[smrtby123]

I finished my buffered Cmoy and it works great (other than the IC opamp adapter, which is an unstable pain in the ass), but to get it working I had to tear out the BUF634 I had in parallel with my TLE2426. The voltage was split across the TLE but the buffer was dumping it to one side. The only thing I can think of is that I overheated it, but I didn't think I abused it that bad while putting it in. Has anyone else had problems with finicky BUF634s or ones that have died on them?

 

[error401]

How can you parallel a BUF634 with a TLE2426? They have different functions and that notion doesn't really make sense.

 

[rds]

Probably meant 'series'. The BUF634 would be buffering the ground. If it really was parallel then the dead chip seems reasonable.

 

[smrtby123]

I'm not sure exactly parallel or series when there is a virtual ground center, I just followed the power section of Ben Feist's A47 schematic, and it was hooked up correctly.

 

[ph33lix]

I have the exact same problem.
My buffered opa ground driver shifted the TLE's split by an average of -0.5v which seems to be quite a lot.

 

[amb]

When the BUF634 is used in an open loop manner (without an opamp and feedback loop), it will have significant DC offset.

 

[FallenAngel]

Quote:

Originally Posted by amb /img/forum/go_quote.gif When the BUF634 is used in an open loop manner (without an opamp and feedback loop), it will have significant DC offset.

What about a design like the A47?

 

[amb]

Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif What about a design like the A47?

Yes, that buffered ground output will have a DC offset from the TLE2426's virtual ground, because the BUF634 is in open loop. Btw, the same is also true of many other open loop buffers (e.g., sijosae) unless extra circuitry is added to trim/adjust or servo the DC offset.

 

[smrtby123]

So is there any way to make the BUF634 work in a Cmoy or such without turning it into something way more complicated? How did Ben Feist get it to work in his A47 amp, I just copied and pasted his power section into a Cmoy.

Quote:

Originally Posted by ph33lix /img/forum/go_quote.gif I have the exact same problem. My buffered opa ground driver shifted the TLE's split by an average of -0.5v which seems to be quite a lot.

It was more like a split of +6 and -1.2 on mine. so I think it was beyond just DC offset.

 

[ph33lix]

Quote:

Originally Posted by smrtby123 /img/forum/go_quote.gif So is there any way to make the BUF634 work in a Cmoy or such without turning it into something way more complicated? How did Ben Feist get it to work in his A47 amp, I just copied and pasted his power section into a Cmoy. It was more like a split of +6 and -1.2 on mine. so I think it was beyond just DC offset.

No idea yet on how Ben Feist got it to work. In any case, it looks like you've got quite a problem on your hands. If you're using the 8 pin package, you might wanna triple check your wiring and the datasheet.

The only time I saw such a huge offset was when I was using a resistor rail splitter on a pair of single OPA627s with low impedance cans.

If the connections you made are all checked, you might want to make a discrete version of your buffer so that you can swap your BUF634 out for testing.
http://www.head-fi.org/forums/f6/sij...buffer-212625/
Currently, I'm using this diamond buffer which is pretty simple to make on a perfboard and is meant to be pin-compatible with the BUF634.

http://www.head-fi.org/forums/f6/mon...er-pcb-345035/
This is the BUF634 replacement and should be pin compatible too should you leave the leads at the 4 yellow points uncut.

 

[smrtby123]

I got the amp to work but the solution was to tear out the buffer and rebuild the damned thing. At a certain point I might as well just go for a mini^3 or another amp.

I really wanted to use the buffer simply so I could try and use an AD8397 chip for its higher current output since I have a set of grados, but the chip is a bitch to work with. I have to do some mods to it, like decoupling caps and the other suggestions that amb made in other threads.