paralleling chips

[odysseybmx414]

ok maybe this is a stupid question or maybe not

ive always wondered why each chip has to have its own parts and why you cant just put one chip on top of the other(connect the inverting input to the other chips inverting input etc instead of having to use 2X as many resistors etc. if anybody has an answer to this please let me know as im very curious. i have never tried this before as in i have no speakers i am willing to blow to test this. but if anybody tried it let me know the results.

thanks.

 

[tiltedhalo]

paralleling chips??

Let me get this straight, you mean paralleling the chips and useing a global feed back loop for all parallel chips, same with input net works and output also???

 

[tiltedhalo]

parralleing chips

Sorry didnt read the post close enough! these chip are not unity gain stable, boostrapping wont work

 

[odysseybmx414]

opa549 is unity gain stable so would it work

and jus to make sure what i mean i mean just connecting all the pins from one chip directly to the other chip and just using one fb, input resistor, etc.

 

[tool49]

Due to difference in chips, doing so would cause imbalance. You would to the very least need to have a current sharing method to insure that they're all working to the same degree.

Hope this helps!
Sébastien

 

[odysseybmx414]

ive decided im going to try it with 2 opa549's i will repost later

 

[tiltedhalo]

Holy chip, it works! I tried it with two LM3875,S I had laying around. Split the input with 2 1K resistors and 1 22K to ground used a common feed back loop with a 22.1K .1%, 47uf tantalum cap, and a 1k .1% to ground fed the outputs to two .15 ohm 1% soldered directly two the pins and took the feedback of the combination of outputs. Might have to take a closer look at this! I got the least amount of DC offset of any chip amp Ive made!!

 

[troystg]

tiltedhalo-

Picture please.....

 

[neutron7]

i know you can do it with BUF634 for headphone amps etc.


tiltedhalo can you try it without the tantalum cap? lots of GCs dont use a cap in the feedback.

 

[tiltedhalo]

cap removal

The tantalum should not be removed for DC offset, you also need current resistors, but you can use a common feed back network. I would post pics but my wife is out of town and has the digital camera.
I had this strange idea of potting the hole thing in epoxy and soldering the chip to a piece of copper and basically building a parallel module

 

[Nordic]

Posted this exact question before... got no answer.... good work

 

[odysseybmx414]

so let me get this straight you can do it, as long as you have a 47uF in the fb? would that be parallel to the fb resistor or series with it? im a bit confused.

the output .1 ohm resistors i kinda figured they would be used in the first place for current sharing but i do not understand the 47uf in the feedback loop.

 

[neutron7]

well could you at least use a electrolytic instead of the tantalum one? and put a nice non polarized cap in parralel with it.

 

[jaudio]

Holy chip, it works! I tried it with two LM3875,S I had laying around. Split the input with 2 1K resistors and 1 22K to ground used a common feed back loop with a 22.1K .1%, 47uf tantalum cap, and a 1k .1% to ground fed the outputs to two .15 ohm 1% soldered directly two the pins and took the feedback of the combination of outputs. Might have to take a closer look at this! I got the least amount of DC offset of any chip amp Ive made!!

How hot does the chips get? I have a hard time believing it is that simple . I mean why would manufacturer who uses chips in parallel use so many extra parts? I hope you are right and your amp can be duplicated by other.

 

[tiltedhalo]

parallelling chips

In all honesty your only saveing 2 resistors, everything else remains the same. You still need current shareing resistors ect., the only differance is your useing a global instead of local feedback network, this has been used for years with other op amps.

 

[jaudio]

Ok the tantalum cap can correct the internal dc offset of each chip? Have you measured each chip's offset separatly? It seem to me that if you dont get each chip's offset down before paralleling them,you will still get quite a bit of curent share even if you have resistors at the output.