Wireless Scale Conversion-open source collab

[Richard Hybels]

What is a dead head unit?
I just bought a set of Intercomp Kart scales so my hillbilly deal will be more hillbilly than yours.
Thanks for sharing. I know guys that would not be caught dead with anything but a $5k brand new alu platten.

[Rick Kirchner]

What is a dead head unit?

The power supply and display part. If that goes your scales are boat anchors unless Longacre can (or wants) to fix it.

[Steve Davis]

Yeah.. NOW I understand... dead "head unit"

[J Leonard]

The power supply and display part. If that goes your scales are boat anchors unless Longacre can (or wants) to fix it.

Wouldn't the "head unit" just need to read the millivolts from the strain gauge/ load cells in the scale pads then convert to a weight number?

[Rick Kirchner]

Wouldn't the "head unit" just need to read the millivolts from the strain gauge/ load cells in the scale pads then convert to a weight number?

You need to be able to calibrate to zero and set the scale factor. That functionality is built into the unit I posted.

Scorp997 was working on turning his scales into wireless. I have been doing a hardware design but I suck at code. Basically you need something like a Raspberry Pi Pico W at each scale with a battery, a charge controller and a HX711 load cell interface. The HX711 I posted will do everything but the wireless part and each scale has to be calibrated separately.

So now you have 4 scale units and if you take another Raspbetty pi or even write an app for your phone, you can put all 4 displays in the same place as well as trigger a cal cycle.

That's really all the scale manufacturers have done, just different parts.

[Mike B]

You need to be able to calibrate to zero and set the scale factor. That functionality is built into the unit I posted.

Scorp997 was working on turning his scales into wireless. I have been doing a hardware design but I suck at code. Basically you need something like a Raspberry Pi Pico W at each scale with a battery, a charge controller and a HX711 load cell interface. The HX711 I posted will do everything but the wireless part and each scale has to be calibrated separately.

So now you have 4 scale units and if you take another Raspbetty pi or even write an app for your phone, you can put all 4 displays in the same place as well as trigger a cal cycle.

That's really all the scale manufacturers have done, just different parts.

I talked to this company at PRI last week: https://creativeracing.com/ and they said they could convert any scales to wireless for only $1,500.

[Rick Kirchner]

For an all-pi solution, its about $225 in parts, plus 5 3d printed enclosures. If you do the display off a phone, its about $145 in parts, and 4 enclosures.

If you just do the HX-711 modules with individual displays, It's about $120 in parts.

It seems our sport is chock full of people funding their racing habit via other racers...$1500 bucks? I think once designed, you could probably crank them out for a half-hour of labor per corner. So yeah, say $225 in parts, maybe two hours of labor (whatever you think that's worth) might get you to $500. Then you throw in a fancy box, a website, shipping, the potential for warranty, and double it, and $1500 doesn't seem like such a reach.

I'd rather make something open source for the community.

[Mike B]

I think once designed, you could probably crank them out for a half-hour of labor per corner. So yeah, say $225 in parts, maybe two hours of labor (whatever you think that's worth) might get you to $500. .

I'd rather make something open source for the community.

I'm in.

[jcolley]

You need to be able to calibrate to zero and set the scale factor. That functionality is built into the unit I posted.

Scorp997 was working on turning his scales into wireless. I have been doing a hardware design but I suck at code. Basically you need something like a Raspberry Pi Pico W at each scale with a battery, a charge controller and a HX711 load cell interface. The HX711 I posted will do everything but the wireless part and each scale has to be calibrated separately.

So now you have 4 scale units and if you take another Raspbetty pi or even write an app for your phone, you can put all 4 displays in the same place as well as trigger a cal cycle.

That's really all the scale manufacturers have done, just different parts.

