Gain on AD623 not per datasheet - why?

[GreenLeader]

Hi guys
I am using an AD623 instrumentation amplifier to boost the output of a pressure transducer fro 15mV/psi to about 260mV/psi for input to the Picaxe ADC.

I computed the resistor needed to set the gain using the data sheet formula:

Rg = 100/(G-1)

So for a gain of G=17.39 I need Rg=6.1kOhm

However, when I check the actual gain on my breadboard circuit, I get a gain of about 34.2, pretty much double what I thought I should get.

I increased Rg to 13.2kOhm, which should give G=8.6, but I actually get 19.1, also double what I expect.

What's going wrong?:
(a) am I nuts?
(b) is the datasheet wrong?
(c) is there something I have missed?

 

[MFB]

Maybe the error being X2 is significant. Are you getting single-ended and differential gain confused?

 

[GreenLeader]

Maybe the error being X2 is significant. Are you getting single-ended and differential gain confused?

Could well be - what's the difference?

I assume my signal would be single ended - it has a common ground. But how does that affect the gain?

I'm using a 1.25V reference for the AD623 to get an offset so that so I can let my signal go from a real -1.25V to 3.75V, shifted to 0V to 5V for the picaxe ADC.

 

[inglewoodpete]

When I get strange results, I go back to small steps. Try the 1.25v offset with a gain of 1. Then a gain of 2 etc. The problem may become obvious.

Otherwise option (a) may be correct

 

[GreenLeader]

Actually MFB is right, and the answers is (c)
Re-reading the data sheet, I see this wording :

"Also for bipolar inputs (that is, input range = 2 VDIFF), the maximum gain is half the value yielded by the previous equations because the REF pin must be at midsupply"..

While I haven't figured out exactly what that means, or if it applies to only the "maximum" gain as opposed to the "gain", it does seem to confirm MFB's comment. I'll have to read the data sheet a few more times before it sinks in properly.

I think what I wrote about my signal being single ended is rubbish. It is indeed a differential signal, coming from a bridge type device...

Thanks !

 

[MFB]

Its one of those things that only tends to fool you once. After that you are on your guard against ‘errors’ that are significant, like twice or half what you effected.

 

[GreenLeader]

Would you believe that I finally got to the bottom of this?

It turned out that the gain was in fact as I originally expected it to be from the data sheet equation.

The reason I thought it was coming out wrong was that my calibration procedure was flawed! As luck would have it the error was about a factor of 2. Had it been 1.5 or something I might have searched a bit harder!

I had hooked up a manometer to the pressure transducer that had the AD623 to boost the signal level. So I was indirectly measuring the gain by comparing pressure to expected volts. Had I measured the gain directly by measuring the input and output volts to the AD623, I'd have spotted it straight away. (I only have one multimeter so that option was not convenient - next time I'll use WOMAI's oscilloscope).

Instead, it was only when I was calibrating the full prototype over a bigger pressure range that I realised my procedure mistake! (The mistake itself is another embarrassing story altogether!!)

 

[BeanieBots]

Glad you got it working. Just goes to prove that going back to basics and checking the fundamentals always pays off in the long term even when it appears to be "inconvenient" at the time.