The BUF634 amp.

[Avro_Arrow]

This is an offshoot of the thread "Some LME49600 implementations".
 
This little board is meant for the BUF634 in TO-220 package and
a dual channel op amp such as LME49720 in DIP. In fact, the schematic
is pretty much the one found in the LME49600 data sheet.
It is done with almost all through hole components with just
a few smd caps.
 
The schematic:
 
 
 

 
 
 
I kept the resulting layout to one layer to make it easy to etch at home:
 
 

 
 
Any thoughts, mistakes or interest?
 
Edit: changed servo values

 

[G.Trenchev]

You don't need that servo.

 

[Avro_Arrow]

I know.
 
If you want to have a servo, build this.
If you don't want a servo, build something else.
 
Quote:

You don't need that servo.

 

[G.Trenchev]

Sorry,I didn't want to sound offensive.It's just something already in the opamp itself.You'll never see anything over 1mV with or without it.
Oh,well,it a fine design anyway.Good luck.

 

[Avro_Arrow]

The servo is not meant to null out on board offset.
The servo nulls offset coming from the device connected
to the amp.
 
For instance, my Chinese DAC has almost 20mV offset.
The servo would enable the amp to be DC connected  with
no offset being passed through from the DAC.
 
 
Quote:

Sorry,I didn't want to sound offensive.It's just something already in the opamp itself.You'll never see anything over 1mV with or without it.
Oh,well,it a fine design anyway.Good luck.

 

[wakibaki]

It's just something already in the opamp itself. You'll never see anything over 1mV with or without it.

 
The LME49720 datasheet states 0.7mV input offset voltage max. With a gain of 2 that would result in an uncorrected output offset of 1.4mV. The gain of the circuit as drawn is 2.1.
 
More important is the input offset current of 65nA max., which, with the impedance looking out of the servo inputs being 1M, will result in a worst-case offset of 65mV at the servo output. Even taking into account the 1:10 servo authority, this is why the servo should use a different opamp and the amplifier really requires the use of a 2-to-1 adapter instead of a dual opamp, as discussed in the 'Some LME49600 implementations' thread.
 
With or without, you're wrong on both counts.
 
w

 

[Avro_Arrow]

Thanks for reminding me...I should change the servo values to 100k and 10u.
But yes, as Wakibaki mentions, I had originally intended this to have LME49990
in the voltage gain spot and OPA211 in the servo spot.
 
I used that combination in another amp and the result was less than 1mV
using 100k and 10u in the servo.

 

[qusp]

for servoing an input? give me a nice cap anyday. i'm all for DC coupled outputs and signal path where possible as i'm not too keen on electros in the signal path if the output is driving headphones, but for a dac output feeding a very high impedance opamp input, i'll use a decent quality film before a servo, or just build a dac with no DC.personal taste there is evil all around us

 

[Avro_Arrow]

I agree
I intend to do a version with input caps and no servo and a version with no cap or servo...DC all the way through.
I have to be careful though, eh....don't want to get too close to reproducing "The Wire" now do I...
 
Quote:

for servoing an input? give me a nice cap anyday. i'm all for DC coupled outputs and signal path where possible as i'm not too keen on electros in the signal path if the output is driving headphones, but for a dac output feeding a very high impedance opamp input, i'll use a decent quality film before a servo, or just build a dac with no DC.personal taste there is evil all around us

 

[G.Trenchev]

OK,I've tried with LME49990,LM4562,OPA1612 and OPA2132.On two different boards.
Less than 0.1 mV every time at gain of 3.
 
Quote:

 
The LME49720 datasheet states 0.7mV input offset voltage max. With a gain of 2 that would result in an uncorrected output offset of 1.4mV. The gain of the circuit as drawn is 2.1.
 
More important is the input offset current of 65nA max., which, with the impedance looking out of the servo inputs being 1M, will result in a worst-case offset of 65mV at the servo output. Even taking into account the 1:10 servo authority, this is why the servo should use a different opamp and the amplifier really requires the use of a 2-to-1 adapter instead of a dual opamp, as discussed in the 'Some LME49600 implementations' thread.
 
With or without, you're wrong on both counts.
 
w

 

[wakibaki]

Quote:

OK,I've tried with LME49990,LM4562,OPA1612 and OPA2132.On two different boards.
Less than 0.1 mV every time at gain of 3.

 
That means nothing, other than that you've been lucky.
 
Do you imagine Nat Semi (TI now) would put 0.7mV in the datasheet if the device could be relied upon to achieve <0.1mV?
 
Data sheets are what professional designers rely upon to keep their jobs and what prudent amateurs use to ensure professional results. They detail the performance the manufacturer expects on the basis of both design experience and the test of thousands of devices. Learn to depend on them or expect some unpleasant surprises.
 
w