Switchable inverting/non-inverting opamp circuit?

That would admittedly be much easier, but one of the prerequisites in this case is that the lowest and highest output values can be set (semi-) independently and accurately. I found no way of doing this using adjustable gain and offset.

Richard Rasker

How about having your switchover points fixed as you want. Then just derive the integrated current values from the switch points deltaV.

Trim to get 1Hz | G=1 inst amp | || | .----||-. HC4053 V | || | |\ +--------o ___ | |\ | Vlow >-------|-\ | .----. \o-|___|-+--|-\ | | >-+-| -1 |-o | >-+ Vhigh >-------|+/ '----' ^ GND-|+/ | |/ | |/ | | | | /| | | .---. /+|------< Vlow | | S|--< | | | | | \-|--+ '--o| R|-. \| | '---' | | | /| | | /+|--' '< | \-|------< Vhigh \|

Fred, Not to put too fine a point on it, but this is pretty much what I came up with in the first place (refer to

- U3 p8/9/10 provides the difference between Vlow and Vhigh

- U3 p1/2/3 is the switchable inverter/non-inverter, with a MOSFET instead of a 4053

- U3 p12/13/14 is the integrator

- U2 p1/2/3 and p12/13/14 are the comparators

- U2 p5/6/7 is configured as flip-flop

But thanks for your effort anyway -- you've come up with some real gems in the past, among which a very elegant log converter circuit for a mass air flow sensor, which is still in use :-)

Richard Rasker

Ahem, OK, I missed it. Pretty much the same indeed.

Glad it helped :-)

Sorry for resurrecting an old (and entertaining!) thread here, but I have a very similar design problem. I have a 0???4 V analog control signal ("A_abs") that I need to selectively invert with a 5V output line from a µP. A_abs is updated at 40 Hz so no fancy high speed stuff should be needed (but phase/time delays from filters are not ideal also).

It seems like the circuit on page 2 of this PDF would do the trick (basically the same as the one shown in the sweep-page1.ps document), but A_abs is usually less than 100 mV and the error introduced by the V_CE?0 adds an offset onto the output. Any ideas that I can breadboard before I send my PCB out for fab? I may just go with the MUX idea mentioned up front here.

--Sean Clarke

--------------------------------------- Posted through

Hi sclarke. Forget the bipolar transistor. Although some have a very low VCE(sat), especially when the collector current is small, it could still be significant w.r.t. 0.1V.

Consider a MOSFET, or an IC providing analog switches.

With some analog switching/multiplexing you could also implement a solution whereby you split the signal between a buffer and inverter, and then multiplex on the output side.

Thanks for the response, Kaz. The MUX on the output is probably the way to go for me. I haven't done electronics design in a long time and I haven't had much luck getting the SPICE model to work correctly when I use a MOSFET.

Does anyone have a suggestion for a MUX? Are there "Make before Break" models? My 'A_abs' signal is actually an input to a flight control system on an aircraft, and after this follower/inverter stage it goes through an instrument amp and another buffer stage so I don't want there to be a risk of the MUX terminal developing a stray voltage for the brief moment that it's floating during switching as this could be amplified. As a fall back I may just add a large pull-down resistor.