Instrumentation Amp Interest?

[geo_leeman]

Hi all,

I've had a little project that I created to scratch my own itch as an experimental geophysicist.  I often work with load cells, DCDTS, and other instruments that need to be buffered and/or amplified in a differential way before going into my control/recording equipment. Turns out this seems to be a pretty common problem for lots of people using any kind of bridge based transducer.

Everyone seems to roll their own, but I've ended up making several for friends and others who find them to be useful prototyping/one-off tools. 

Do you all think there's enough interest to make a small run?

A few of the keys:
* Input is clamped to the rails
* Really nice INA128
* Low-pass filter that can be put in/out with solder jumpers
*4 gains (including unity) selectable with solder jumpers
* On-board regulators up to 1A to power the differential transducer as well as amplify it

Would you be interested? Why or why not? Board photo attached.

Thanks for any reality checks or suggestions!

[ez24]

Hi all,

Would you be interested? Why or why not? Board photo attached.

Thanks for any reality checks or suggestions!

My need is to determine if my building is sliding down a hill.  It was built on landfill (compost pile) and the building does not have a foundation and is strong enough to withstand a fall over.  I cannot get power equipment to the base of the "hill".  I have spent years putting concrete at the base.  Now I spend $1,000 a year on materials and labor.

For years it settled and I jacked up the building to level it.  For the last 2 years the building has remained level.   I have a level pole at the base and over the years it moved one way (settlement) then moved the other way (slide).  Someday I would like to sell the property and if I could show a "movement" chart, it might help.   So I am very interested.

[geo_leeman]

What kind of sensors would you be wanting to use with it? Strain meters or something similar?

[ez24]

What kind of sensors would you be wanting to use with it? Strain meters or something similar?

I will PM you with some pictures in the next couple of days.  thanks

[nctnico]

I have build something very similar but with BNC inputs and a 5V to +/-15V DC-DC converter. Also the input clamping is likely to mess with small signals due to leakage currents. I used an Analog devices instrumentation amplifier which has somewhat ruggedized inputs.

[CM800]

In regards to your landslide issue, could I suggest something they did for the london underground that might help?

Get yourself a load of fairly fluid cement and pump it into the ground using some steel tubes with holes at the end (hammer it in then begin pumping.

To pump you could probibly get an old hydraulic pump and a hydraulic actuator and some valves, then build a very rough compressor out of steel and a couple of one way valves to pump the concrete into the earth at high pressure. It then hardens, solidifying the ground below.

[Brutte]

This is a two channel device and it has a 1.5A excitation? That looks like 100x oversized.
Have you considered any overcurrent protection, in case of wiring failure/shorts? These 7805 shut down after exceeding 2A..

Why do you think a symmetrical supply is of any use in such application? The GND won't be tied to anything, I am afraid.

Also, not sure what configuration you want the bridge/load cell to operate at (looks like a full bridge) but you still have to null out the output somehow. Then you can use full op-amp output to match ADC that follows.

[geo_leeman]

I have build something very similar but with BNC inputs and a 5V to +/-15V DC-DC converter. Also the input clamping is likely to mess with small signals due to leakage currents. I used an Analog devices instrumentation amplifier which has somewhat ruggedized inputs.

Have you ever ran into problems with this? I added this as a ruggedizing measure per the data sheet, but the inputs are already protected to +/- 40V, so maybe it's unnecessary?

This is a two channel device and it has a 1.5A excitation? That looks like 100x oversized.
Have you considered any overcurrent protection, in case of wiring failure/shorts? These 7805 shut down after exceeding 2A..

Why do you think a symmetrical supply is of any use in such application? The GND won't be tied to anything, I am afraid.

Also, not sure what configuration you want the bridge/load cell to operate at (looks like a full bridge) but you still have to null out the output somehow. Then you can use full op-amp output to match ADC that follows.

Yes, the regulators are way overkill. A couple of the sensors I use need ~100mA each and I didn't want to even think about heat. In the next rev I'll probably go with a smaller SMT regulator if I can find a suitable one.. Which could be a challenge for 15 V.

I use a symmetrical supply since our strain gauge bridges are very well balanced, so using +/- 15 means a very near zero bridge output, even after amplification. Hence why I don't need nulling. I think adding nulling and making this a single channel device could be a great idea though. Currently the way we do this is by first passing the differential input through a unity gain buffer, then offset it with another op-amp and pot, and finally amplify. More devices to pass through though. 

[Brutte]

Yes, the regulators are way overkill. A couple of the sensors I use need ~100mA each and I didn't want to even think about heat. In the next rev I'll probably go with a smaller SMT regulator if I can find a suitable one.. Which could be a challenge for 15 V.

