Understanding INA125 - Strain Gage Application
mgbolts
Posts: 2
Hi, I have some questions regarding the INA 125 which I would very much appreciate some advice on. I am quite new at this and little AR. I wish to improve my understanding and the overall accuracy/reliability of the unit or at least what its weaknesses are. I am using the INA125 on a load cell (simple bridge type 5kg).
I have successfully connected it (copying someone else's fine work http://cerulean.dk/words/?page_id=42). In regards to the setup, the INA is powered with 5V, the same 5V supply as the bridge and the IC. The Vref5 (pin15) is connected to the VrefOUT(pin4), but this is not powering the bridge. The questions are as follows:
1. I am using a cheap resistor for the trim (pin 8 and 9) to get the desired gain. I presume a cheap resistor will drift more with time and temp and therefore change my gain. Any suggested resistor type to improve this. I am not concerned with drift between uses as I will 'zero' the reading periodically with the IC maths.
2. My circuit already uses a precision voltage reference (LT1021-5) for the voltage reference on my IC. The idea was to improve the accuracy of the IC ADC readings from the INA125. Could I also use this as a more accurate voltage reference for the INA125 (pin 4)? It looks like the LT1021 also cant handle the load for the bridge?
3. My bridge is rated 5-10v. Is it not better to power the bridge at 10V then 5V? I says this on the premise that I should have greater resolution (2x)?
4. Similarly, will my resolution improve by running a 10V reference (INA125 pin4) as compared to a 2.5v or 5v?
5. If I go to a lot of hassle setting up the INA125 as best as possible, that's great but... at the end of the day, all it does is do a great job at gain. If the voltage out of the bridge is unreliable because it is merely a function of the quality of the bridge power supply. I note that the TI datasheet (Figure 1) has pin 4 (&16) providing power to the bridge. I have read elsewhere on this forum that bridges can pull hard and drop the voltage supplied by the INA125. The proposed TI solution is the use of a transistor (TIP29C) in Figure 4. Is this good way to go?
I look forward to your thoughts? Any other ideas?
Kind regards
Mark
I have successfully connected it (copying someone else's fine work http://cerulean.dk/words/?page_id=42). In regards to the setup, the INA is powered with 5V, the same 5V supply as the bridge and the IC. The Vref5 (pin15) is connected to the VrefOUT(pin4), but this is not powering the bridge. The questions are as follows:
1. I am using a cheap resistor for the trim (pin 8 and 9) to get the desired gain. I presume a cheap resistor will drift more with time and temp and therefore change my gain. Any suggested resistor type to improve this. I am not concerned with drift between uses as I will 'zero' the reading periodically with the IC maths.
2. My circuit already uses a precision voltage reference (LT1021-5) for the voltage reference on my IC. The idea was to improve the accuracy of the IC ADC readings from the INA125. Could I also use this as a more accurate voltage reference for the INA125 (pin 4)? It looks like the LT1021 also cant handle the load for the bridge?
3. My bridge is rated 5-10v. Is it not better to power the bridge at 10V then 5V? I says this on the premise that I should have greater resolution (2x)?
4. Similarly, will my resolution improve by running a 10V reference (INA125 pin4) as compared to a 2.5v or 5v?
5. If I go to a lot of hassle setting up the INA125 as best as possible, that's great but... at the end of the day, all it does is do a great job at gain. If the voltage out of the bridge is unreliable because it is merely a function of the quality of the bridge power supply. I note that the TI datasheet (Figure 1) has pin 4 (&16) providing power to the bridge. I have read elsewhere on this forum that bridges can pull hard and drop the voltage supplied by the INA125. The proposed TI solution is the use of a transistor (TIP29C) in Figure 4. Is this good way to go?
I look forward to your thoughts? Any other ideas?
Kind regards
Mark
Comments
Firstly using ratiometric ADC will remove the need for an accurate voltage reference - strain guages (and other wheatstone-bridge sensor
arrangements) are inherently ratiometric, so using or deriving the ADC reference from the bridge supply voltage (using a 4-wire connection
scheme?) ought to to solve that issue. Failing a ratiometric arrangement simply measure the bridge supply with the same ADC chip and
perform the division in software This is an application where differential input ADC is a natural fit BTW.
The gain-setting resistor should preferably be fixed and high-stability and accuracy (0.1% or better?) - failing that a cermet 10-turn trimmer is the minimum
spec to go for I think.
10V will give more resolution but will increase self-heating of the sensor - I'd say pay careful attention to noise filtering and ground
routing and use the lower voltage if its more convenient - but check the noise and offset-drift performance of the sensor and INA125.
As I understand it, you are recommending the voltage reference is driven by the same voltage supply to the bridge. On this basis, we get a symmetry of input and output, essentially an 'auto balancing' negating the need for absolute accuracy, rather a relative accuracy irrespective of the absolute voltage supply. Assuming I got the concept right I will walk through the circuit and see how it hangs together.
Alternatively, you are suggesting I measure the brige voltage supply via the IC ADC. On this basis if I make an assumption of the mv/V and the gain, I can mathemtacially adjust for movements in the input voltage.
On the noise filtering, this is a little ahead of me at this stage. Will take your advice and research it further.
Also, thanks for the recommendation on the resistor.
Kind regards
Mark