I'd like to sim a load cell circuit using the following example from the AD8221 instrumentation amplifier datasheet:

​

Good news is that the AD8221 is included in LTSpice's library, so that simplifies things significantly. But my question is, how do I sim the load cell? How much does the upper right resistance shown in the 350Ω bridge varies under normal working conditions?

 

hi cm,
This is one option.
Use a 5Vext
E

 

How much does the upper right resistance shown in the 350Ω bridge varies under normal working conditions?

Welcome to the wonderful world of load cells and strain gauges. With any luck, you'll emerge from this adventure with your sanity intact-- more or less.

In a practical application, look for a resistance change of no more than a couple of tenths of a percent, though it depends on the load cell design and the characteristics of the strain gauge bonded to it. Any more than a few tenths of a percent, and for most materials you start getting in the realm of plastic deformation.

 

Welcome to the wonderful world of load cells and strain gauges. With any luck, you'll emerge from this adventure with your sanity intact-- more or less.

In a practical application, look for a resistance change of no more than a couple of tenths of a percent, though it depends on the load cell design and the characteristics of the strain gauge bonded to it. Any more than a few tenths of a percent, and for most materials you start getting in the realm of plastic deformation.

Thanks for the warm reception... but I've been digging in and out of this subject for a few years already, and I'm afraid to say that the only reason I haven't sought therapy yet is because my medical insurance does not cover this sort of thing.

That being said, many thanks for getting involved, Dave. Your help is always thoroughly appreciated. I've built a few circuits over the years, and they all have had mediocre results, at least from my perfectionist perspective. But there's something new that I've now mastered (more or less) that wasn't in my arsenal back then. I can now do quite a good job when soldering SMT components.

Most high quality amplifiers and ADC's are now offered in that presentation only. And before I started playing with my air gun and soldering paste stencils, I would solder those components using an ordinary soldering iron with a very sharp tip. Needless to say, but more than 95% of the time those components ended up being damaged from overheating. I didn't even know what was going on back then until I accidentally did something right and retraced my steps to find out what it was that I was doing wrong.

Anyway, I think I'm now ready to revisit this area. I'm going to start by making sure that the circuit I build has been tested in LTSpice as best as it can, and then I'll very carefully design the PCB layout and so on...

 

I've been practicing SMT soldering with a hot air gun and Maker Paste low-temp solder, on little header boards from Adafruit. Seems to work OK, with some practice. As with your experience, going the soldering iron route didn't work out too well.

As for simulation, I can't help much with LTSpice; I use a different simulator package, and the way I've done it is by parameterizing the resistor value and doing a .DC Sweep on that parameter. I assume there's something similar in LTSpice.

 

hi cm,
This is one option.
Use a 5Vext
EView attachment 214273

Thanks, Eric.

I'm trying to understand what you did there. You made the value of R2 a function of V4, which in turn is a voltage source that increases with time, right? Why not simply make R2 directly proportional to Time?

 

maybe https://www.hardysolutions.com/tenants/hardy/documents/5FactorsA.pdf
+ https://www.fierceelectronics.com/components/strain-sensor-basics-and-signal-conditioning-tips

 

Ok, here's the first part of my plan, the load cell's excitation.

​

• V1 is purposely noisy. I added a 100mV rippl to it, since it is meant to represent an ordinary wall wart with a supposedly regulated output.​
• L1 and C1 are there as an LC filter, which has worked wonderfully for me in previous designs, using said wall wart.​
• RA, CA are an RC filter, which is being "multiplied" by Q1. One of Dave Jones videos explains it thoroughly.​
• RA1 and Q2 are there as a way to boost LT1021-5 current output. This is done as suggested by its datasheet.​
• RB, D1 and U2 are there to trim LT1021-5's output, also as per its datasheet.​

 

Why bother adding pain and torment to your life with analog front-ends? They only add error, noise and drift to an already perfectly good analog signal. Just run the bridge right into a good delta sigma with a built-in PGA and be done with it.

 

maybe https://www.hardysolutions.com/tenants/hardy/documents/5FactorsA.pdf
+ https://www.fierceelectronics.com/components/strain-sensor-basics-and-signal-conditioning-tips

Thanks for those two excellent sources of info, C...

Question, on the second website it's stated that:

... a 350 V, 2 mV/V strain gauge delivers full output when one leg drops by 0.8% to 347.2 V

I plan to build this thing using a 1.5 mV/V gauge. How do they know that full output from the cell is delivered when the sensing leg drops by 0.8% when the cell is of the 2mV/V type?

 

Why bother adding pain and torment to your life with analog front-ends? They only add error, noise and drift to an already perfectly good analog signal. Just run the bridge right into a good delta sigma with a built-in PGA and be done with it.

ok ... I know you're not fond of suggesting specific chips when asked ... but here's a question, would that sort of chip have zero-drift capabilities and temperature compensation?

also, without asking for specifics, what brand/family of chips would you suggest for that?

 

I've had great success with the TI ADS1242.

 

I've had great success with the TI ADS1242.

Thank you , Joey... one of the reasons why I was working on the separate components approach was cost. For instance, the AD770 boasts quite a few desirable features, but it's in the range of more than $20 dlls each.
The chip you've suggested costs less than half of that. I'll look it up.

 

Interesting.... even though the ADS1242 has an internal PGA, its datasheet suggests using an external amplifier circuit for maximum accuracy:

​

 

Interesting.... even though the ADS1242 has an internal PGA, its datasheet suggests using an external amplifier circuit for maximum accuracy:

View attachment 214312​

No. High precision.

You'll get 50K divisions without the IA.

 

No. High precision.

You'll get 50K divisions without the IA.

Right... the difference between precision and accuracy is important. Thanks for clarifying.

Anyway, by IA you mean "Internal Amplifier" right? If that's the case, that's less than 16 bits, why would resolution suffer so much?

 

BTW: 50K divisions is very likely far more than your load cell is capable of.

 

by IA you mean...

Instrumentation Amplifier.

 

why would resolution suffer so much?

Dynamic range.

 

BTW: 50K divisions is very likely far more than your load cell is capable of.

Yet another important piece of info, thanks. The specific cell I'll be experimenting with is the LSP-2 from Transducer Techniques:

​

It's not a bad cell, as far as I know. How do you know the number of divisions that cell is capable of,? Is it related to its hysteresis?