Former job we collected a lot of data with ESP32 IOT devices. WiFi connected and used MQTT protocol to publish messages (data points, in our case temp and humidity) to the MQTT broker (a WiFi Connected Raspberry Pi) where they were stored in an InfluxDB database. On the Pi, it hosted NodeRed and we created dashboards that would show all the incoming data. All of it done with open source software, Pi was $50 (at the time) and the ESP32s were WeMos D1 Minis https://www.amazon.com/Organizer-ESP8266-Internet-Development-Compatible/dp/B081PX9YFV/ref=sr_1_3?keywords=wemos+d1+mini&qid=1702500224&s r=8-3

The D1 Mini has a single 0-3.2V analog input, so a little voltage divider might get you there. Not sure how accurate the reading would be and how calibration would go, but shouldn't be too complicated. The InfluxDB might be overkill.

NodeRed is hosted on a web server on the pi, so connect phone to the same wifi network and navigate right to the dashboard from your phone.

Maybe in the spring if I get some free time I'll see what I can knock together.

[Rick Kirchner]

The problem with talking to me about software was the last thing I wrote, beyond writing a little labview piece to evaluate its use on a project in 1996, was a heart rate monitor done using a Rockwell 64 SBC for my senior project in 1982. It was written in opcode and was less than 100 LOC, but it was stored as FM tones on cassette tape to make loading it easier....

I can design software, partition between software and hardware, I just can't write code...

Bought Pi for dummies, a book on python on the pi, even downloaded Labview for Pi and a couple of Pi toolkits. My problem is everybody that writes this stuff starts in the middle. I need it Barney-style.

The Pi Pico W is $6 each and the and the Zero 2W both have 2.5Gig Wi-fi and BT. I have heard rumblings it doesn't work that well but who actually knows, and under what circumstances? I'd think a static layout where all the parts are less than 15 ft apart is pretty easy.

Dumping to a phone much preferred over the Pi zero solution - much cheaper of course but more of a hassle to get the app on the phone, and then you have the android vs apple thing. A decent display for the pi is about $38, although I was looking at some e-ink displays that are cheaper and of course don't use much power at all, the do have slow refresh rates. Its just that most of the cheaper displays are very small.

[Mike B]

The problem with talking to me about software was the last thing I wrote, beyond writing a little labview piece to evaluate its use on a project in 1996, was a heart rate monitor done using a Rockwell 64 SBC for my senior project in 1982. It was written in opcode and was less than 100 LOC, but it was stored as FM tones on cassette tape to make loading it easier....

I wrote some rudimentary PLC ladder diagrams in college. Does that help?

[jcolley]

The problem with talking to me about software was the last thing I wrote, beyond writing a little labview piece to evaluate its use on a project in 1996, was a heart rate monitor done using a Rockwell 64 SBC for my senior project in 1982. It was written in opcode and was less than 100 LOC, but it was stored as FM tones on cassette tape to make loading it easier....

I can design software, partition between software and hardware, I just can't write code...

Bought Pi for dummies, a book on python on the pi, even downloaded Labview for Pi and a couple of Pi toolkits. My problem is everybody that writes this stuff starts in the middle. I need it Barney-style.

The Pi Pico W is $6 each and the and the Zero 2W both have 2.5Gig Wi-fi and BT. I have heard rumblings it doesn't work that well but who actually knows, and under what circumstances? I'd think a static layout where all the parts are less than 15 ft apart is pretty easy.

Dumping to a phone much preferred over the Pi zero solution - much cheaper of course but more of a hassle to get the app on the phone, and then you have the android vs apple thing. A decent display for the pi is about $38, although I was looking at some e-ink displays that are cheaper and of course don't use much power at all, the do have slow refresh rates. Its just that most of the cheaper displays are very small.

Rick, I've looked around a bit and can't find anything for wiring diagrams on the Longacres. I'm looking to buy an older set from a guy near me and might want to play with this idea. Do you know what power/signals the scale cables carry?

[Rick Kirchner]

Rick, I've looked around a bit and can't find anything for wiring diagrams on the Longacres. I'm looking to buy an older set from a guy near me and might want to play with this idea. Do you know what power/signals the scale cables carry?

If your longacres are like my rebcos, there should be 4 screws holding the upper plate to the strain gauge beam. Take those off, and you should see a beam with the gauge bonded to it. There are probably 4 wires there and the color code is pretty standard - red, green, black, white.

