What's new
What's new

Building differential electronic levels?

NateA2

Cast Iron
Joined
Sep 4, 2012
Location
Ann Arbor MI.
I have been contemplating building a set of electronic differential levels. For the past few months I have been reading up on how they (supposedly) work and thinking about how difficult it might be to make a set.

The best information I have found on how they work is in this 8 page PDF.

http://www.google.com/url?sa=t&rct=...QD3Ntbi79PkcRdFzw&sig2=3Bh8MBabXKlU5i3lwgQt6g

Looking into how my electronic indicators (knowing they may prove to be used as the sensing portion of a home-brew electronic level system) I took one apart (30.00 on ebay and was worth every penny) I learned that its really just a simple LVDT Linear variable differential transformer - Wikipedia, the free encyclopedia of sorts.

for testing purposes I attached a 6" rod to the indicator and put a weight at the end. the indicator thus became part of a pendulum that used the indicators lever pivot (as the pivot of course) and allowed me to set it up as a crude and uncalibrated level for testing. I was able to easily see very minute changes in deviation from horizontal, and thus I concluded that as a POC, this test was a success.

Since electronic indicators appear to be nothing more than a conveniently laid out LVDT of sorts with a mounting portion and a lever, I began looking into LVDT's. they are readily available- at relatively high cost. They additionally lack the needed form format for use in an electronic level. For an electronic level I need a completely non-contact LVDT. Well...turns out they are actually simple and cheap to make in a wide variety of forms.

See here for how easy they are to make. LVDTs

Now, electronic indicator DC readouts range from cheap to outrageous second hand on ebay, that said, they appear to be build-able with some available components like the phillips NE5521. These aparently contain most of whats needed to drive a LVDT and have it coupled to a DC meter. IF this IS the case, then making a display with two DC meters and the appropriate nulling ability (not to mention output that can be input into a PC) should be quite do-able.

So...if the LVDT's are cheap and simple to make, the NE5521's are available at ~30 USD ea (I bought 3 from china for testing) DC meters are ~15.00 each, I should be able to make a set of levels for under a few hundred dollars (if I don't include my time)

Building a LVDT that has a non contact core riding on 2 reeds and weighted as a pendulum should be fairly simple.

Am I crazy, or does this look as possible as I think it is?
 
Yes NateA2, you frick'n crazy...NOT!

When you get to that point, look into the Protolabs(?) Cool Idea program for assistance. Or Kickstarter...or both.

Good luck to you!
 
I have been contemplating building a set of electronic differential levels. For the past few months I have been reading up on how they (supposedly) work and thinking about how difficult it might be to make a set.

The best information I have found on how they work is in this 8 page PDF.

http://www.google.com/url?sa=t&rct=...QD3Ntbi79PkcRdFzw&sig2=3Bh8MBabXKlU5i3lwgQt6g

Looking into how my electronic indicators (knowing they may prove to be used as the sensing portion of a home-brew electronic level system) I took one apart (30.00 on ebay and was worth every penny) I learned that its really just a simple LVDT Linear variable differential transformer - Wikipedia, the free encyclopedia of sorts.

for testing purposes I attached a 6" rod to the indicator and put a weight at the end. the indicator thus became part of a pendulum that used the indicators lever pivot (as the pivot of course) and allowed me to set it up as a crude and uncalibrated level for testing. I was able to easily see very minute changes in deviation from horizontal, and thus I concluded that as a POC, this test was a success.

Since electronic indicators appear to be nothing more than a conveniently laid out LVDT of sorts with a mounting portion and a lever, I began looking into LVDT's. they are readily available- at relatively high cost. They additionally lack the needed form format for use in an electronic level. For an electronic level I need a completely non-contact LVDT. Well...turns out they are actually simple and cheap to make in a wide variety of forms.

See here for how easy they are to make. LVDTs

Now, electronic indicator DC readouts range from cheap to outrageous second hand on ebay, that said, they appear to be build-able with some available components like the phillips NE5521. These aparently contain most of whats needed to drive a LVDT and have it coupled to a DC meter. IF this IS the case, then making a display with two DC meters and the appropriate nulling ability (not to mention output that can be input into a PC) should be quite do-able.

So...if the LVDT's are cheap and simple to make, the NE5521's are available at ~30 USD ea (I bought 3 from china for testing) DC meters are ~15.00 each, I should be able to make a set of levels for under a few hundred dollars (if I don't include my time)

Building a LVDT that has a non contact core riding on 2 reeds and weighted as a pendulum should be fairly simple.