Yes, the heat, well it does not matter how big the regulator is - same heat has to be dissipated. But I get your point - lets use 1.5A for now and decrease it to reasonable size later.

I have never seen a 100mA * 15V = 1.5W excitation sensors. This must be a serious gear.. I wonder how big is that sensor able to dissipate 1.5W?? The constantan strain gauges that I use are 3x3mm and are excited with 10mW max.

I use a symmetrical supply since our strain gauge bridges are very well balanced, so using +/- 15 means a very near zero bridge output, even after amplification.

The bridge arrangement + opamp means it does not matter where the common voltage is. The differential voltage is amplified, common voltage is rejected. 120dB. Besides, the 780x are 3% at best so I would not expect that GND to be nowhere near the midpoint of power rails.

Hence why I don't need nulling.

I do not follow. Can you tell me where do you connect that beefy GND in a bridge + differential opamp application?

That opamp has differential signal input (two pins) that are tied to the bridge and nowhere else and output signal+reference (two pins) that are tied to ADC and nowhere else. Four pins in total take part in that play (and some powering, filtering, gain setting, NC and other stuff irrelevant here).

I do not see the place for your 1.5A GND there..

[geo_leeman]

I have never seen a 100mA * 15V = 1.5W excitation sensors. This must be a serious gear.. I wonder how big is that sensor able to dissipate 1.5W?? The constantan strain gauges that I use are 3x3mm and are excited with 10mW max.

Not all sensors we use are strain gauges. We also use DCDTs (displacement transducers that are differential transformer based) and others as well as sometimes powering other bits of rigs off what's available.

OK yes on the bridge - brain fart. In our lab we've traditionally always used +/-15 V excitation for everything (all this equipment was designed long before I was around), so I suppose that basically guided this part of the design. We can record +/-10 V with our ADC system (as is common for many NI data acquisition systems used) and like to maximize our use of that to get the best resolution possible. Many of the bits of gear we use do let us offset to near the bottom rail, but as I said, I haven't implemented that here yet.

[nctnico]

I have build something very similar but with BNC inputs and a 5V to +/-15V DC-DC converter. Also the input clamping is likely to mess with small signals due to leakage currents. I used an Analog devices instrumentation amplifier which has somewhat ruggedized inputs.

Have you ever ran into problems with this? I added this as a ruggedizing measure per the data sheet, but the inputs are already protected to +/- 40V, so maybe it's unnecessary?

In my board I wanted a 1M Ohm impedance. With diodes between the power supplies there will be a leakage current through the diodes and the resulting voltage will not be exactly zero due to variance in the diodes and power supply voltages. The result is that you will introduce a different DC offset at each leg of the instrumentation amplifier and you might also couple in noise from the power supplies.
If you want clamping diodes it is better to use (low leakage) TVS diodes to ground but in these kind of situations I'm inclined to use series resistors and rely on the ESD clamping diodes inside the instrumentation amplifier. Using series resistors is also tricky becuase the input bias current can introduce a small DC offset as well.

[geo_leeman]

If you want clamping diodes it is better to use (low leakage) TVS diodes to ground but in these kind of situations I'm inclined to use series resistors and rely on the ESD clamping diodes inside the instrumentation amplifier. Using series resistors is also tricky becuase the input bias current can introduce a small DC offset as well.

Right. Seems like putting series resistor footprints with 0 ohm jumpers could be a good way to do it. Folks could modify if they needed. Or make it cut-trace under the footprint.

[nctnico]

This is a picture of my creation:

For convenience I used BNC connectors to connect current shunt resistors or a regular oscilloscope probe.

BTW I used a relatively big case because I had a few of these on hand including laser cut front panels from another project.

[geo_leeman]

This is a picture of my creation:

For convenience I used BNC connectors to connect current shunt resistors or a regular oscilloscope probe.

BTW I used a relatively big case because I had a few of these on hand including laser cut front panels from another project.

Very nice! How do you like that DC-DC converter. Spec sheet looks pretty decent.

[ez24]

In regards to your landslide issue, could I suggest something they did for the london underground that might help?

Get yourself a load of fairly fluid cement and pump it into the ground using some steel tubes with holes at the end (hammer it in then begin pumping.

To pump you could probibly get an old hydraulic pump and a hydraulic actuator and some valves, then build a very rough compressor out of steel and a couple of one way valves to pump the concrete into the earth at high pressure. It then hardens, solidifying the ground below.

I am not skilled enough to design my own pumps.  It would be very very expensive to rent one because it would have to go about 300 feet and get a lot of attention.  I cannot afford to get attention. 