My rebcos have HUGE mil-spec connectors, so I plan to solder a much smaller connector in parallel to it in order to make the attachment of the wireless unit to it a lot easier.

https://tutorials-raspberrypi.com/di...-sensor-hx711/

Question - would anyone be interested on taking this to a separate thread to create a crowd-sourced solution? I can certainly do the hardware design - that's just box pluggin.

[jcolley]

If your longacres are like my rebcos, there should be 4 screws holding the upper plate to the strain gauge beam. Take those off, and you should see a beam with the gauge bonded to it. There are probably 4 wires there and the color code is pretty standard - red, green, black, white.

My rebcos have HUGE mil-spec connectors, so I plan to solder a much smaller connector in parallel to it in order to make the attachment of the wireless unit to it a lot easier.

https://tutorials-raspberrypi.com/di...-sensor-hx711/

Question - would anyone be interested on taking this to a separate thread to create a crowd-sourced solution? I can certainly do the hardware design - that's just box pluggin.

Thanks, will take a look if I pick them up.

Yep, I'm in. It's not really coding intense if you use something like NodeRed. It's drag and drop coding for idiots like me. Made a bunch of medical devices manufacturing engineers think I knew what I was doing.

[scorp997]

I built an Ardrino head unit a while back, should be on another thread. While not wireless, it did allow for calibration, corner weights, cross weights and F/B and L/R distribution. Ode came mostly from the Ardrino forum l, modified for my use and single 4 line display.

[redcar9]

I'm game to help with this project. I have access to a PCB assembly line.

Meaning if someone designs this and sends Gerber / BOM details. I can run them in small batches.

I can work with a board house + assembly + test.

[Garey Guzman]

I'll ask my CS&E major son if he has any ideas on this. I also have a dead head-unit with good scales so maybe he can make/test something.

[Rick Kirchner]

So here's what I was starting with: https://www.amazon.com/gp/product/B09Z29TK7J/

Now the nice thing about this module is that it can work as a stand alone unit. It has a display, it can be calibrated, and you can power it off of a 3.5mm jack or a small USB. I plugged it in today and the display lights up, so that's a start. At the bottom of the amazon page there's a link to user's manuals, and in the reviews a guy does a better job of translating operation from chinese.

When using these with a Pi there are a number of places where wires can be soldered between boards or you can use the header connector . There's also a header for the strain gauge input.

I also purchased these connectors to make linking the smaller components easier. https://www.amazon.com/gp/product/B09Z29TK7J/

Might not have been the greatest decision. they're a bit short, which limits where I might put the box for each scale. They are pretty inexpensive, high quality, keyed, and relatively weatherproof though. I was thinking about those snap connectors used in AIM systems, and they would work well but are surprisingly expensive.

There are a couple of ways to go with the Pi. At each scale, we could use a PiPico W: https://www.amazon.com/DVOZVO-Raspbe...dp/B0BMP5546H/

I've heard the wireless and BT don't work that well, but don't know enough to make a credible judgement based on the comments made. We could also use a standard pi pico: https://www.amazon.com/Raspberry-Pi-...dp/B09437S9X4/
with a HC 05 BT module plugged into it: https://www.amazon.com/AEDIKO-Blueto.../dp/B09M6ZVJVR

To run it off of a battery we'll need a LiPo shim: https://www.pishop.us/product/lipo-shim-for-pico/ or you have to get a "pirate pico" - those are made by a different manufacturer but they have LiPo charging built in.

Then for a battery: https://www.amazon.com/gp/aw/d/B08214DJLJ/. A 2000ma/hr battery would last for days running a scale and BT. However, these are 3.7 V and the shims are designed for that. The HX711 requires 5V. I believe the pis have a boost converter on them to supply 5v to other devices, but I need to check.

I think it would be simplest to use the HX711 module above, a Pi PicoW with a lipo shim and a 3.7 battery. If the HX711 board above will not work off of 3.7, then a standard hx711 board would be required. I'm betting it needs the higher voltage to power the display.