Am I crazy, or does this look as possible as I think it is?
.
a cheap chinese precision level of 0.02mm/m is $30-60. the main advantage of the electronic levels is the software connected to a program on a laptop computer that draws out the measurements with error magnified. the ones i have used showed 0.0001" per 4" readings and you slide it along. if sensor is picked up and tilted it rarely when set down will repeat. i find to get repeatable readings they must be treated very gently.
 
So that would be ~.00078" per meter or ~.00026" per ft. Add to that that the repeatability might not be the greatest on a precision chinese level, and the fact that it would not have an analog readout, and I am unconvinced that they would be in any way a suitable solution for a set of differential electronic levels.

I'm all for cheap solutions, but I don't think they have the necessary resolution or repeatability for measuring in the manner that I am looking for.

The commercial differential electronic levels use a gage amplifier, a LVDT (or similar transformer setup) and measure with great sensitivity.

How hard could it be to make a program in excel or some such that captures a digital output and then graphically represents it? seems that many business software graphing programs should be able to do this very easily...
 
Hmm... after doing some quick searching, I cannot seem to find electronic levels that spec that resolution at prices like you state above. Were you talking about spirit levels?

I'm after the same level of precision as a 10K+ set of differential electronic levels by Mahr, Federal or the like. They have .1-.2 arc second resolution.

Anyone know offhand (lol) what .1 arc sec is in inches per foot?
 
Yes Nate you are crazy...but dont let that stop you, go for it and share with us how it works out for you. :)

Charles

What's wrong with crazy? "All are crazy but me and thee, and sometimes I'm not sure about thee."

Probably most of the innovators and inventors were at some time considered as crazy by some who knew them. Nate, it will be interesting to see where your idea leads. As described it sounds quite rational to me.
 
Hmm... after doing some quick searching, I cannot seem to find electronic levels that spec that resolution at prices like you state above. Were you talking about spirit levels?

I'm after the same level of precision as a 10K+ set of differential electronic levels by Mahr, Federal or the like. They have .1-.2 arc second resolution.

Anyone know offhand (lol) what .1 arc sec is in inches per foot?
.
1 arc second is .006" per 100 feet
.0006" per 10 feet
.00006" per 1 foot
.00005" per 10" give or take
........ personal experience pick up one of those levels and tilt them and set back down and 90% of time they do not repeat. i find sliding gently is the only way to get repeatable readings. They do not repeat if handle roughly and carelessly.
 
.
1 arc second is .006" per 100 feet
.0006" per 10 feet
.00006" per 1 foot
.00005" per 10" give or take
........ personal experience pick up one of those levels and tilt them and set back down and 90% of time they do not repeat. i find sliding gently is the only way to get repeatable readings. They do not repeat if handle roughly and carelessly.

Biggest concern for me re: electronic levels would be how to verify its' reading - a quick, simple, and 100% reliable procedure for a spirit level. Would you need a spirit level to verify the electronic level??

And the first time an electronic level proved non-repeatable, it would be returned to the vendor or go in the trash. "Trust is a very important part of a relationship." One strike and you're out.
 
Then there is the differential part of the problem.

The sensitive elements have to be matched in some way so tilt for tilt, they put out the same signal. The solution might be a simple gain control.

Nulling would be simple: an op-amp is essentially a differential amplifier. Feed the DC volts output from the signal conditioner chips into the + and - inputs. The output suitably scaled is fed to the meter or numerical readout. When tilrts ae equal the meter reads null.

Another part of the problem is scaling. Here again, a switched voltage divider.

Sorry, I imagine these functions would be better managed digitally but my old fart mindset is analog and so I write in analog.

LVDT's are fairly easy to make if you have good eyes and hands for working with spider web sized wire. The core can be a ferrite rod and the moving element can be a simple copper washer acting as a shorted turn in the flux created by the primary. The center-tapped secondary voltage will be influenced by the movement of the washer. Maybe a sleeve instead of washer but what the aspect ratio should be, I haven't the foggiest. Maybe a teenie ferrite toroid.

Tiny insulated lead wire may be harvested from defunct wired computer mice. It's #26 ga stranded or there abouts. I'd rather solder #34 magnet wire to #26 stranded than honkin big #16.

Here's a source for coil forms:

Phenolic Tubes | U.S. Plastic Corp.
 
