Custom FG Tail Lights!

[MattLacey]

Ok, those are looking pretty damned cool IMO!

[biddie_fiddler]

Been slow going on the lights since the power supply died.

 

Waiting on a whole lotta parts from aliexpress to make this job easier. Most of it has arrived, just got one more connector left to arrive before I can wire it all up properly, and easily this time

[biddie_fiddler]

Nothing exciting, just a little bench test box for testing tail lights. It'll save me having to put them in to the car constantly to test. Rather just bench test them  

Will also be an easy way to show the difference between a factory set and my custom set. 
Stickers to label the switches have been made since this photo was taken. Putting the Mrs Cricut Explore 3 to good use 🤣

 

For anyone curious how this thing works:

It uses an STM32 dev board as the brains, it takes the switches as inputs (with some s/w debounce to reduce components) and triggers some relays to switch the higher current outputs.

I did it this way so I could eventually run stress test patterns and I can also set the indicator switching speed.

 

[biddie_fiddler]

First fit in a tail light didn't quite go as planned. Yes it fits, I never doubted that.

But, the brake LEDs protrude too far out, causing a collision with the lens. Unfortunately that means I'll need to remake the brake section and glue the LEDs in a way so that they don't stick out too much.

Will need to ponder this one a bit more.

 

 

With that said. I have a completely functioning code build, so my main focus now will be these LEDs. Once that's done I can get the first prototype set assembled and start testing  

Here is a cut video of the start-up and shutdown sequence. Its a little old but the pattern is the same. I've made the DRL pattern a bit faster, and have implemented a smooth dimming transition between the two brightness levels. 

Will continue the project once I'm back from my holiday. With how long I've spent on these so far, I think its safe to assume that the prototype set will only be done toward the end of the year/ beginning of 2023.

[Panda Eyes]

looks awesome mate. very high cool factor. Personally I reckon the start up sequence is a bit long- twice the speed would be perfect though imo.

[biddie_fiddler]

Cheers mate!  

 

And yea I agree with you! In fact, since taking the video, I have increased the speed haha

Faster and more smooth  

[biddie_fiddler]

Wow its been a while since I started this! Quite a bit has changed since the last post. I had the proof of concept working but wasn't happy with it, the controller I designed was too restricted. I ran out of RAM and so needed to change the microcontroller. It was a big design change. I pondered for a while and thought about what I wanted to have as an end product. What I had was good, but not what I wanted. 

There were some big drawbacks in my design that other controllers on the market had. 1. PC software to configure the thing. 2. Independently adjustable channel brightness. I could only adjust global brightness and my patterns are hard coded. Not ideal.

 

So, I decided to redesign the controller to add more features and the ability to easily configure it with some PC software.

It now has LED drivers capable of controlling the brightness of every channel. I've still retained the extender to add an extra 32 channels on the same board, still on the fence if I want to do this or not though.

It also have a UART port so I can communicate through a basic serial to usb adaptor (can get them for like 5 bux!). I chose this because the other ones on the market actually compile code and require you to own a programming tool to configure. You also need an internet connection because the code is compiled on a remote server. What a crap idea, I want it to be completely local. The only benefit this has is it reduces RAM and flash usage, which is not worth it imo.

 

Its now also smaller, 39mmx52.5mm.

Debugging is going to be a pain due to the footprints of the drivers and mcu I've chosen, but its all that was in stock and met my requirements so I have to make do.

 

  

 

I'm aiming to have something I can sell as a standalone unit, that isn't car specific. Just a controller that you can configure to do whatever you want with!

[Puffwagon]

Can it run Crysis?

[biddie_fiddler]

What's the best thing to do when you successfully create a proof of concept? Start all over again

 

The project was all round a success, but was both very limited, and too complicated at the same time. My patterns were hard coded into the MCU, which means they'd only work for only the falcon lights. My original goal was to have a product that could compete with the Ghozt Sequencer, something I can't achieve use hard coded patterns. So it was back to the drawing board!

 

I redesigned the board, removing the ability to add extra boards - a feature that is too complicated for the benefits. I also chose a different LED driver chip that was similar is price but has way more functionality, kinda annoyed I didn't discover this one first.

This chip can individually control the brightness of all 16 channels per chip, and they can be daisy-chained like the other chip, so I now have 32 channels that are completely independent.

 

I have now used a much beefier MCU that has more RAM and Flash so my program can be more complex. Why do I need this? Because the board can now be connected to a PC via USB and have all patterns programmed using PC software  

that's the theory anyway, board has been designed but not made yet, code has been written but not tested. I've instead spent many, many months learning how to develop windows applications, thanks to Chat GPT its been a relatively successful, albeit slow, process.  

 

If money wasn't an issue I'd have assembled a few of these boards and started real world testing. But sadly I've never had the money to, essentially,  waste on this project for some time. I'm going to ditch the idea to re-use the internal brake light red diffuser from the tail light and instead opt to have a piece cut out of acrylic, like the indicator insert. Its just so much easier to do and I can fit it in better.

 

Here's some pics of the new board, same same but also very different

 

 

 

The PC app is where I've spent most of time, its dense, not the best design but will get the job done! It still needs a lot of work and testing but the bones are there!

So far it can:

- create patterns for all 5 inputs by adjusting the brightness of each led per pattern, I've limited it to 4 brightness levels to keep it more simple

- configure each channel for your needs

- send packets via uart (tested using uart simulators), it handles the packet creation, CRC checksums, sending packets, receiving responses and handling errors

- save projects to a file so your designs are saved. It will inform you when a unsaved change has been made and gives you the ability to save the file. 

 

Todo:

- encrypt the save file so it can't be tampered with

- implement an "open project" feature

 

Then test it! I'm sure there is a lot of bugs in it but that's part of the journey.

 

 

 

I must say, doing this completely alone has been extremely overwhelming, I have stopped working on it for weeks, then come back to it with massive ambitions, only for the cycle to repeat itself. I have so many aspirations for this and I really want to see it through. I've spent too much time on this to give up now. I just don't know what to do once its "finished". I would love to have the code reviewed but others but not sure how to go about this. For now, once I have a working board I can do some real-world testing and start fixing some bugs I'll definitely find.

 

The board itself is pretty much done, as well as the code, its always the 10% that takes the longest though. From memory, I just need to implement saving data to flash for it to be ready for testing. Then I can finalise how the patterns are handled in code.

 

The whole point of this is that I wanted to be able to sell the PCBs so that anyone can do custom lighting themselves, but realistically that's a long way away. I wouldn't know how to even sell something like this lol

[k31th]

12 minutes ago, biddie_fiddler said:

I would love to have the code reviewed but others but not sure how to go about this.

Best way to do something like this is to take out identifiable features and upload the rest to github to see if your architecture is reasonable. Even split out different "functions" you want checked into separate folders/projects in github. If you're fine with your entire programming project being online, then just upload it as-is. I might even send some pull-requests your way to help fix stuff up  not making any promises, though, as I do plenty of open-source software dev as it is haha