[Rick Kirchner]

Here's a reb-co load cell naked. You can see the color coded wire if you download the instructions for the HX711 module I posted from Amazon it has the bridge schematic.

[Rick Kirchner]

When finished, we could probably write an article for Grassroots Motorsports.

[Rick Kirchner]

https://www.anyload.com/load-cell-fo...-how-it-works/

[Rick Kirchner]

I'll ask my CS&E major son if he has any ideas on this. I also have a dead head-unit with good scales so maybe he can make/test something.

Good class project......

[Rick Kirchner]

Gave some thought to flowcharting the software.

At each scale, as soon as the power comes on it needs to place BT in discover mode and wait for comms from the head unit.

When the head unit powers up it needs to connect to each scale and show a graphic of connected scales. probably needs a button to re-start trying to connect.

First time on, once comms are established each scale needs to tell the head unit which position it is in and if not named, ask to be named. Head unit allows you to name each scale (needs to go into flash). each scale is assigned a repeating reporting time associated with a prime number (23, 37,59, and 71 ms). this will assure that as each scale reports it will minimize any data collisions.

First time on each scale needs to tell the head unit it is not calibrated. Head unit prompts user to put a known weight on the scale and tell the system how much it weighs. Scale uses this data to calibrate. and places the value in flash. Head unit walks user through each scale position.

Each scale now tells the head unit is is not zeroed. Head unit prompts user to zero all scales, or each scale individually.

Head unit now displays each scale weight.

[Rick Kirchner]

I just verified that the HX711 board I found on Amazon will run on 3.5 Volts. Have not seen how far down it will run, nor did I evaluate if he calbration changes with power supply voltage.

[chrisw52]

Gave some thought to flowcharting the software.

At each scale, as soon as the power comes on it needs to place BT in discover mode and wait for comms from the head unit.

Once you have each scale connected to a Bluetooth BLE connection, you can use any android or iOS device as a head unit. Apple devices are better at multiple BLE connections than android, but both platforms can easily be supported with the standard development software available.

I have used this tutorial to read a load sensor used in automotive scales. If you can match the colors on the wires, you're ok. That is well standardized. https://tutorials-raspberrypi.com/di...-sensor-hx711/. I got stalled on the work because the prototype board I used could not support multiple BLE addressing, so I couldn't support the multiple BLE connections.

The cool part begins when you start connecting the load sensor to the device over BLE bluetooth. Now you have a wireless bluetooth connected device that could be expanded to have weight, camber, caster, etc..

[Rick Kirchner]

I bought some headers, a power supply, soldered the smaller connector in parallel with the big ones on my scales, and tried to get the Amazon modules I described in another post working. No dice. the manufacturer even sent another. Still no dice. I think its just bad instructions.

I can get them to power up and zero but I can't set the scale factor.

I was following instructions posted in the amazon reviews section for the product, but still can't get it going. Tried to contact the guy on FB (he's a professor) but no reply.

If anyone would like to try to solve this problem I'll ship them my two modules and the data I've collected on them, and a couple of connectors. Heck, I'll even throw in a breadboard and some hookup wire.

[scorp997]

I will check my notes and see if I can get an answer for you. I remember the instructions were not super helpful and I think I used instructions from another manufacturer to get mine to work (I then gave up on them in favor of the full Arduino project that used different modules)

[Rick Kirchner]

I did not know you used the same unit at first. To get that working is super cheap if you don't mind doing the simple math and walking around the car to look at the displays.

[scorp997]

I went back to the Amazon product page you linked to earlier and one of the reviewers, D. Maciuca, wrote a very nice instruction outline in his review. Look that over and it may make more sense and provide you with a starting point for your trial and error tests.

[Rick Kirchner]

I went back to the Amazon product page you linked to earlier and one of the reviewers, D. Maciuca, wrote a very nice instruction outline in his review. Look that over and it may make more sense and provide you with a starting point for your trial and error tests.

Yeah, that was the stuff I tried. I'm not getting it somewhere but I'll make another attempt. I reached out to the guy on FB messenger and never got a reply.