The sensors I have seen in this application have used peizo-electric motors to excite a pendulum damped resonator, with the strain on the pendulum varying the resonant frequency of the assembly.

Piezoelectric / Sensors And Actuators / Applications / OOFELIE::Multiphysics - Engineering Standard Multiphysics Solutions - Open-Engineering

A number of off the shelf devices are available as MEMS components.

MEMS-Based Sensor Technology|Freescale
MEMS and Sensors - STMicroelectronics

Once a sensor configuration is built the raw data must be processed and the sensor 'trained' to its purpose. The software side is as challenging as conceiving and assembling the hardware.


A MEMS sensor with attached pendulum:

mems.jpg


A quick glance shows the first problem would be attaching a suitable pendulum to a commercially available MEMS device.


.
 
While mems sensors certainly look interesting, I think we may be making this a good bit more complicated than it needs to be.

commercially available (analog) DEL's appear to use a transformer setup that is the same type (possibly in a different form) as in LVDT's.

So... a pendulum or flexing reed setup and a suitable gauge amplifier, and Voila! an electronic level has been created. Admittedly, I have not gotten as far as figuring out how to make them act in a differential manner, and how to data record and plot usefully, but the hardware side does not appear to be very complex.

As far as data capturing, aren't their simple circuits that take an analog input and output digital? I know their are.

My first step is to produce a functional electronic level based upon a reed-hinge mechanism and a LVDT style transformer. I have several LVDT's already ordered for disassembley and testing. I think I will try taking out the core and cylindrical grinding it down a few thou for clearance and attaching the body of the LVDT to the chassis of a level and the core to the reed-hinge mechanism. With decent alignment, and with the very short stroke (swing) of the reed-hinge, I should be able to make a non-influenceing interface between the LVDT and the core.

Then its just a question of scale and lengths- the width of the level, the height, the strength and form of the reed-hinge etc...
 
While mems sensors certainly look interesting, I think we may be making this a good bit more complicated than it needs to be.

Frankly, MEMS postdates my actual experience in the matter. (An accelerometer for inertial navigation). The device we used was a ring actuator with a filament in tension where the horizontal beam is in this gif. The resonating filament was the sensor element. The pendulum would be attached to the lever acting on the top arch.

apa2.gif


The device was very small, but not microscopic like today's MEMS devices. I'm more than a little surprised that a high precision level is not an off-the-shelf MEMS component.

As far as data capturing, aren't their simple circuits that take an analog input and output digital? I know their are.

Data capture could be as simple as an Ardurino, but I think a device with an embedded FPGA would be much more flexible. Something like the Altera DE1-SoC development board has 8 channels of 12-bit ADC with a 1MHz sample rate, and 6 PLLs.

Raw sensor inputs are uncalibrated, non-linear, subject to hysteresis effects and their performance is environment dependent. You will have to model the raw data from a newly designed sensor and compare the models predictive performance against a known standard to build a sensor that gives a repeatable output in calibrated units, rather than a device that reads more or less. That is an involved iterative process.

.
 
Out of curiosity I did a quick survey of current off-the-shelf inclinometer technology.

While there are quite a few MEMS based inclinometers available their best performance seems to be limited to about 1 arc-second resolution and 2 arc-seconds repeatability.

Sub arc-second sensors seem to be dominated by macro scale servo-feedback designs.

A web search led me to the web site of Singer Instruments (not Singer-Kearfott where I once worked), where they have a series of tutorials on LVDT sensors, inclinometers and signal conditioning devices.

Singer-Instruments - Introduction to LVDT - Basic Principles

.
 
That Singer information makes me wonder whether a project might be ahead by utilizing one of their small LVDT units and focusing on the mechanical pendulum part. Maybe it's just my inclination since I take a mechanic's approach even to electronics.
 
Out of curiosity I did a quick survey of current off-the-shelf inclinometer technology.

While there are quite a few MEMS based inclinometers available their best performance seems to be limited to about 1 arc-second resolution and 2 arc-seconds repeatability.

.

I build fluid mechanics demonstration devices --- Little River Research & Design --- we need very crude tiltmeters; 1/100 degree is fine; but the best simple electronic level I've found is good to about that. And no MEMS devices that precise, more like 1/10 of a degree --- I'd really like to know what you found there. The one we like now uses an aluminum dome, about 3 inches across, with a bubble in oil, and works on capacitance. Sorry to go OT, but this whole thread is super relevant to my work -- thanks.
 








 
Back
Top