Strain Gauges - Signal Conditioning with programmable gain

martin

Hello Victor I've done designes very similar to your requirements. I'll put something to gether on my web site, once I get it working. I will probably have some more questions, which I will also post on the site

as soon as I can find some spare time. Robert

Thank You! I=B4ll continue studying and as soon as you=B4ve done with your site i=B4ll check this out!

Hello Victor, I have posted some questions on web site. Regards Robert

Hello Rob,

Thanks for supporting me with my strain gage project!

I=B4ll try to answer your questions the best I can:

1 - The output of my strain gage assembly will be +-10mV. 2 - When the gage is not under load, the output should be zero, shouldn=B4t it? I=B4m sorry if im being too ignorant but i=B4ve started studying electronics at college about an year ago and I haven=B4t built anything real, I just know theory. I know the gage might have some voltage even if no load in applied but we intend to calibrate it to zero. 3 - Actually, there are two projects that I want to use this signal conditioning circuit. One will use a micro the other will just have the signal to input a device that was bought to measure the deformation. I think i=B4ll handle the nulling of the equipment with a potentiometer. What I need is to amplify and filter it the best way and then input it ito an A/D converter. Basically thats what I need. But I don=B4t know how I can find the best frequency of my filter. Should I have a spectrum analyser or just through the strain gage=B4s manual I can find out which frequencies it gets?

Thank you again for helping me. I don=B4t want you to waste your time with my stuff, so, when you have a spare time you reply me!

I appreciate! Victor

1 - The output of my strain gage assembly will be +-10mV. 1-Answer:- OK, I will assume that you mean that when you apply your load to the strain gauge assembly, each of it's outputs will change by 10mV, so a 20mV change overall.

2 - When the gage is not under load, the output should be zero, shouldn't it? I'm sorry if I'm being too ignorant but I've started studying electronics at college about an year ago and I haven't built

anything real, I just know theory. I know the gage might have some voltage even if no load in applied but we intend to calibrate it to zero.

2-Answer:- In an ideal world, the output is zero for zero load. In practice, it never is. So all I really need to know is how you intend performing this null. It sounds like you intend performing this manually, so I will amend the circuit accordingly. I will however need to know what the approximate error is likely to be. This should be specified as part of the specification of the assembly.

3 - Actually, there are two projects that I want to use this signal conditioning circuit. One will use a micro the other will just have the

signal to input a device that was bought to measure the deformation. I think I'll handle the nulling of the equipment with a potentiometer. What I need is to amplify and filter it the best way and then input it into an A/D converter. Basically that's what I need. But I don't know how I can find the best frequency of my filter. Should I have a spectrum analyser or just through the strain gage's manual I can find

out which frequencies it gets?

3-Answer:- OK, that should be fine, see answer 2 for nulling. Regarding the filter, what is the maximum rate of change of load that you wish the system to respond to? In other words for example, you might apply the load and read the result 1 second later or 5 seconds etc.

Thank you again for helping me. I don't want you to waste your time with my stuff, so, when you have a spare time you reply me!

No problem, this is much more interesting than the crap I have to do for a living!!

Hello Rob,

What I mean by +-10mV is that when we compress the object we should read a negative value, when we pull the object we should read a positive value. But the overall is 20mV as you mentioned. Unfortunatelly I won=B4t be able to tell you an approximate error. It should be small but i know "small" depends on the application. I=B4ll find out as soon as i can and I=B4ll tell you. What I=B4m most worried about is the filter part. One of the projects, we apply the load and read the result 1 second later. The other one, the whole thing lasts miliseconds, so, we should have to read 1 result and then read the other one 10microseconds(100 Khz frequency) later. As I need to input the signal in a A/D I need to project an anti-aliasing filtering. 40khz LP filter should be good for anti-aliasing, but I would still have lots of noise because of the high frequency, wouldn=B4t I? As time goes by things seem to get more complicated instead of getting easier! :-) That=B4s because I=B4m learning new concepts which are important for the project! Thats it, thanks a lot again and I hope things get more exciting at your job! :-)

Victor

What I mean by +-10mV is that when we compress the object we should read a negative value, when we pull the object we should read a positive value. But the overall is 20mV as you mentioned.

Answer:- This is an important bit of additional information. You want the output to be 0-10V. But as you are indeed applying force in both directions to your gauge assembly, the output with no load will be +5V. At full positive load the output will be +10V and at full negative load the output will be 0V. Or do you want the output to go negative for negative applied loads? I don't think you do, as your ADC is probably not able to accept this.

Unfortunately I won't be able to tell you an approximate error. It should be small but I know "small" depends on the application. I'll find out as soon as I can and I'll tell you.

Answer:- In my circuit I have assumed that the bridge supply is 10V (a common supply voltage). The bridge output is commonly specified as so many volts per volt of bridge supply. So if your assembly has an output of +/-10mV with a 10V bridge supply, it will have an output of 1mV/V. The 'Null error' is needed so that I can set the amount of null adjustment needed. I will allow a reasonable amount of adjustment (say

20% (4mV null error) of the full scale output. Then we can refine it later.

