Is it a loadcell? I don’t think it actually says it is. It says “pressure”. Could be a piezo sensor like the E3D product.
I’m just going off the four wires on it hoping that it’s a full bridge strain gauge. Maybe half bridge at worse. If it’s not a strain gauge I’ma be really sad let’s put it that way lol
Looking at the construction it appears to be a full bridge load cell AND assuming the put the gages in the right place it should be (almost) immune to side loads. The documentation leaves a lot to be desired.
Probably wont be in my hands till late next week, Ima guess 7/17 at the earliest. Anticipation will be killing me until then lol.
So close yet so far away. It will work, probably better than any other I’ve seen due to the optimum mounting location. But it is going to have more nonlinearity than it could have had and some side load sensitivity. Gages are too big and are laid on areas where the strain fields are not uniform. And on this sample, off center and crooked. What a waste of excellent machine work. The solder debacle is not Mellow’s fault.
Based on CAD and that pic, I have in hand two full bridge strain gauges exactly like that. I dont have any of the special glue, or the special white protection stuff though. But, if mine doesn’t meet expectations I may have a go at replacing the gauge myself and using two instead of one. Just have to wait and see I suppose. If it can set decent first layer personally I’m not super picky. You wont catch me doing full bed first layer tests and picking apart the results basically lol
Gauge is probably the wrong type too. These things need FEA. It’s non trivial.
No special glue required, water thin CA works fine.
You do however need cellophane tape NOT vinyl scotch tape. Cellophane is non elastic so you don’t pre-stretch the gages. The white is simply acid free RTV. I always paint the grids with polyurethane varnish first as I never trust the “acid free” claim.
This device begs for 2 half bridge gages, one centered on each beam. Even better would be 4 1/4 bridge, top and bottom of each beam.
For what it’s worth, if you’ve never laid a gage and you only have 2 available then there is a 50% chance you don’t have any 
you’re probably right I bet I’ll mess up the two I have. I was just looking at other gauges and visually I’d say you’re spot on with using two half bridge so that the grid area stays in the flexing area better.
I mean Finite Element Analysis, simulation. To understand where the peak strain is to locate the gauge, and pick a sensor orientation.
Oh I totally agree. But I’m working with what there is available. I’m sure it’s not perfect, but I’m also not designing my own nor investing tons of effort. “Perfect is the enemy of good enough” and that’s near enough for me 
2 Likes
Don’t write this off, it is an EXCELLENT solution to the endstop detection function. I was just disappointed in the poor execution of the gage placement. It won’t be as accurate as it could have been in the role of filament force monitoring that @garethky wants to eventually implement.
Lets circle back to that $13 circuit board
@mykepredko Is it feasible to flash firmware that would allow this to attach to a Klipper host as a stand alone MCU and send force readings?
The documentation does mention a “programable gain amplifier” which I assume is the goldish device below the STM32. A metal case makes sense for a analog amplifier working with low level signals.
The firmware (stock Klipper or custom) would need to be able to set the gain as well as interface with the ADC.
Most likely the answer is “yes” but I can’t find schematics for the board to confirm that.
There are “R” and “B” buttons on the PCB that I presume are for “RESET” and “BOOT0” which implies you can get into DFU mode and then you’re halfway there.
If you have one of these boards, could you connect via USB to a Linux/SBC host, press and hold down “B”, press and release “R”, then release “B” and then execute lsusb on the Linux/SBC host? If you get back a device with 0483:df11 then you can definitely add your own firmware.
If you’re going to be working new hardware then I’d like to see if you could wire up an SWD interface but I don’t know about that until I can see schematics.
I don’t have a board. I buy one and ship it to you if you have time/interest to poke at it.
Mellow has schematics for many of their boards on git… but not this one. I haven’t asked them for one as of now you’d have to reverse engineer it.
I have Zero experience with writing or modifying firmware beyond compiling a few Marlin variants using their very excellent tool chain. Commenting and uncommenting lines in a well written configuration script is NOT writing firmware.
Let me know if you want a sample.
I really don’t have the time to go through the board and try to suss out the circuitry. It’s not hugely complex that will take a bit of time.
If you can get a schematic then I’ll take a look at it again (I don’t mind buying one for myself - it’s not that expensive and the approach taken is interesting).
I fully understand the issue of too many projects and not enough time.
Anybody else wants a sample to poke at, let me know.
1 Like
I can almost guarantee there’s nothing stopping from flashing klipper to the mcu, or any other firmware you want(I doubt they’ve blocked the bootloader) from dfu mode. I’ve got all kinda stm mcus running klipper over USB connections, its really no big deal. Looking at the tech doc on the adc it uses I can almost guarantee it’s connected over spi to the mcu, just have to probe and find which pins are what(which I plan to do). All that is no different than using the currently supported ads1220 connected to an mcu and configured in klipper(I have my ads1220 connected to a pi pico flashed with klipper, same concept). Issue being we don’t have klipper support for the ads131m02 that this uses. If support were added I’m confident this could be a standalone probe module.
I don’t think so. To me, it looks like the clock source for the STM32.