I have a quick design that I've been tasked with. The design involves data logging a narrow band of temperature readings from a TC. I know the Analog AD595 chip has a 10 mV/degC out but sine I am working with a narrow range of temperatures, I'd like to increase the slope. My first thought was to use a TI INA128 instrumentation amplifier but so far that hasn't worked and I can't figure out why.

The quick summary of the circuit is: The AD595 output is connected to INA128 Vin+. Vin- is connected to ground. Rg=5.6K (G=10), Vref is grounded. Both chips are running from the same 0/+15 power source.

I can get movement of the AD595 output so I know that's working and thus limits my issue to the in-amp. Can anyone point me in the right direction of what I have done wrong with the IA setup/design?

There are two other designing issues: 1-I'm stuck with what we have around already (Lots of K-type TCs), 2-I don't have access to an oscilloscope.

Thank you for your time.

 

Tell us what the problem is might help us out. BUt to me it seems like the INA128 needs a bipolar supply. Since voltage is all relative, this is meaningless for two pins. BUt you do have a reference pin on the op-amp, however, that is normally connected to ground. THis would appear to be the third pin that makes the op-amp have a bipolar requirement rather than a unipolar requirement that is twice as big. So what happens if you connect REF to V/2 through a resistive divider (such that +V and GND are bipolar as far as V/2's perspective is concerned). THe minimum voltage requirement of +/-2.25V means that Ref must be at least that much away from both +V and -V. That is the first problem.

THe second problem I can see is that the INA128 is NOT rail-rail input with a minimum input voltage of -V+1.7V, and a AD595 sensitivity of 10mV/C you need to be measuring at least 170C for the AD595's output to reach the minimum input voltage for the 1NA128 which is 1.7V in this case since your -V =GND = 0V.

 

I knew I forgot something.

I can't get a response of the IA. It stays steady at 0.100V while Vin+ changes from .280 to .350. At a steady temperature (28degC), the AD595 output is 0.280 and the IA should be 2.80V but it's "stuck" at 0.100.

 

I've edited my post because I've spotted another problem in the circuit and have tried to improve my explanations.

(THe post should refer to the minimum input voltage of the op amp as a problem, as well as the requirements between +V, -V, and REF which are imposed by the op-amp's minimum voltage requirements.)

THe resistive divider will have to have a low enough source impedance (or be buffered by another op-amp) so that the internal 40K resistors inside the TI INA128 do not skew it's operation. Easiest thing to do right now would be to use bipolar supplies on both AD595 and INA128 and leave REF connected to ground. Because with a unipolar INA128, you cannot refer it's output to ground and it will be at V/2 or something like that and you've just lost half your scale. THis would also reduce your minimum input voltage to the 1.7V above the negative rail (and below the minimum output voltage of the AD595) and allow your AD595's output to work with the op amp at low temperatures.

Or you could just get another amp.

 

It's a little late for me right now, but if read you correctly switching to a +/-5V or +/-12V supply with Vref at ground that should solve my problems. Vref will be more than 2.25 from the rails if it remains at ground and the output of the AD595 that will be between 0-1.25V will be more than 1.7 above -V (at -5 or -12).

I will give that a shot tomorrow.

For amps on hand, we have the INA126, INA128, INA129, and a few useless surface mount ones that I can't prototype with.

 

Hehe, I think you responded after reading my old second post but before I updated my second post again (I update my posts a lot so long posts aren't valid until about 5 minutes after they are first posted). But yeah, the easiest thing to do is to use bipolar supplies. Bipolar 5V would work too. since -5V+1.7V is far below the minimum output voltage of an AD595.

 

Thanks for the thoughts and help. I'll let you know how things turn out tomorrow.

 

I used a -5/+15V supplies and it works great. Another question, if I knew the only input voltages that I would see were between 0.6 and 1.0 and I set Vin- to 0.5 then at G=10, my output would swing from 1 to 5 volts correct? How high will the output of the INA128 go with a +15V supply? 13V?

 

See attachment for the area of the datasheet that tells you the limits of the output voltage. WHen selecting amps or op-amps, some of the first things you should be looking at on the datasheet are the supply voltages (as well as unipolar or bipolar supply), the input voltage limits, and the output voltage limits, all of which you seem to have missed when initially designing your circuit.

Yes, it will amplify the difference between the inputs since it is a differential amplifier (and is also true for an amplifier with a changeable bias/reference, but there is a subtle but important difference with how common-mode signals are treated and I won't get into it).

Attached Images

Image1.jpg (29.1 KB, 4 views)

 

You seem to be up to date on all of this. Designing this stuff was second nature to me at one point but unfortunately my career has managed to wander away from circuit design and now I've found that I'm a little rusty at it. Nowadays I make drugs, that is, I work in the automation group at a pharmaceutical company now. I miss the circuit design and working in the lab. I was lucky enough to score this project due to my background (MSEE whereas most in the group are ChE). I was of course given full access to what ever was on hand. I wonder quietly why the system maintenance guys even had this stuff in the shop, they don't tend to repair the equipment at this level.

I have finished the circuit and it works great. Thank you for your time and help. With the additional power supply range and Vin- set to 0.5 I boosted the gain to 20 to increase the resolution in the area of interest. I know that I have also introduced some error and the data logging program takes care of that.

I've heard grumblings that they will want a PWB designed (I miss that too) and the output configured to set off an alarm on low temperature. I have chosen the LM393 and wired it up on sly already so I can do it quickly if the plan goes forward.

Again thank you for your time.