INA163 instr amp from single rail?

Hi,

attempting to get one of those TI's INA163 dual rail inst amps to work
from a single rail, but it doesn't quite work, and I'm a bit puzzled why
it doesn't...

Right now I'm using +10V as the rail (=+V) and -V = GND. The reference
pin is connected to a small +5V regulator (instead of gnd as would be in
the dual rail setup). The differential inputs are both biased to +1.5
Vdc, onto which the signal is then coupled. When I test with e.g. a
50mVpp 1kHz sine signal, a slightly attenuated version appears on the
external gain set resistor, none on the intermediate Vout1 and Vout2
pins, and the actual output pin just stays quite solid at +5.0Vdc,
though with a 1kHz sine with a fraction of a microvolt.

Snip from the schematic:
http://www.hut.fi/~jwagner/ele-tmp/INA163-circ.gif (+1.5Vdc bias not
shown) Any thoughts what might be wrong with that setup?

In a crude simulation http://www.hut.fi/~jwagner/ele-tmp/INA163-sch.gif
it works fine, none of the internal signals ever falls below GND
(http://www.hut.fi/~jwagner/ele-tmp/INA163-sim.gif) except for the
Vout1-Vout2 signal to which however the +5V reference is added, so even
the output opamp stage should not have any problem with clipping (?) Any
ideas? Or does anyone happen to have a schematic with the INA163 working
from a single rail? (google didn't turn up anything...)

- Jan

 

Subject: INA163 instr amp from single rail?

Hi, Jan. Your ersatz GND (Vcc/2) can probably only source current. How about
using an op amp with a voltage divider and cap to get Vcc/2. That's the first
choice for a quick and dirty Vcc/2 "GND". The op amp can source or sink
current.

Ersatz GND
VCC
+ .-----------.
| | |
.-. | VCC |
R| | | + |
| | | |\| |
'-' '--|-\ |
| | >-----o------o Vcc/2
o----o-----|+/
| |/|
.-. | |
R| | C--- ===
| | --- GND
'-' |
| |
=== ===
GND GND

created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

Good luck
Chris

 

Good and reasonable suggestion, thanks! I tried it and tinkered around a
bit, but, too bad it did not help... So, the INA163's still refuse to
amplify. Any further ideas?

There shouldn't be anything obviously flawed with my single rail +
mid-rail reference approach, right?

Instrumentation amps should work even with a "negative" common mode bias
on the inputs?

Right now the +5V reference and +10V rail are decoupled with 100nF chip
capacitors right next to the INA163, both look quite noisefree, and
there's plenty of ground plane around, on both top and bottom layers.
Loading or not loading the output pin didn't change anything either.
Even all the internal resistors measure out ok. Of course it could be
that (unlikely, but...) the two sample parts were dead...

thanks,
- Jan

 

As long as all the inputs are within the specified common mode voltage
range, with respect to the supply rails, there is no good reason it
should not work.

 

Subject: Re: INA163 instr amp from single rail?

Hi, Jan. It sounds like you're doing everything right now, but you're getting
the wrong result. I'll bet there's a wiring problem in your prototype.

Sketch the circuit you want on a sheet of paper, if you haven't done that
already. Make a photocopy, and then go over your prototype node by node,
checking off connections on the copy with a colored pencil. Don't spare
yourself here - do the whole job. Also, it might be possible that you plugged
both samples in to a circuit, and then discovered a wiring problem and fixed
it. Unfortunately, you can wire this IC to let out the smoke (you didn't
reverse power connections at any time, did you?). By the way, don't measure
the internal resistors.

Well. You've picked a rather unusual part that isn't stocked at this moment by
the TI e-store, Digi-Key, or most of the distributors. However, Avnet has
them, and the TI e-store/samples department might be able to talk Avnet into
breaking a tube for you.

Good luck
Chris

 

Thanks for your help, and thanks John for confirming there's nothing
fundamentally wrong in the single rail + reference approach.

Well, wouldn't know about pen and paper, I'm mainly using ERC+DRC,
copper track highlighting and extensive datasheet specified pin-out vs
library&PCB pin checks, and scoping the actual circuit - helps against
drawing imaginary things in sketches, IMHO. Anyway, I'm still a bit
sceptical but it looks like both chips _do_ have dead output stages, as
a x-acto mod into dual supply with gnd reference plus all checks didn't
bring these to live either, while internal input amps appear live and
kicking. The sample chips were fresh, good ESD handling + no reversed
rails or such things, so, go figure...

Luckily this isn't for a production line - well it wouldn't even have
been suggested for the approved parts list in the first place... Now,
ADI's SSM2019 apparently comes close in BP/UGBW and noise specs, and
surprisingly it's even stocked, e.g. should the INA163 sample order not
work out...

cheers,
- Jan