So anytime you get a reading from the bridge it is unbalanced. The lab grade daq units balance the bridge by injecting a resistance parallel to the sensor. This is typically done in the beginning of the test, it does not affect the response of the bridge, just the DC value at 0 load.
In your case you can zero the reading in the software
1 Like
Right, I understand that point. What I meant is that it could be unbalanced at the rest, before it should be providing any kind of reading. And since the amplifier chips only detect imbalance going one way, the bridge could be moving around under load, but just in a non measurable zone.
EDIT: And the reason I want to get it working in the forward direction is just that the range of the output signal looks to be wider than when reversed, so I would get more throttle resolution. Not a dealbreaker though.
2 Likes
Yep, that’s what I got.
So I took apart some 30 gauge silicone wire, separated the strands, and twisted them back together into pairs that measured around 1 ohm for every 15 centimeters. I added 1 ohm at a time between the ground node and the low side tension gauge, and it did bring the array closer to being balanced at rest, I think. For each test I began by setting the gain such that the output was 3 volts at rest, and as I kept adding resistance it needed more gain to reach that value. With no extra resistance added, full application of my weight lowered the output to 2.75 volts, but by the time it was up to 4 ohms of extra resistance the output range was 3 volts through 1.4 volts, which is quite good. And it’s similar to what I was getting on the #0 and #3 sensors, which I trimmed to give me an output range of 2 volts through 3.5 volts.
I could keep going and maybe get it working in un-reversed mode, but the wire length is getting a little bit difficult to manage. So I think I’ll just stop there and set a negative calibration constant like you suggested. I’m kind of stumped about how I could have ended up with a bridge that was out of balance at rest by an amount in the single-digit ohms, though. As I was building all of the bridges I measured the resistance of each gauge after soldering, and they were all consistent to within a tenth.
EDIT: Found some 3/4 turn pots that are 10 ohms max instead of the usual 10k ohms, I’ll get some of those instead, much easier. See ya in 3 or 4 days when they show up.
1 Like
Applying gages is also an art 
, so don’t beat yourself up.
2 Likes
I really love the idea and project tho I have some remarks, I think the speed range should be lowered 65kph is a very high speed for a board relying on a new input method, surely it could be tested and run without issue, but that is probably for the generation 2 or 3 board as any issue with the strain gauges can make for a very sad day.
So my personal reccomendation would be Pneumatics like 7" or 6.5" tires and a top speed of around 45 km/h and stick in a BMS into the system for charge only to make your life easier 
, good luck with the project and I hope it turns out as good as you hope !
I won’t, thanks 
So it could have gotten messed up during the adhesion process? Interesting. I only had them taped down by their wire leads while the glue was curing so no pressure on the guages themselves, could it have been shrinkage in the glue?
@baein Thanks for the kind words. The hardware is firmly set now, I’m not sure how switching to pneumatics would help with potential sensor problems? And do you mean replace my hobby charger with a BMS?
It’s not like I’ll be doing 65 all the time, it’s about having the option to do so when I find suitable locations for it. I’ve hit those speeds on skates and bikes before, I’m aware of how stupid it is.
1 Like
This can be one of the reasons. The proper way to glue them is to use appropriate adhesive, with 2-3 additional chemicals to apply before. Then you use constant pressure over the gage for good adhesion . Our technicians have these tiny sand clocks to time themselves.
1 Like
Dang, i need to get me some sand clocks. Thanks, good to know.
1 Like
Well I was thinking pneumatics for steadier ride, less “false” input from the terrain and more input from you, just skipping the hobby charger for a BMS just makes life easier and the board more easy to go as if you ask me having a board ready to go quickly helps those “odd” rides every now and then.
Just trying to look out for you because most fun is had below 50 km/h and above that is just pure adrenalin usually, this is a new tech branch and you just try it out alot and for long distances before you start hitting “crazy” speeds .
1 Like
Ah, i see. Road inputs will be dampened by a software filter, they might not be problematic. Or rather, I can filter as heavily as I need to make them go away, but that’s a trade-off with control lag. Good point about pneumatics though, that certainly would make it easier on the system. But it could also be that urethane is fine, we just don’t know yet. (EDIT: I just remembered that physical filtering introduces latency just like computational filtering does. Pnummies may be equivalent to an LPF, IDK.)
I plan on using my local park for a burn-in test, has a huge wide road (low consequences for wiping out) and some patches of gnarly pavement then I can use to put the system through its paces.
That’s an upside to a bms that I hadn’t thought of. I think I’ll stick with my hobby charger because I already have it and I like the level of control it has though.
1 Like
Holy shortcakes, I’ve made a mistake. Gotta check the product dimensions next time, more soldering nightmares.
3 Likes
I can now balance all of the bridges! They’re all working correctly and in the non inverted direction. Still going to need to spend some more time dialing them in better, but once they’re good I will glue them in place so that vibration doesn’t change them. The next obstacle to overcome is my programming skills or lack thereof lol, my code didn’t exactly work properly on the first try.
(If I ever try to use components this small again though, I do want one of you to come over here and slap me)
5 Likes
GUESS WHAT
24 Likes
Nice! Now for the painful tuning process lol
3 Likes
Haha yup, no way am i taking my laptop out with me for testing once i get this rolling (crunch), so if I wanna change any settings I’ll need to come back inside.
3 Likes
Potted all the control electronics today, which messed with the calibration a bit, I straightened it out in software and it’s back to normal. Lost some of the resolution on sensor #0, but oh well.
I put these shields over all of the control electronics too, to protect against EMI. Just a fish paper sandwich with aluminum tape in the center, connected to a ground wire with some electrically conductive tape.
Power system is next, which ought to be the first bit of actually charted waters this build will sail through
4 Likes
Drives are sorted. Retaining compound on all the screws, threadlocker on all the gear teeth, a light coating of red n tacky between the motor shaft and pinion, and 2 papers worth of space between the inner and outer housing cases.
Also wanted to share a hack for those too lazy to get circlip pliers: you can get them off by bending out the prongs of a pair of tweezers, and then forcing the tweezers apart with regular pliers. Guaranteed to work by at most the 14th attempt.
Had to take care of a scraping noise inside of one of my motors, turned out just to be the end of the sensor wire sheath being a bit loose.
6 Likes
That gif is really cool! Nice demo
2 Likes