a NAP250 or more specifically a pair of them active in SBL's.
 
Hi Jonnoshore,
Martin is the expert in this field, and he knows how an extra component added to the circuit here and there can affect things like roll-off at the top and bottom ends of the frequency range, and the DC offset at the speaker terminals. The nice thing about these LM3886 amps is that you can tweak them very simply.
I don't know about a 250, but I compared one of mine to a 140; same room, same speakers, etc. Without a pre-amp, with just a volume pot, the LM3886 was a bit different from the 140, but not necessarily better or worse. Then I connected it downstream from the NAC72 pre-amp and it sounded very similar to the 140, only very slight differences that I cannot really describe. Maybe a bit "sharper" and "colder" than the 140, but not that much in it at all.
 
Ah, the copper. I don't know, but I would not think it makes any difference at all compared to an aluminum heat-sink. Just more expensive and harder to work with.
Some people advocate very short, point-to-point connections, others use printed circuits. Probably not that much difference. I've always used PTP on a perforated board. But you have to be very careful no get everything connected right without making extra contacts, and in the end it can look a bit messy and cobbled-together. If you have a PCB like that supplied by Chipamp you can't go wrong, easier and safer.
 
Thermal conductivity of copper per unit volume is marginally better than aluminium, but is much denser and more expensive. Aluminium is actually better per unit weight or cost.
 
There's no special magic in direct wiring. Short signal paths are a good thing in general, as the longer the path the more chance there is to pick up up noise, plus added inductance in the feedback loop can cause instability. But as long as the layout is reasonably compact, it's not going to make much difference in a normal speed audio frequency amp. Very fast amps can be more critical.
 
You must log in or register to reply here.


advertisement


Back
Top