The precision of INA105 for voltage addition and voltage subtraction

I'm doing a simple project which will do voltage calculations(A+D)-(B+C)/(A+B+C+D) which some of you have known through my previous post.

I'm using INA105 for the addition part. With Vcc +5 (+10) and Vdd -5 (-10), i obtained very poor results. For example, A=-5, B=-5, A+B=-10 which is very inaccurate. Based on your experience, are there any problems with the precisions of the INA105 itself which is not 100% unity gain or some noise may be incurred?

I'll add a VR in the feedback loop to fine tune. I may be wrong. Do you have any suggestions/opinions which may help to quicken the design process?

Fyi, the link to technical info of INA105 is as below:

Regards, Albert

I don't understand your terminology. Is Vcc +5 or +10? What is Vdd? And is it -5 or -10?

From the datasheet for the part (which is a Texas Instruments part now that TI bought Burr-Brown) V+ (Vcc) should be +15 Volts if you want to input and output signals up to 10V. And V- (Vee) should be -15. Also, the part doesn't perform addition. It performs subtraction.

From your post I get the impression that you set A=-5, B=-5, and measured the result of A+B at -10. Why do you call that inaccurate? That sounds correct to me.

What do you mean by VR?

--Mac

through

-5

A+B=-10

I tested with 1) Vcc =+5, Vdd=-5. 2) Vcc=+10V, Vdd=-10V and

now

to

the

The IC is designed for subtraction but it can be modified to do addition also. You can refer to the datasheet. There's one diagram about addition and etc.

Sorry. The A+B i got is actually -9V.

you

Variable resistor.

I think you mean "Vcc" and "Vee." I suggest you stick with +/-15 Volts.

"Vdd" is usually used in place of "Vcc" in MOS parts. For bipolar process parts, you would use "Vcc" for the positive power supply, and "Vee" for the negative supply if there is one. In MOS parts, the negative supply would be Vss.

I guess you are referring to figure 13?

Oh, OK. The chip itself should be capable of much better results than that. It might be a good idea to start a new thread, and post a full schematic showing your whole circuit (including resistor tolerances) and detailing the results you got. I don't think anyone is watching this thread anymore. If you can put the schematic on a web page you control somewhere, that would work. If not, use savefile.com

Either way, post a link to the schematic.

You should be able to get very close using 1% (or better) tolerance resistors without resorting to a potentiometer (variable resistor).

--Mac

Be sure to keep all intermediate and final voltages well inside the output range.

Bye

Dear Uwe Bonnes,

thanks. Quoting your post, i assume that you meant the intermediate and final voltages must be within the Vcc &Vee. Am i right? Correct me if i'm wrong.

Dear all,

Well... After testing the circuit and connecting the circuit to the correct ground, i was able to obtain the more desirable results., which is +/-3%.

Now, i'm facing a problem regarding Vref and GND. My circuit is basically designed to obtain 4 signals (namely A, B, C and D) from 4 testing points of a CD drive and do calculation of output ((A+D)-(B+C))/(A+B+C+D). All the 4 signals and output must be connected to Vref as a reference point instead of GND. Apart from that, the Vcc and Vee of the op-amps used are also connected to the Vref.

Vref and GND mentioned above are both points on the CD drive also . When we take GND as the reference point, everything works perfectly. However, if Vref is taken as the reference point, some component(s) which are in between Vref and the GND is/are burnt.

My question is that how can we take Vref as the reference point without causing damage to the CD drive while getting a desirable results? Hope to get your response soon. The Vref for different drives are different.

Fyi, the requirement is to take Vref as the reference point. I'm not sure my explanation here is clear. If not, please feel free to highlight to me. I'll try my best to make the picture clearer.

Sorry that my circuit diagram is not yet ready in soft copy. Thanks.

Regards, Albert