Every year I and hired help dig a 3 foot square into the ground.  Remove about 2 feet of topsoil and 2 feet of clay.  I have to use a rotary hammer ( 1 1/2 inch) with a spade bit to get the clay out. This alone takes us a day.

 Once I hit the sandstone I drill 1 1/8" holes into it about 12 inches deep and drive in 6 foot sections of pipe  (I can drill 2 holes a day - it is very hard), about 20 of them.  Then place about a yard of hand mixed concrete into it and build up from there by stacking concrete bags to 6 feet above ground level.  This takes several months and costs about $1,000 in labor and materials.  I try and do one a year (I have done 4 so far).   But the "slide" width is 30 feet.  It is like putting a finger in a dike.   There is no road access nor even wheelbarrow access.  Everything has to be hand carried to the site (downhill)

If I were to use a strain gauge, it would go on the upside of one of these pilings.  I think I would use two plates of some sort and put the gauge between them and bury it and place it on top of the bedrock.  One problem is I do not know what force it would take to shear one of these pilings.

The mass is about 30 feet wide, 21 feet long and 15 feet high, which in ruff terms is about 500 tons.  Divide this by 30 and that gives 17.5 tons a foot (for 3 foot piling).  I think a 20 ton strain gauge would work because the force is about 30 degrees (slope of hill)

[nctnico]

This is a picture of my creation:

BTW I used a relatively big case because I had a few of these on hand including laser cut front panels from another project.

Very nice! How do you like that DC-DC converter. Spec sheet looks pretty decent.

Those DC-DC converters are nice but for analog circuits they can really use extra filtering hence the electrolytic capacitors (generic remark: check the maximum capacitance a DC-DC converter can deal with) and common mode chokes.

[nctnico]

In regards to your landslide issue, could I suggest something they did for the london underground that might help?

Get yourself a load of fairly fluid cement and pump it into the ground using some steel tubes with holes at the end (hammer it in then begin pumping.

To pump you could probibly get an old hydraulic pump and a hydraulic actuator and some valves, then build a very rough compressor out of steel and a couple of one way valves to pump the concrete into the earth at high pressure. It then hardens, solidifying the ground below.

I am not skilled enough to design my own pumps.  It would be very very expensive to rent one because it would have to go about 300 feet and get a lot of attention.  I cannot afford to get attention. 

You make it sound very spooky. And do you really think all your digging and drilling isn't going to draw attention? Anyway, there are portable drilling machines the size of a pallet which can vibrate steel tubes into the ground which are then filled with concrete. These machines are specifically designed for your problem: to fix the foundation of a house with minimal disturbance. Some even fit in the crawl space!

[ez24]

You make it sound very spooky. And do you really think all your digging and drilling isn't going to draw attention? Anyway, there are portable drilling machines the size of a pallet which can vibrate steel tubes into the ground which are then filled with concrete. These machines are specifically designed for your problem: to fix the foundation of a house with minimal disturbance. Some even fit in the crawl space!

You make it sound very spooky.

  Only if it moves while I am at the base.  I do get a spooky feeling when I look up.

And do you really think all your digging and drilling isn't going to draw attention?

  Less than concrete trucks.  Plus I am very weak and slow.

portable drilling machines the size of a pallet which can vibrate steel tubes into the ground which are then filled with concrete.

  sounds expensive and they would not get into the bedrock

Do these machines have a name that I can search on?

thanks

[nctnico]

You make it sound very spooky. And do you really think all your digging and drilling isn't going to draw attention? Anyway, there are portable drilling machines the size of a pallet which can vibrate steel tubes into the ground which are then filled with concrete. These machines are specifically designed for your problem: to fix the foundation of a house with minimal disturbance. Some even fit in the crawl space!

You make it sound very spooky.

  Only if it moves while I am at the base.  I do get a spooky feeling when I look up.

And do you really think all your digging and drilling isn't going to draw attention?

  Less than concrete trucks.  Plus I am very weak and slow.

portable drilling machines the size of a pallet which can vibrate steel tubes into the ground which are then filled with concrete.

  sounds expensive and they would not get into the bedrock
Do these machines have a name that I can search on?

Mini pile drivers are the best term I guess. These are a pictures of the machines I'm talking about:



In the NL concrete pilings are driven into the ground until they hit a sand bed several meters deep (12 meters / 36 feet isn't an exception).

[geo_leeman]

This is a picture of my creation:

For convenience I used BNC connectors to connect current shunt resistors or a regular oscilloscope probe.

BTW I used a relatively big case because I had a few of these on hand including laser cut front panels from another project.

I made a little 3D printed case for mine. We used DB9 connectors for the sensors, and BNC for the output to our ADC system.