Linkwitz’s chip amp – dead bug or other?

[rick57]

Has anyone built SL’s 3886 layout?

I’ve just started my first project, and chose the 3886 for it’s ability to drive a wider range of loads. Having heard the Orion and seeing the degree of thought in his work, I decided to use SL’s variation of the 3886 layout.
http://www.linkwitzlab.com/images/graphics/3886amp.gif

Having read the virtues of the gainclone, I decided to assemble his circuit the dead bug–minimum leads way.

When I try to fit the geeting up to the 10 components (twice as many as some of the 3875 implementations) around the 8 pins, while the first few are not too difficult for a newbie, fitting them all in, without risking a short on fire up progressively became a real PITA.

A friend dropped by, a former technician, and thought that some of the low value caps in SL’s layout, including a 100 p ‘loop’ cap off the feedback resistor, are probably for RF filtering, effectively (PS aside) that the SL design achieved the Gaincard advantages.

He felt that with > say 6 components, there would be more hassle than advantage in persisting with dead bug style. He suggested assembly on matrix board.

Opinions/ experience?

[GregGC]

http://www.diyaudio.com/forums/showt...184#post291184

http://www.diyaudio.com/forums/showt...939#post272939

[rick57]

Greg

Maybe I’ve missed something, but the first thread appears to be about getting a 3875 working correctly, rather than fitting 10 components in.

The 2nd thread show very neat wiring - you’ve spent a lot of time thinking of the best layout. I agree this is critical. I spent quite awhile too.
Though how many components are you using - five?

A friend is doing a 3875 with five components, and it seems a lot easier then 10 components. Just getting the components into the congested space is a major challenge!

FYI: to avoid one of my problems - previously soldered components dropping off while soldering the next one on:
someone offline suggested using a ‘heatsink’ between the old solder joint and the new, eg self closing tweezers.

Another option at the moment is to change to the simpler 3875.


Has anyone done a 10 component dead bug?

How effective might a 220pf cap on the rca plugs be to silence RF hash?

[Nuuk]

Has anyone done a 10 component dead bug?

There are eleven components in the design below (two feedback resistors). I would not use all those solder tags again either!

And moving the sockets further away will provide even more room if space allows.



When you work with P2P, you should bend connections so that they stay in place without soldering. That way, when you have to solder something 'on top' nothing moves. It helps top stick components down with a small blob of hot melt glue as well.

If you rely on the solder to hold things in place, there is always mechanical stress on the solder joint.

[GregGC]

Originally posted by rick57
Greg

Maybe I’ve missed something, but the first thread appears to be about getting a 3875 working correctly, rather than fitting 10 components in.

The 2nd thread show very neat wiring - you’ve spent a lot of time thinking of the best layout. I agree this is critical. I spent quite awhile too.
Though how many components are you using - five?

A friend is doing a 3875 with five components, and it seems a lot easier then 10 components. Just getting the components into the congested space is a major challenge!

FYI: to avoid one of my problems - previously soldered components dropping off while soldering the next one on:
someone offline suggested using a ‘heatsink’ between the old solder joint and the new, eg self closing tweezers.

Another option at the moment is to change to the simpler 3875.


Has anyone done a 10 component dead bug?

How effective might a 220pf cap on the rca plugs be to silence RF hash?

Rick,

I's more than 10 components in total, but the connected directly to the IC are 5. I used SMD resistors and utilized som of the NC pins of the IC (I connected them to the GND star) to provide solder points to GND. The components associated with the input are connected directly on the input RCAs, the speaker connectors and accross the attenuator/pot. Use the same aproach. If your layout is compact (not lage distances between the connectors and pots...). The components that need to be connected directly to the IC are the FB resistors and the caps (1000.0). The other ones can be soldered to the connectors and the pot and not directly to the IC.

Greg

[GregGC]

Here is the back view of the chip.

[Morse]

Hi GregGC;

Terrific work as always!

One question - what's the rating/type/brand of SMD resistor you used? I'd like to order some up from Digikey on my next order, but there are so many different types of SMD resistor that it's a little confusing - I want to be able to span the leads of a chip the way you're doing. I'm using 1/8 w axials and they work okay, but those SMD's look terrific there!

Thanks in advance for any advice or info.
All the best!
Morse

[GregGC]

Originally posted by Morse
Hi GregGC;

Terrific work as always!

One question - what's the rating/type/brand of SMD resistor you used? I'd like to order some up from Digikey on my next order, but there are so many different types of SMD resistor that it's a little confusing - I want to be able to span the leads of a chip the way you're doing. I'm using 1/8 w axials and they work okay, but those SMD's look terrific there!

Thanks in advance for any advice or info.
All the best!
Morse

Don't bend the pins that carry high current (they may crack). I just straitened the ones that will be connected to signal GND and then reinforced all the pins with some thick layer of solder to prevent stress and breakage. Don't overheat the pins too much.

The res I used are 5% but I have a lot on hand so I measured and matched them to 0.1%. Try the Panasonic ERJ 0805 type 1%.
0805 fits well between the pins of the IC.
http://ca.digikey.com/scripts/DkSear...313888&Site=CA

I have to admit I'm not into the use of exotic components.

/Greg

[Morse]

Thanks Greg;

>>>...Don't bend the pins that carry high current (they may crack). I just straitened the ones that will be connected to signal GND and then reinforced all the pins with some thick layer of solder to prevent stress and breakage. Don't overheat the pins too much...<<<

That's some food for thought - I was a bit cavalier bending the pins on the TDA2030A's for my (I guess it qualifies) NIGC. One signal lead did indeed shear off after an accidental slip, necessitating a bit of careful soldering to the remaining half of the pin. The V+ and V- leads had to be opened up a little to allow for the blocking diodes (hopefully the diodes' bodies will provide some reinforcement of the pins, preventing flexing and failure). Still, I'm looking forward to building with the LM series chips, since they don't require blocking diodes - which will let me follow your advice and NOT bend those pins.

Thanks again - Digikey's going to be getting more of my business...

All the best,
Morse

[rick57]

Nuuk:
“P2P, you should bend connections so that they stay in place without soldering”
Ah, my first P2P – I simply tried to solder straight pieces directly.

“It helps top stick components down with a small blob of hot melt glue”
(Another simple question) What are ‘top stick components’ or ‘hot melt glue’?


Greg:
“components that need to be connected directly to the IC are the FB resistors and the caps (1000). The other ones can be soldered to the connectors and the pot and not directly to the IC”

To minimize lengths, I had been trying to connect all components to the IC, and then connectors to those components. Your approach is a lot easier.

Greg, your labelled photos are excellent.
I gather that (the case of) the resistor from pins 4-9 appears to be contacting the pins in between doesn’t’ matter?


I’m also still curious if anyone has built Linkwitz’s 3886 layout, and eg knows the purpose of the components in his design, that are not in the National datasheet – if any of the parts are not essential


Thanks again guys