Strain Gauge to INA125 Troubles

So I purchased a 2kg scale with the intention to link it to an arduino to send the scale readings to the serial port for further use. I obtained a INA125P In-amp with the hopes to amplify the signal to do this. There are several examples out there of folks hacking scales in this manner. I am running into some issues where the output sweep seems to be limited and I'm not sure why.

The resistance on the bridge is 1k ohm (reading between V- and V+). The excitation V is 3.3V. The voltage to the INA125P is 3.3V. The reference is the INA125P on-board 2.5V. I have had trouble finding a solid gain to amplify the signal, and it seems a resistor of 22 ohm gives a reasonable sensitivity of the scale response to minimal weight but the amp seems to saturate at about 500g. When weight is applied the voltage drops from the max voltage to zero at about 500g and anything beyond that just sits at a little above zero. I was hoping to replicate the scale at the full 2kg. Why would the amp be saturating at only 500g? I've tried other gains with pots and resistors, but nothing really outputs a reasonable sweep beyond 500g, or requires too much force to sense readings from the sensors.

Greatly appreciate any insight & help. I am by no means an EE and am trying to learn! :)

This is essentially the circuit minus the shown RL resistor (from the data sheet).

Firstly you seem to be powering the INA125 from 3V - why? That really limits its output range. Use 5V.

Also you will need to lower the gain somewhat - if the range you get is 0..500g, and you want 0..2000g then you'll have to increase Rg by a factor of 4. But if you power the INA from 5V you'll get more output range anyway, so try that first.

You mention you are using the built in 2.5V reference, but that's not what the circuit diagram says - is that diagram exactly what you have?

What is the no-load output voltage? You want the output to range across as much of the supply range as possible, if the bridge isn't balanced there are various schemes to balance it out.

Well, the scale originally only used 3V, so I figured why bump it to 5V if it isn’t necessary? I put it to 5V just to see if it improved.

I knocked the Rg up to 88 ohms and it seemed to lower the no load voltage from 3.02V to 0.88V. I noticed the circuit is much more sensitive to interference as well. If i put my hand near the scale or IC it will change the Vo by lowering it (seems to be the case for both 22ohm and 88 ohm, but stronger at 88 ohm).

After the changes, there seems to be no improvement with the range. The Vo still drops to 0.07V (lowest it goes) at around 500g weight applied.

I wired up the INA125P just as the datasheet shows in the pic attached. I used the 2.5V ref pin 14 into 4 as shown.

Even with a simple INA128 I had the same results as this. Is it the load cell? It has the E+ E- S+ and S- outputs like any other.

Try switching the bridge connections to the + and - inputs round - then you output will increase with increasing load.

You probably might then need a balancing circuit to trim the zero reading to something around 0.4V or so (at low range of INA125's output, but not saturated).

If the circuit is sensitive to your hand it likely to be oscillating - you need a ground-plane for high-gain amp circuits like this and keep inputs shielded and away from outputs.

Some good and bad news. I flipped the + for the - on the bridge inputs and this seemed to help in allowing the voltage to go beyond the point it did before. I noticed that when I place the excitation voltage on the 5V instead of pin 4 as suggested by the datasheet, the sweep is larger. So, the good is the overall sweep is pretty well dialed in as far as allowing 2kg of force, but (the bad) there seems to be no way to allow a sensitive reading. The scale is capable of sensing 1/2 a gram but I can't seem to get it to change voltage from the no load point until about 50g or so of weight. Could this be an indication of a bad amp selection? Other possibilities?

No, its just that the zero-point isn't within the output voltage range of the amp - you need to trim the bridge offset to get it within the active range - various trim circuits are out there on the web, might be as simple as one potentiometer.

So the no load voltage should be zero, and to accomplish that, I would need to modify the circuit to trim it to zero? I tried googling around but mostly found circuits to balance the bridge. Is that the same? I was assuming the bridge was factory balanced as there is no additional components on the original circuitry from the scale. The leads go strait into the microcontroller under the edge of the sealed black blob.

Any verified examples I could compare my situation to to attempt to zero out my bridge? BTW, the bridge appears to be made of the apposing strain gauges (3 wire each) crossed into the 4 outputs of excitation and sense, both +/-.

Thanks again for helping me with this..

No! The no-load voltage should be something that doesn't saturate the INA125, which if I recall the datasheet is 0.4V or more. Not too much more or you'll lose useful range. How about 0.5V as a compromise?

Having the amp saturated means it won't change at all until the input comes back into range.