Current feedback Op-Amp

[krw]

I have an application for a fairly high power op-amp (50MA - 24Vp-p
into 300ohms) at audio frequencies. For several reasons, likely
some not very good, I selected the AD810 for a test. After doing
some more research I'm having second thoughts about a current
feedback amp in this application (Howland current source), though
the simulation looks good. Specifically I'm concerned about the
zero input impedance at the inverting input. I still have (a
little) time to change it to a voltage feedback amp, but would
rather not for this test. Not having been an analog type for
<mumble> years (and having never used a current feedback amp), I
thought I'd get the opinion here. Thanks.

 

[Martin Griffith]

I have an application for a fairly high power op-amp (50MA - 24Vp-p
into 300ohms) at audio frequencies. For several reasons, likely
some not very good, I selected the AD810 for a test. After doing
some more research I'm having second thoughts about a current
feedback amp in this application (Howland current source), though
the simulation looks good. Specifically I'm concerned about the
zero input impedance at the inverting input. I still have (a
little) time to change it to a voltage feedback amp, but would
rather not for this test. Not having been an analog type for
<mumble> years (and having never used a current feedback amp), I
thought I'd get the opinion here. Thanks.

I came across this, earlier today
http://www.national.com/pf/LM/LME49600.html
probably overkill though

martin

 

[krw]

Maybe post your schematic? Vagueness gets you garbage solutions ;-)

I don't have the exact schematic here, other than it's a classical
Howland current source driving two loads (a transmission line and an
impedance matching leg) connected as a bridged trans hybrid. My
concern is with the current feedback amplifier in the Howland
current source.

I haven't thought through the ramifications of the CF amp's input
characteristics WRT a VF amp. I'm having a bit of trouble with the
whole concept. I guess next weeks job is wrapping my arms around
CFAs.

 

[krw]

I came across this, earlier today
http://www.national.com/pf/LM/LME49600.html
probably overkill though

Kinda neat. I'll file it away, but I don't see how it's going to
help me here and it's kinda expensive (almost $5).

 

[krw]

I didn't realize it was April 1...

Didn't think it was that funny, though I'm out of my normal water
here.

Howja like that +10dB peaking ?

I ignored it because the intended use is almost a full page left of
the "peak". The application is audio, and not for phoolz.

 

[krw]

That peak portends instability in the feedback loop(s).

Understood. Five orders of magnitude is a lot though.

 

[krw]

Only if you configure the feedback loop to guarantee that the peaking
occurs _well_ after the closed-loop response crosses zero dB.

Sure, but five decades is a lot of space to roll the gain off to
below 1. A single pole squishes 100dB in that space. ;-) The
nominal gain of this thing is only 1 (though it is in the feedback
path).

 

[krw]

It's the overall loop gain that matters, so you simply need to add
this "unity gain" (in dB's) to the that of the LME49710 or whatever it
is you're using.

Sure, that's what I meant by it being in the feedback path. Does
anyone use output stages for a lot of voltage gain? Sounds
dangerous.

Loop phase is what will get you ;-)

Which, to me is why that hump has "Danger Will Robinson" written all
over it, but again, it's 5 decades away for where I care.