To recap the previous state of things: the first strain gauge installation showed an extremely weak, positive sensitivity to weight. New strain gauges arrived yesterday, so I wired up one set outside of the foot pads and directly connected to the main tube. This now shows a strong, positive sensitivity to weight, which is fantastic! I can set the gain such that it responds from 2 volts to 5 volts, although I can probably add a tiny bit of extra resistance to one of the legs of the H bridge to trim down that 2 volts to 0 volts to get more usable data from it.
And just for kicks, I decided to take the old sensor installation and wire up the output leads backwards, just in case that did anything. It produced a strong, negative sensitivity to weight, (starting from 5 volts and going down with weight) which I totally was not expecting. I thought that it would simply be unresponsive, which I guess goes to show how much I know about how op-amps work. I could hypothetically use such a reversed sensor to control the board, I’d just have to flip something around in software, but given the fact that the output jumps around when I put pressure directly over the sensors, I’m still not going to use them. In hindsight, gluing sensors underneath the most flexible part of the foot pads was a mistake, and the output from the tube based sensors looks more reliable.
All six of the sensor arrays are glued down, and connected to the amplifiers with shielded wire. Next is to connect them to the Arduino and use those NiMH batteries to make an uninterruptible power supply. Once I’m through with that I’ll start testing and calibrating the sensors, and hopefully do a full system test if they all work. Assuming it does, everything will get more permanent weather and EMI-proofing.
(Before you ask, the next video will be after the system test. Thanks for all your encouragement!)
Each array is split between the top and bottom of the tube, since those are the areas of maximum compression and tension when weight is put on the board. The wiring diagram is copied from this one, where diagonally opposed strain guages on the diagram get placed next together on the tension or compression sides.
Thanks, nice work. Excited to see what comes out of it.
I’m a little late for the party, just saw this thread. Skimming through it - don’t worry about the voltage at zero load, it’s something that’s typical of every bridge based transducer.
Where do you get your gages from? What resistance? With what do you glue it?
I’ve been doing structural testing since 2004, so this one is interesting to me🤪
Thanks! Nice to have an actual pro checking my work.
The guages are BF350’s from Elecrow, which is the only retailer I found that sells them individually instead of by the hundreds. Everything else is Digikey.
Im gluing with ZAP Z-poxy, just cuz I’m familiar with it from my RC plane hobby. It’s got a properly hard cure, and can be bought in small quantities. I know you’re supposed to use a specialized hot cure resin for permanent installation of strain guages but my structure would be damaged by autoclave temps.
So you have the 350ohm gages, which are better choice for composites.
There are plenty gage adhesives, CA like, that don’t require heating. If your gages are reacting you’re fine, cause you don’t need the precise strain, you just need proportional reading.
Looking forward for the progress, and feel free to reach out if you need some inputs structure wise.
Oh yeah? On the all of the gauge manufacturer’s product pages I read, it said CA was only appropriate for temporary installations. I figured that might be down to the lower shear strength and the brittleness of CA.
Thanks for the help! I’m glad I have the correct resistance, because I had no idea which one I should have gotten starting out, so I just got the “medium” option lol.
Been a long day but I have the control system circuitry fully assembled. Sensors, arduino, uninterruptible power supply, and piezo buzzer (to indicate armed/disarmed). Pending some multimeter checks to try to make sure it won’t set itself on fire when I plug it in, I’ll start testing each of the sensors tomorrow.
Not sure what to make of today, pretty bummed tbh. Of the sensors (0 through 5), #0 and #3 were responsive, the others output 80mV at minimum gain (like normal), but were not responsive to gain adjustments or weight. Later on in the day as i made adjustments, number 3 stopped working, number 0 became erratic, and the rest began to output zero as if they were disconnected?
The main difference from before is that the amplifiers are now running on 5V from the Arduino regulator instead of 12V from a battery. I started off by powering the gauge bridges directly from that 5V supply instead of the amplifier’s internal 5V reference like it had been before, but since I had 4 guages unresponsive I decided to change it back. I mistakenly connected the 5V reference to the 5V supply, tested it, I realized my mistake, fixed it and tested again. It was at that point that I started noticing with the erratic behavior, so I’m concerned I fried something I guess? I have 1 spare amplifier from the initial test that i could sub in.
Check the resistance over your bridge (w/ disconnected amplifier)
Getting 350ohms over in or out pins should indicate that the gages are at least not fried.
Sorry it took so long for me to write back. TBH I had some problems with getting motivated to work when I realized that I’ll need to wade ass first into some headachy troubleshooting of black box electronics.
To recap, after installing all six sensors, #0 and #3 are functional and the other 4 are unresponsive, they only produce the 80mV rest voltage and their output does not change when adjusting the gain. After finding this out I realized I may have made a mistake in how I wired the power supply for the amplifier chips, then changed it back, but that didn’t fix anything. I said in my last post that the functional ones later became erratic, but I would like to strike that from the record now, because they have been fine today and I think I just had a spotty ground connection to my meter last time (lol).
All of the bridges are 350ohms to within a few tenths, which is as good as my cheap meter can tell me (they’re highly consistent though, all have the exact same reading). I took the non-responsive #5 sensor, and swapped out its amplifier chip for my spare one, which was confirmed working back when I used it in the first successful sensor test. The spare chip was also not subjected to whatever wiring errors I made during the first pass of setting up the 6 sensors, but it was still unresponsive after being installed in the #5 location. I confirmed with my meter that the chip is being fed 5 volts, it has a ground connection, the top of the bridge is connected to the 2.5 volt reference supply on the chip, the middle points on the bridge are at 1.25 volts, the bottom of the bridge is at 0 volts, and the gain potentiometer is working correctly. There are no shorts to 5 volt or ground on any of the pins when the system is powered off, either. And all of these readings are exactly the same on the functional #0 sensor. Ugh.
So I asked myself the all-important question, “how would I have borked this up if I were a mathematically optimal idiot?” and then tried switching the pin connections for the left and right sides of the bridge. After I did that to the #5 sensor, it exhibited the weird behavior (that I had seen before when trying to set up the initial test sensor) where I can set the gain such that the resting output voltage is large, and then application of weight to the board results in the output voltage decreasing. My guess right now is that this implies the bridge is unbalanced or I hooked up something in it backwards somehow, does that sound right to you @dani ?
Wheatstone Bridge summarizes the strains in such a way that you have two opposite ones with + and the other two with -.
I your case the gages are applied in a way as to increase compression and decrease tension. If you can show me close up of the gages and directions I can probably tell where each one should go.
Other option is to set a negative calibration factor and proceed with your plans.
Yep, it certainly is good to know that it’s responsive. By setting a negative calibration factor, do you mean flipping the behavior around in software? That’s definitely doable.
Here’s a close-up of the #5 sensor array. Top of the picture is tension.
When the twisted wire pairs come together into solid black, red, and white pairs, they correspond to +Vin (red), -Vin (black), +Vout, and -Vout (both white) in this diagram. The white pair that comes from lower left to upper right in the previous picture is supposed to be -Vout and vice versa.
It seems like the amplifier chips can only measure a positive difference between the output nodes, not a negative difference. So if the resistances of the four legs are out of whack such that the difference at rest is negative, that would explain the behavior I’m seeing, no? I have some resistive wire that I could use to try to trim the bridge back to level. Scratch that, it’s not really solderable. Probably not to a vibration resistant degree anyway.
