From: Winfield Hill
Subject: Re: Current Feedback Amp Questions
Date: 27 Sep 2002 07:12:39 -0700
Organization: Rowland Institute
References: <email@example.com> <3D92F99F.firstname.lastname@example.org> <3D944FC0.email@example.com>
X-Newsreader: Direct Read News 2.98
Fred Bloggs wrote...
> Winfield Hill wrote:
>> Fred wrote...
>>> Are you driving these indicators with current or voltage, or do you know
>>> the difference. Whatever, the Nationla Semiconductor LM675 would be the
>>> part to use, a single chip fully protected, solution at low cost, except
>>> from Digi-Key, but still 25% of the LT and BB candidates which will be
>>> layout sensitive among other things.
>> A nice part, but be careful with its minimum gain of 10 requirement.
>> Be sure to add extra parts to create a noise gain of at least 10x.
> The data sheet gives many examples of how to do this. The circuit below,
> which was declared a "trick question" by several distinguished members
> of the "S.E.D. staff," is an example of a practical application of this
> type of follower:-))
> R | \
> +---/\/\/\------|+ \
> | | \
> Vin----+ R | / -----+--Vout
> +---/\/\/\--+---|- / |
> | | / |
> | | / |
> | |
> | 10R |
> +----/\/\/\ ----+
Right. Very good. However a few improvements can be had. First, if
the circuit ia used as shown, 0.9R for the + input better balances the
bias current portion of the input-offset error. Better yet, add a cap
to eliminate the 10x DC gain, with its multiplication of the DC offset
error, up to 100mV max at the output. The change cuts it down to 10mV.
NSC shows this form, which is more conventional for unity-gain followers
using non-unity-gain opamps. The values suggested are fairly conservative,
but are well suited for our O.P.'s needs. R2 = Rs = 3.3k is about right.
. LM675 _
. Rs | \ f(-3dB) = 50kHz
. in o--/\/\--+----------|+ \
. | | \ +/- 20V
. R | >--+--+-------o out +/- 2A
. | C | / | |
. '--||--+---|- / | '--||--/\/\---,
. | |_/ | 0.22uF 1R0 |
. | R2 | gnd
. R = (Rs + R2)/10 '---/\/\----'
. C = 1/(R 2pi 500kHz)