What I'm most worried about is the filter part. One of the projects, we apply the load and read the result 1 second later. The other one, the whole thing lasts milliseconds, so, we should have to read 1 result

and then read the other one 10microseconds (100 kHz frequency) later. As I need to input the signal in a A/D I need to project an anti-aliasing filtering. 40khz LP filter should be good for anti-aliasing, but I would still have lots of noise because of the high frequency, wouldn't I?

Answer:- For the application requiring a 1 second response time, filtering should not pose a problem. Much more than 4 samples a second and the human eye/brain is not capable of following what's going on anyway. Your other application is more of a challenge. If the equipment is properly built, the noise should not really be as much of a problem as you anticipate. But in order to capture a fast moving event (the breaking of a rod under tension?), your ADC needs to be able to capture samples at the rate you mention, and ideally much faster than 10uS. Then you can do software averaging on a number of separate samples. I will configure the filter as a 40kHz low pass.

As time goes by things seem to get more complicated instead of getting easier! :-) That's because I'm learning new concepts which are important for the project!

As the saying goes, "the devil is in the detail" Robert

Hello Rob,

Sorry for getting too long to reply. The last days were too confusing here. You=B4re right, i don=B4t want the output to go negative for negative applied loads. Since I=B4ve got a 5V bridge supply, and my assembly output is +-10mV I have an output of 2mV/V. The fast moving event that I reffered to is a pendulum that hits a big iron bar. It lasts milliseconds. I have a question. Do you know any article on the internet or something you have at your cpu that explains there kind of stuff? If you can recommend something I=B4d appreciate! Thanks again!

Hello Victor, I will reconfigure the circuit on the web to use a 5V bridge supply. I will configure two circuits:- one for use with a DVM or similar and to have a low frequency response and have zero load as 0V, as a DVM will indicate both positive and negative strain outputs. The second will have a 40kHz LP filter and an output voltage of 5V for zero load. Thus positive full scale loads will be +10V and full scale negative load will be 0V. Regarding good sources of information. I can't think of anything better than the book "The Art Of Electronics" co-authored by Win Hill, a regular contributor on this board. Then there is always 'Googling'. I will have a quick go and let you know anything I find. Regards Robert

While you are at it look up load cell, i suspect you are trying to make something like one. Besides, it should improve your understanding of strain gauges a bit.

--
JosephKK
Gegen dummheit kampfen die Gotter Selbst, vergebens.  
--Schiller

Hello Rob!

I saw you=B4ve updated your site! Thats awesome! The configuration you are going to make is perfect and I think it=B4ll be very useful for me! Regarding the book, I=B4ll order it soon! Thanks a lot!!! Victor

Hello Victor, I've updated the first schematic and added a second. I hope they do what you want. In order to get the best out of them, you need to pay attention to the physical construction, e.g. common mode rejection, etc Regards Robert

Hello Rob,

I checked the site you=B4ve updated! That=B4s really nice you=B4re helping me with this. I=B4ve already ordered the book you(and almost everyone) recommends - The Art of Electronics. As I=B4m from Brazil, it may take about 3 weeks to get here.I told you I studied electronics the last months but I was using Sedra and Smith book. It=B4s good but there=B4s too much theory in it. I know what that book teaches but I can=B4t apply it to anything real. I also put the schematic on Pspice to see how it works. It seems perfect. I know in the computer is much different than in real life, because of many things including the physical construction as you said. The "full load" condition went to 10,005V(something like this) and the no load condition to 4.9995V and the negative one to 1.XXmV. I feel a little embarrased to ask you what I=B4m going to ask but, since i=B4m still learning electronics, I think It would be good for me if you just explain a little bit of what each part of the circuit does. I don=B4t want you to explain everything because I know it might be too long and you could spend too much time on it. I know that the in-amp have a x50 gain and as its supposed to do, it must reject the common mode noise. I also know that the in amp gain is set through the resistor. The last op-amp is the filter part. What about the other op-amps? Another question,(i promise it will be the last one :-) , in case I change the strain gage assembly so that its output is +- 20mV, is it too hard to modify the circuit? In which parts I have to focus? Thanks a lot again! I don=B4t know how to repay this! It was awesome! If you send me your addess I=B4ll send you something from Brazil, like a caipirinha, don=B4t know if you know it, its a traditional drink from Brazil. :-) Anyways, I don=B4t even know if you drink, so if you dont, never mind that! Thank you!!! Victor

Hello Victor, No problem, I will break down the circuit down into parts and explain the operation of each one. Also how to make changes so that you can change the gain and so on. AofE is good in that it explains a wide number of design areas in a simple, easy to understand way, with the least amount of mathematics I always have my copy to hand, it's my bible, as far as electronics is concerned. Thanks very much for offer of traditional Brazilian drink, that would me very appreciated. I will post my address on the web site. Regards Robert