Massive GC? Suggestions please

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
with the sanding of the heatsink, i heard
"that everything above 500 grit is just polishing it and in extreem cases causes a loss in thermal conduction becuase the thermal interface goop that you use will cuase it to 'float' as there are no small pits for it to go into"
does this have any basis in reality?
 
Just to clarify a bit about my transformers.
It is a 1000VA toroid with 2x24vac secondaries.

And i also have 4 EI transformers with 250VA each. 24VAC single secondaries. The EI transformers are going to be used as monoblocks with 2 in each block. Each EI weighs.... 6kg!!:cool:

Nicks
 
autoexec said:
with the sanding of the heatsink, i heard
"that everything above 500 grit is just polishing it and in extreem cases causes a loss in thermal conduction becuase the thermal interface goop that you use will cuase it to 'float' as there are no small pits for it to go into"
does this have any basis in reality?

From my experiments and other's reports when overclocking computer CPUs, I would say that they're wrong (or maybe using a thermal paste I haven't..). Hardcore overclockers would lap their heatsinks and the tops of the heatspreaders on their CPUs to a mirrored finish, then very carefully apply a very small amount of heatsink paste. This could give as much of a drop of 2-3 degrees.

The hardest part of doing this wass getting a perfectly flat surface to "lap" the heatsink/chip on. A lot of folks used a thick piece of glass, wrapped in sandpaper.
 
ok, my first attempt in Eagle....

Its just 2 LM3886 in parallell. I intend to parallell 2 of these board.

Comments appreciated.

Nicks
 

Attachments

  • bpa400.jpg
    bpa400.jpg
    61 KB · Views: 508
Thanks for your comments!
Comments are always welcome!

i was thinking of making a separate regulated PSU for the opa134.
is this wrong in your opinion?

As i wrote: first attempt ever! :)

This is how i learn.

i will modify and then get back with an update.

Best regards

Nicks
 
Tlmadsen, thank you for your support. :)

i was thinking of parallelling 4 chips on each side.
Then use the SSM2142 opamp to "split" the signal, giving 4 chips inverted signal and 4 chips noninverted.

Much like the BPA 200. but using 8 chips per channel.

so, i cant see the "need" for bigger ground tracks since the speaker never will be connected to it. I hope i am correct here.

which opamp do you recommend instead of the opa134?

Best regards

Nicks
 
Hi Nicks

You are right about the ground track if you connect your speaker (-) straigt to your PSU. However having low-resistance ground tracks/path is always good for your design.

I take it from your PCB that the opamp in question is the one used for the DC-servo ??

Using a OPA627 or AD8610 is totally wast of money and I will be VERY impressed if you can hear any difference.

Use something like LF412 or AD712 (dual, so you only need one with your two amps on each board))

Have a look at http://www.diyaudio.com/forums/showthread.php?s=&threadid=19157

2 x 4 LM3886 will give you Power, but make sure you get propper cooling, otherwise you can't take advantage of all your LM3886 (and running at +/- 42 V DC is OK).

Have fun :)

Thomas
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.